diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,216253 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.9925603945498622, + "eval_steps": 200, + "global_step": 35800, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 7.8656, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 1.0282, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 1.9432, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 11 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 2.9279, + "step": 12 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 1.9644, + "step": 13 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 1.0658, + "step": 14 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 15 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 16 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 3.9697, + "step": 17 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 18 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 6.6126, + "step": 19 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 7.3985, + "step": 21 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 7.784, + "step": 22 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 23 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.7991, + "step": 24 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 7.3936, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 1.6084, + "step": 26 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 10.303, + "step": 27 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 28 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 8.3689, + "step": 29 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 30 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 31 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 1.1597, + "step": 32 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 33 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 34 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 35 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 36 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 8.2998, + "step": 37 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 3.3718, + "step": 38 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 39 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 4.0934, + "step": 40 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 4.7952, + "step": 41 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 42 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 43 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 44 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 45 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 46 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 47 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 48 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 3.808, + "step": 49 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.6558, + "step": 50 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 51 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 7.064, + "step": 52 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 53 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.7119, + "step": 54 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 55 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 56 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 57 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 4.075, + "step": 58 + }, + { + "epoch": 0.0, + "learning_rate": 0.0, + "loss": 0.0, + "step": 59 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 60 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 61 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.5647, + "step": 62 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 8.1373, + "step": 63 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 64 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 65 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 66 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 67 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 5.8967, + "step": 68 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.442, + "step": 69 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 70 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 71 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.9519, + "step": 72 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 73 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.3925, + "step": 74 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 3.4677, + "step": 75 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.9471, + "step": 76 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.7176, + "step": 77 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 78 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 5.1744, + "step": 79 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 80 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.1019, + "step": 81 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 82 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 6.9118, + "step": 83 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 84 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 85 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 3.1822, + "step": 86 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.4597, + "step": 87 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 88 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.1091, + "step": 89 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 6.3657, + "step": 90 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 8.4478, + "step": 91 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 92 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 3.7869, + "step": 93 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.7626, + "step": 94 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 95 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 96 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 97 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.0969, + "step": 98 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 99 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 3.9912, + "step": 101 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 102 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 6.7689, + "step": 103 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 9.1101, + "step": 104 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 5.1302, + "step": 105 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.5629, + "step": 106 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 107 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.3772, + "step": 108 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 109 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.9704, + "step": 110 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 111 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 112 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 113 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 114 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 115 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 116 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.3458, + "step": 117 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 118 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 7.34, + "step": 119 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.8325, + "step": 120 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 121 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 122 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 123 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 124 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 6.2913, + "step": 125 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 126 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 127 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.9513, + "step": 128 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.9344, + "step": 129 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 7.5585, + "step": 130 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.3171, + "step": 131 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.23, + "step": 132 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.0497, + "step": 133 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.3475, + "step": 134 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 135 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.7963, + "step": 136 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 8.3855, + "step": 137 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 138 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 139 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.051, + "step": 140 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.139, + "step": 141 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 142 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 6.0531, + "step": 143 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 144 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.9902, + "step": 145 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 146 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 8.1704, + "step": 147 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.5269, + "step": 148 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 149 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 11.6175, + "step": 150 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 151 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 152 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 3.4356, + "step": 153 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 6.543, + "step": 154 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 3.7644, + "step": 155 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 5.096, + "step": 156 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.1356, + "step": 157 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 158 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.1688, + "step": 159 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 160 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 161 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.1989, + "step": 162 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 4.0312, + "step": 163 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 164 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 165 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 7.6709, + "step": 166 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.0191, + "step": 167 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 3.2597, + "step": 168 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.0086, + "step": 169 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 1.8913, + "step": 170 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 171 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 7.6918, + "step": 172 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 173 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 174 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 175 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 176 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 177 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 2.6726, + "step": 178 + }, + { + "epoch": 0.01, + "learning_rate": 0.0, + "loss": 0.0, + "step": 179 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.8788, + "step": 180 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.9036, + "step": 181 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.7778, + "step": 182 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 183 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 184 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 7.2872, + "step": 185 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 186 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.9714, + "step": 187 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 188 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.2156, + "step": 189 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.7094, + "step": 190 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 191 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 4.5634, + "step": 192 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 193 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.8668, + "step": 194 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.8239, + "step": 195 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 196 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 6.1047, + "step": 197 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 3.049, + "step": 198 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 199 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 200 + }, + { + "epoch": 0.02, + "eval_loss": NaN, + "eval_runtime": 39.8975, + "eval_samples_per_second": 62.661, + "eval_steps_per_second": 7.845, + "step": 200 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 7.5773, + "step": 201 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.959, + "step": 202 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 203 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 6.9605, + "step": 204 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 205 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 206 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.0732, + "step": 207 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 208 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 209 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 3.9526, + "step": 210 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 7.3479, + "step": 211 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.8793, + "step": 212 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 8.3999, + "step": 213 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 214 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.094, + "step": 215 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 216 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 3.6086, + "step": 217 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 218 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 6.6629, + "step": 219 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 220 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 12.4961, + "step": 221 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 8.853, + "step": 222 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.2854, + "step": 223 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 3.8395, + "step": 224 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 225 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.0311, + "step": 226 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 227 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 228 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.0992, + "step": 229 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 230 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.651, + "step": 231 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 232 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 6.1443, + "step": 233 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 234 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 4.3989, + "step": 235 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 3.7411, + "step": 236 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.0457, + "step": 237 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.3211, + "step": 238 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 239 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 240 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 241 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 242 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 243 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 244 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.7661, + "step": 245 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 246 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 247 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 6.5682, + "step": 248 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 249 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 9.3318, + "step": 250 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 7.0875, + "step": 251 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 252 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 253 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.2082, + "step": 254 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 3.917, + "step": 255 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 256 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.9379, + "step": 257 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.065, + "step": 258 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 259 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 4.9534, + "step": 260 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.9632, + "step": 261 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.579, + "step": 262 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.9179, + "step": 263 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 264 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 265 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.5389, + "step": 266 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 267 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 9.6248, + "step": 268 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.9248, + "step": 269 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 4.1094, + "step": 270 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 271 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.0121, + "step": 272 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.1852, + "step": 273 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 4.1785, + "step": 274 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 275 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 5.9483, + "step": 276 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.9695, + "step": 277 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 278 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 279 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.5464, + "step": 280 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.8343, + "step": 281 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 282 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 283 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 8.6875, + "step": 284 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 8.2669, + "step": 285 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 286 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 6.296, + "step": 287 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 288 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 5.4732, + "step": 289 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.131, + "step": 290 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 3.0949, + "step": 291 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 2.1698, + "step": 292 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 293 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 294 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.9374, + "step": 295 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 296 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 1.0929, + "step": 297 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 298 + }, + { + "epoch": 0.02, + "learning_rate": 0.0, + "loss": 0.0, + "step": 299 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.9852, + "step": 300 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 301 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.9438, + "step": 302 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 3.3896, + "step": 303 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 304 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 305 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.125, + "step": 306 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 307 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 10.782, + "step": 308 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.2746, + "step": 309 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 310 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 311 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.9245, + "step": 312 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.5496, + "step": 313 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 314 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 315 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 316 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 317 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 318 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 319 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.8616, + "step": 320 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.8013, + "step": 321 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 3.6248, + "step": 322 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 6.314, + "step": 323 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 324 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 325 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 6.263, + "step": 326 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 3.5941, + "step": 327 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 328 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 3.9027, + "step": 329 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 7.2188, + "step": 330 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 5.0104, + "step": 331 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 3.4421, + "step": 332 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.2332, + "step": 333 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.0827, + "step": 334 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 5.3278, + "step": 335 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.9879, + "step": 336 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 337 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 6.867, + "step": 338 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.9943, + "step": 339 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.6983, + "step": 340 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.8157, + "step": 341 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.792, + "step": 342 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 343 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 344 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 345 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.8973, + "step": 346 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.0653, + "step": 347 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 348 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.8235, + "step": 349 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 350 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.8159, + "step": 351 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 352 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 353 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.6773, + "step": 354 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 355 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 356 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 357 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.7282, + "step": 358 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 5.9502, + "step": 359 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 360 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 361 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 362 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.753, + "step": 363 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.6362, + "step": 364 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 9.6134, + "step": 365 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 366 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 6.971, + "step": 367 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 368 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 369 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.2781, + "step": 370 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.4037, + "step": 371 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 3.5748, + "step": 372 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 373 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.8172, + "step": 374 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 375 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.0415, + "step": 376 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 377 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.0685, + "step": 378 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 3.2433, + "step": 379 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 380 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.0042, + "step": 381 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 6.782, + "step": 382 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 7.3923, + "step": 383 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.0649, + "step": 384 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.2409, + "step": 385 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.7474, + "step": 386 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.9989, + "step": 387 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 388 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 389 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 10.665, + "step": 390 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 5.0467, + "step": 391 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 392 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 393 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.0216, + "step": 394 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.2566, + "step": 395 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 396 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.9763, + "step": 397 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.0852, + "step": 398 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 399 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.8061, + "step": 400 + }, + { + "epoch": 0.03, + "eval_loss": NaN, + "eval_runtime": 39.8297, + "eval_samples_per_second": 62.767, + "eval_steps_per_second": 7.858, + "step": 400 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 7.5152, + "step": 401 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 402 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 403 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 404 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.2591, + "step": 405 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 406 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 407 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.8126, + "step": 408 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.1018, + "step": 409 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 410 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 8.0506, + "step": 411 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.8737, + "step": 412 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 413 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 2.0774, + "step": 414 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 415 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 1.9694, + "step": 416 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 4.9445, + "step": 417 + }, + { + "epoch": 0.03, + "learning_rate": 0.0, + "loss": 0.0, + "step": 418 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 10.2866, + "step": 419 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 11.7945, + "step": 420 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 421 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 422 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.9741, + "step": 423 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 424 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.8931, + "step": 425 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.6194, + "step": 426 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.0959, + "step": 427 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 8.3362, + "step": 428 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.5821, + "step": 429 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 430 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 8.6856, + "step": 431 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 432 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 7.0903, + "step": 433 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 434 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.0673, + "step": 435 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.0861, + "step": 436 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 9.6113, + "step": 437 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 438 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 439 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 440 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.9422, + "step": 441 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 6.954, + "step": 442 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.2373, + "step": 443 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.9292, + "step": 444 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 445 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 446 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.7804, + "step": 447 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 448 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.2207, + "step": 449 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 450 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 10.1459, + "step": 451 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.9618, + "step": 452 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.8807, + "step": 453 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.0898, + "step": 454 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.0712, + "step": 455 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.7826, + "step": 456 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 8.3758, + "step": 457 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.8666, + "step": 458 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 459 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.1852, + "step": 460 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 461 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.021, + "step": 462 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.9276, + "step": 463 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 464 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.9801, + "step": 465 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 466 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 467 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 5.1888, + "step": 468 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 7.3664, + "step": 469 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.4276, + "step": 470 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 471 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.4786, + "step": 472 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 473 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 474 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.0148, + "step": 475 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.5472, + "step": 476 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 477 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 478 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.5864, + "step": 479 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 480 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.2661, + "step": 481 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.6547, + "step": 482 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 483 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.1045, + "step": 484 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 485 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 12.0126, + "step": 486 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.772, + "step": 487 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.4677, + "step": 488 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.1773, + "step": 489 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 490 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 491 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 492 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 493 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 494 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 495 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 496 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.8717, + "step": 497 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 498 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.1468, + "step": 499 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 500 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 501 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.0958, + "step": 502 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.7644, + "step": 503 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 504 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.0106, + "step": 505 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.8965, + "step": 506 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 3.3549, + "step": 507 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.9415, + "step": 508 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.845, + "step": 509 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 510 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 511 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 512 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 513 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 514 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 515 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 2.281, + "step": 516 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 517 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 4.2067, + "step": 518 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 519 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 1.9827, + "step": 520 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 521 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 522 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 523 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 524 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 525 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 8.4011, + "step": 526 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 527 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 528 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 529 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 530 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 531 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 6.3056, + "step": 532 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.9643, + "step": 533 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 534 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.941, + "step": 535 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 7.1205, + "step": 536 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 537 + }, + { + "epoch": 0.04, + "learning_rate": 0.0, + "loss": 0.0, + "step": 538 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.8026, + "step": 539 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 540 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.0144, + "step": 541 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 542 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 543 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.0393, + "step": 544 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.9106, + "step": 545 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 5.2542, + "step": 546 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.6485, + "step": 547 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 548 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 7.1187, + "step": 549 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 7.8299, + "step": 550 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.4757, + "step": 551 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.9913, + "step": 552 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.8072, + "step": 553 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.8315, + "step": 554 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.8233, + "step": 555 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 556 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.614, + "step": 557 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.8776, + "step": 558 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.6988, + "step": 559 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.0449, + "step": 560 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 561 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.9217, + "step": 562 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.9549, + "step": 563 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 564 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 565 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 566 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 5.1169, + "step": 567 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.5502, + "step": 568 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.2611, + "step": 569 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.7141, + "step": 570 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.5312, + "step": 571 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.0473, + "step": 572 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 573 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 6.8228, + "step": 574 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 575 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.2283, + "step": 576 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.1753, + "step": 577 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 578 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 579 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 6.1476, + "step": 580 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 581 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 582 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 583 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.738, + "step": 584 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 585 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 586 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 587 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 588 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 589 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.8714, + "step": 590 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.4218, + "step": 591 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 6.8856, + "step": 592 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.8931, + "step": 593 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.694, + "step": 594 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 595 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 596 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.0402, + "step": 597 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.098, + "step": 598 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.1381, + "step": 599 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 600 + }, + { + "epoch": 0.05, + "eval_loss": NaN, + "eval_runtime": 39.7217, + "eval_samples_per_second": 62.938, + "eval_steps_per_second": 7.88, + "step": 600 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.0858, + "step": 601 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 602 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 603 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 604 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.9128, + "step": 605 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.4089, + "step": 606 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.824, + "step": 607 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 608 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.0034, + "step": 609 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.0646, + "step": 610 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 611 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.9691, + "step": 612 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 613 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.9835, + "step": 614 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 6.5856, + "step": 615 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 616 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 5.0959, + "step": 617 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.7782, + "step": 618 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 619 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 10.481, + "step": 620 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 621 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.7293, + "step": 622 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 7.9782, + "step": 623 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.3009, + "step": 624 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 7.9604, + "step": 625 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.8977, + "step": 626 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.9511, + "step": 627 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 628 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 6.6955, + "step": 629 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 2.0364, + "step": 630 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 631 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.755, + "step": 632 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.6912, + "step": 633 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 634 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.8998, + "step": 635 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.0269, + "step": 636 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 637 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.4809, + "step": 638 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.3139, + "step": 639 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 640 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.5369, + "step": 641 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.3416, + "step": 642 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 643 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 644 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 1.2059, + "step": 645 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 646 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 9.3014, + "step": 647 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 648 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 649 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.2074, + "step": 650 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 651 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 9.46, + "step": 652 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 7.4575, + "step": 653 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 3.8373, + "step": 654 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.8771, + "step": 655 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 0.0, + "step": 656 + }, + { + "epoch": 0.05, + "learning_rate": 0.0, + "loss": 4.735, + "step": 657 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 658 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 659 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.953, + "step": 660 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.4065, + "step": 661 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.9374, + "step": 662 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 663 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.8956, + "step": 664 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.9916, + "step": 665 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.7454, + "step": 666 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 667 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.0376, + "step": 668 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 669 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.3834, + "step": 670 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.8972, + "step": 671 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.1114, + "step": 672 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 5.0952, + "step": 673 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.8723, + "step": 674 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 9.575, + "step": 675 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.9317, + "step": 676 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.4139, + "step": 677 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.8683, + "step": 678 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.0055, + "step": 679 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 680 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 681 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 682 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 5.524, + "step": 683 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 684 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.9652, + "step": 685 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 8.5498, + "step": 686 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.6898, + "step": 687 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.1857, + "step": 688 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.7663, + "step": 689 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.8917, + "step": 690 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 691 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 692 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.2306, + "step": 693 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.0637, + "step": 694 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.9181, + "step": 695 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 696 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.0722, + "step": 697 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 698 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 5.1628, + "step": 699 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.0704, + "step": 700 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 5.7025, + "step": 701 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 702 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 8.3455, + "step": 703 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.2541, + "step": 704 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 705 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.0088, + "step": 706 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.6556, + "step": 707 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 708 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 709 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 710 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 711 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.9647, + "step": 712 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 713 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.573, + "step": 714 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 715 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 7.9199, + "step": 716 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 717 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.8125, + "step": 718 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 6.4286, + "step": 719 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.1997, + "step": 720 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.1364, + "step": 721 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.0859, + "step": 722 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.9088, + "step": 723 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 724 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.0877, + "step": 725 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 6.7622, + "step": 726 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.3234, + "step": 727 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 728 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.874, + "step": 729 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 730 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 8.474, + "step": 731 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 5.2689, + "step": 732 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 733 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.0912, + "step": 734 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 7.7931, + "step": 735 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 736 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.7319, + "step": 737 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 738 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 739 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.6641, + "step": 740 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 741 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.2979, + "step": 742 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 6.0577, + "step": 743 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 744 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 745 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 7.4101, + "step": 746 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.7423, + "step": 747 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 748 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 749 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 750 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 751 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.9603, + "step": 752 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 10.1696, + "step": 753 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.8995, + "step": 754 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 755 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 756 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 1.0336, + "step": 757 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.2638, + "step": 758 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 759 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 760 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.011, + "step": 761 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 762 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.422, + "step": 763 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 764 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.2189, + "step": 765 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 766 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 767 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 768 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.262, + "step": 769 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.91, + "step": 770 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 771 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 4.7263, + "step": 772 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 773 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 3.58, + "step": 774 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 0.0, + "step": 775 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.4226, + "step": 776 + }, + { + "epoch": 0.06, + "learning_rate": 0.0, + "loss": 2.4815, + "step": 777 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 778 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 779 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 780 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.0737, + "step": 781 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 3.718, + "step": 782 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 783 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 10.1924, + "step": 784 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.021, + "step": 785 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 7.6537, + "step": 786 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 787 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 788 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 4.0077, + "step": 789 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 3.1098, + "step": 790 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 791 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 4.9167, + "step": 792 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.0276, + "step": 793 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.9737, + "step": 794 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 2.8453, + "step": 795 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 796 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 797 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 9.3415, + "step": 798 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 3.7896, + "step": 799 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 3.7318, + "step": 800 + }, + { + "epoch": 0.07, + "eval_loss": NaN, + "eval_runtime": 39.7203, + "eval_samples_per_second": 62.94, + "eval_steps_per_second": 7.88, + "step": 800 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.6272, + "step": 801 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.1496, + "step": 802 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.9595, + "step": 803 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 804 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 7.9664, + "step": 805 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 806 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 807 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 808 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 809 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 810 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.3273, + "step": 811 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 7.559, + "step": 812 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 5.2586, + "step": 813 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.7329, + "step": 814 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 9.0739, + "step": 815 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 816 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 817 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 818 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 819 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 3.8752, + "step": 820 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.1763, + "step": 821 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 822 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 6.2554, + "step": 823 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 824 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 825 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 826 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 827 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 3.1005, + "step": 828 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 829 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 6.5268, + "step": 830 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 5.9297, + "step": 831 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.4135, + "step": 832 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 2.1028, + "step": 833 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 834 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 835 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 6.9263, + "step": 836 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.2022, + "step": 837 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.959, + "step": 838 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.9664, + "step": 839 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.3427, + "step": 840 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 841 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 842 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 843 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 6.0845, + "step": 844 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 845 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 846 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 847 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 6.8882, + "step": 848 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 849 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.9372, + "step": 850 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 4.9255, + "step": 851 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 7.6127, + "step": 852 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 853 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 854 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 855 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 856 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 4.0142, + "step": 857 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 2.4233, + "step": 858 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 859 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.0211, + "step": 860 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 861 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.8784, + "step": 862 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.9552, + "step": 863 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.0253, + "step": 864 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 865 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 866 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.9271, + "step": 867 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.8532, + "step": 868 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.1591, + "step": 869 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 870 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 871 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 10.9965, + "step": 872 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 873 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 874 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 875 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 876 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 2.0918, + "step": 877 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 878 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 2.0958, + "step": 879 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 880 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.1476, + "step": 881 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 882 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 883 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 884 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 885 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 8.6439, + "step": 886 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 887 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 888 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 889 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 890 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 1.5672, + "step": 891 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 2.1427, + "step": 892 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 893 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 0.0, + "step": 894 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 4.1489, + "step": 895 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 4.4625, + "step": 896 + }, + { + "epoch": 0.07, + "learning_rate": 0.0, + "loss": 7.2378, + "step": 897 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 898 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.1308, + "step": 899 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.1094, + "step": 900 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.7182, + "step": 901 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 902 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 903 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 8.5859, + "step": 904 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 8.4973, + "step": 905 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 906 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 907 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.826, + "step": 908 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 2.391, + "step": 909 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 910 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.1296, + "step": 911 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 912 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 913 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.146, + "step": 914 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.6115, + "step": 915 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.2052, + "step": 916 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 917 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 918 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.8033, + "step": 919 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 2.5541, + "step": 920 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 921 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 922 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 6.2779, + "step": 923 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 7.7966, + "step": 924 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.0796, + "step": 925 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 926 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 927 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 928 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.9291, + "step": 929 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.7955, + "step": 930 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 931 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 7.7681, + "step": 932 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.8267, + "step": 933 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 2.9167, + "step": 934 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 935 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 936 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 937 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 938 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 939 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 940 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 7.9761, + "step": 941 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 6.5365, + "step": 942 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.5079, + "step": 943 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 5.8209, + "step": 944 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 945 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 946 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.5998, + "step": 947 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 948 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 949 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 5.9, + "step": 950 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 7.1301, + "step": 951 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 952 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 5.4125, + "step": 953 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 2.1866, + "step": 954 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.7827, + "step": 955 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 956 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.7605, + "step": 957 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.7913, + "step": 958 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.3544, + "step": 959 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.851, + "step": 960 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 6.5161, + "step": 961 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 2.0298, + "step": 962 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 963 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 964 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.1867, + "step": 965 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 966 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 967 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 5.3603, + "step": 968 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.7546, + "step": 969 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 5.3911, + "step": 970 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.4996, + "step": 971 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 972 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 973 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 974 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 975 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 976 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 2.1277, + "step": 977 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.0641, + "step": 978 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 979 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.8771, + "step": 980 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 981 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 982 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 983 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.5656, + "step": 984 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 985 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 986 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 987 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 988 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.7806, + "step": 989 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 990 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 991 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 992 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.6191, + "step": 993 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 5.1858, + "step": 994 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.0849, + "step": 995 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 996 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.0527, + "step": 997 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.4309, + "step": 998 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 7.3899, + "step": 999 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1000 + }, + { + "epoch": 0.08, + "eval_loss": NaN, + "eval_runtime": 39.7753, + "eval_samples_per_second": 62.853, + "eval_steps_per_second": 7.869, + "step": 1000 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 7.126, + "step": 1001 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 6.5259, + "step": 1002 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1003 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1004 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 3.9845, + "step": 1005 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1006 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1007 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 2.104, + "step": 1008 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1009 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 1.9203, + "step": 1010 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1011 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1012 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 0.8637, + "step": 1013 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.7564, + "step": 1014 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 8.2807, + "step": 1015 + }, + { + "epoch": 0.08, + "learning_rate": 0.0, + "loss": 4.6532, + "step": 1016 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1017 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1018 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1019 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1020 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1021 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 2.3651, + "step": 1022 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1023 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.1331, + "step": 1024 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1025 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1026 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 2.9925, + "step": 1027 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.6441, + "step": 1028 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1029 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.8735, + "step": 1030 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1031 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.2519, + "step": 1032 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1033 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1034 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1035 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1036 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.9499, + "step": 1037 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 8.6162, + "step": 1038 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.5775, + "step": 1039 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1040 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1041 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.9575, + "step": 1042 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1043 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1044 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 7.5631, + "step": 1045 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.7992, + "step": 1046 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.9623, + "step": 1047 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1048 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 9.4018, + "step": 1049 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.3445, + "step": 1050 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 8.9074, + "step": 1051 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.5225, + "step": 1052 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.9795, + "step": 1053 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1054 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1055 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.5071, + "step": 1056 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.7272, + "step": 1057 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1058 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1059 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.6842, + "step": 1060 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1061 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.7171, + "step": 1062 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 2.092, + "step": 1063 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 9.223, + "step": 1064 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.6156, + "step": 1065 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 2.0437, + "step": 1066 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 5.8843, + "step": 1067 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1068 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.8849, + "step": 1069 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1070 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1071 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1072 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.1513, + "step": 1073 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 7.1966, + "step": 1074 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.0809, + "step": 1075 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.3983, + "step": 1076 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.6496, + "step": 1077 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1078 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1079 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 2.46, + "step": 1080 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 7.0054, + "step": 1081 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1082 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1083 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1084 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.9511, + "step": 1085 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1086 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.7416, + "step": 1087 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 7.572, + "step": 1088 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.0879, + "step": 1089 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1090 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.8783, + "step": 1091 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1092 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1093 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 5.0588, + "step": 1094 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1095 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.9946, + "step": 1096 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 3.6696, + "step": 1097 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.1833, + "step": 1098 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 6.0019, + "step": 1099 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.7567, + "step": 1100 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1101 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 5.2682, + "step": 1102 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1103 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1104 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1105 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1106 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.7897, + "step": 1107 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.7576, + "step": 1108 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1109 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1110 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1111 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 8.203, + "step": 1112 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 6.2424, + "step": 1113 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.3019, + "step": 1114 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 8.0364, + "step": 1115 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1116 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 7.5254, + "step": 1117 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1118 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 6.9517, + "step": 1119 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.1823, + "step": 1120 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 6.9807, + "step": 1121 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.6061, + "step": 1122 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1123 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.0169, + "step": 1124 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1125 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1126 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1127 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1128 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.1079, + "step": 1129 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1130 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1131 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 4.0705, + "step": 1132 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 1.0774, + "step": 1133 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 7.4625, + "step": 1134 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 6.2086, + "step": 1135 + }, + { + "epoch": 0.09, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1136 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.023, + "step": 1137 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.8596, + "step": 1138 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.399, + "step": 1139 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 5.4579, + "step": 1140 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.1793, + "step": 1141 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.9432, + "step": 1142 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1143 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 6.8042, + "step": 1144 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1145 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1146 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1147 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 6.4026, + "step": 1148 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.8947, + "step": 1149 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.7772, + "step": 1150 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.8168, + "step": 1151 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 4.733, + "step": 1152 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.3644, + "step": 1153 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.7673, + "step": 1154 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1155 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1156 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 4.2358, + "step": 1157 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 6.8441, + "step": 1158 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1159 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1160 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 6.9429, + "step": 1161 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1162 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1163 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.8386, + "step": 1164 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 5.2166, + "step": 1165 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 9.7054, + "step": 1166 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1167 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1168 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 4.6276, + "step": 1169 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.0898, + "step": 1170 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.004, + "step": 1171 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 7.5493, + "step": 1172 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 7.9416, + "step": 1173 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1174 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1175 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 5.2307, + "step": 1176 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1177 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.1535, + "step": 1178 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1179 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1180 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1181 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1182 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.1231, + "step": 1183 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1184 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.1994, + "step": 1185 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1186 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1187 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1188 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1189 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.1505, + "step": 1190 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1191 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1192 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.9432, + "step": 1193 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1194 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.8583, + "step": 1195 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.2744, + "step": 1196 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.9344, + "step": 1197 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1198 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 4.0578, + "step": 1199 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1200 + }, + { + "epoch": 0.1, + "eval_loss": NaN, + "eval_runtime": 40.5076, + "eval_samples_per_second": 61.717, + "eval_steps_per_second": 7.727, + "step": 1200 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 9.2779, + "step": 1201 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1202 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1203 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1204 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.3655, + "step": 1205 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1206 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1207 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1208 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1209 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1210 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1211 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1212 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.0262, + "step": 1213 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 5.236, + "step": 1214 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 8.3536, + "step": 1215 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1216 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 6.3298, + "step": 1217 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1218 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1219 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.7762, + "step": 1220 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1221 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.4798, + "step": 1222 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 4.7311, + "step": 1223 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.8464, + "step": 1224 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1225 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1226 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.8793, + "step": 1227 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1228 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1229 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 5.4276, + "step": 1230 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1231 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.9338, + "step": 1232 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.0619, + "step": 1233 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.2068, + "step": 1234 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 4.4587, + "step": 1235 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.0894, + "step": 1236 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1237 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 5.6093, + "step": 1238 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.7924, + "step": 1239 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1240 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.3638, + "step": 1241 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1242 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.1161, + "step": 1243 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 9.0741, + "step": 1244 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1245 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1246 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 5.2718, + "step": 1247 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1248 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.8295, + "step": 1249 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 1.0564, + "step": 1250 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1251 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 3.4066, + "step": 1252 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 10.3074, + "step": 1253 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 8.9002, + "step": 1254 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1255 + }, + { + "epoch": 0.1, + "learning_rate": 0.0, + "loss": 2.0471, + "step": 1256 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 11.2098, + "step": 1257 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 6.9082, + "step": 1258 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.9038, + "step": 1259 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1260 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1261 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 6.0135, + "step": 1262 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1263 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.0446, + "step": 1264 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1265 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.4024, + "step": 1266 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.8401, + "step": 1267 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 6.8595, + "step": 1268 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.873, + "step": 1269 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.1116, + "step": 1270 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1271 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.5228, + "step": 1272 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1273 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1274 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1275 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.6616, + "step": 1276 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 4.13, + "step": 1277 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1278 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.9729, + "step": 1279 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.0702, + "step": 1280 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1281 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1282 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 7.2575, + "step": 1283 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.4752, + "step": 1284 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1285 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 6.5036, + "step": 1286 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.0173, + "step": 1287 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1288 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.7089, + "step": 1289 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.0601, + "step": 1290 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.9847, + "step": 1291 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.82, + "step": 1292 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.9944, + "step": 1293 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.0917, + "step": 1294 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1295 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 6.4709, + "step": 1296 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 4.6074, + "step": 1297 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.0276, + "step": 1298 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1299 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1300 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1301 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.9882, + "step": 1302 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 4.4335, + "step": 1303 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.9544, + "step": 1304 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1305 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.2525, + "step": 1306 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1307 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 7.6738, + "step": 1308 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1309 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.1261, + "step": 1310 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1311 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1312 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.9128, + "step": 1313 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.5228, + "step": 1314 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1315 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1316 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1317 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.0269, + "step": 1318 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.838, + "step": 1319 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1320 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1321 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.1123, + "step": 1322 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1323 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 4.6046, + "step": 1324 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.4029, + "step": 1325 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1326 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.2083, + "step": 1327 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1328 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.8843, + "step": 1329 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1330 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.9025, + "step": 1331 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.6415, + "step": 1332 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.1407, + "step": 1333 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 7.6286, + "step": 1334 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 8.6104, + "step": 1335 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1336 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 7.0218, + "step": 1337 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1338 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.846, + "step": 1339 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 7.4739, + "step": 1340 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1341 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1342 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.4841, + "step": 1343 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1344 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.1392, + "step": 1345 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1346 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1347 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1348 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1349 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 5.049, + "step": 1350 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1351 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.9964, + "step": 1352 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.9134, + "step": 1353 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.9313, + "step": 1354 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1355 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 8.3294, + "step": 1356 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 7.9764, + "step": 1357 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1358 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1359 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 8.2791, + "step": 1360 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1361 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.8505, + "step": 1362 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 7.2588, + "step": 1363 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1364 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.0424, + "step": 1365 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 8.7602, + "step": 1366 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 4.0012, + "step": 1367 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 6.1672, + "step": 1368 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 5.3927, + "step": 1369 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 3.4777, + "step": 1370 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 6.8195, + "step": 1371 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 1.8054, + "step": 1372 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 8.8954, + "step": 1373 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 8.5657, + "step": 1374 + }, + { + "epoch": 0.11, + "learning_rate": 0.0, + "loss": 2.0889, + "step": 1375 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1376 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1377 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1378 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1379 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1380 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 12.9162, + "step": 1381 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1382 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 2.5925, + "step": 1383 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1384 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1385 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 8.7465, + "step": 1386 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1387 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1388 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.9574, + "step": 1389 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.9249, + "step": 1390 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 4.0046, + "step": 1391 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1392 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.6704, + "step": 1393 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 6.7238, + "step": 1394 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 3.6281, + "step": 1395 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.9162, + "step": 1396 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1397 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1398 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1399 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1400 + }, + { + "epoch": 0.12, + "eval_loss": NaN, + "eval_runtime": 40.6342, + "eval_samples_per_second": 61.524, + "eval_steps_per_second": 7.703, + "step": 1400 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1401 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.7967, + "step": 1402 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.1341, + "step": 1403 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1404 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 6.9185, + "step": 1405 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 5.9346, + "step": 1406 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 3.226, + "step": 1407 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1408 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 5.1887, + "step": 1409 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1410 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1411 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1412 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 5.9302, + "step": 1413 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1414 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 4.697, + "step": 1415 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 3.0753, + "step": 1416 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 2.3678, + "step": 1417 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 2.2983, + "step": 1418 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 4.8017, + "step": 1419 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 5.6217, + "step": 1420 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 7.834, + "step": 1421 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1422 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 4.4776, + "step": 1423 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 7.9156, + "step": 1424 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1425 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 3.8127, + "step": 1426 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1427 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 2.3103, + "step": 1428 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 8.9208, + "step": 1429 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1430 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1431 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1432 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1433 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 2.1302, + "step": 1434 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 2.0479, + "step": 1435 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 2.4117, + "step": 1436 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 4.6789, + "step": 1437 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.2351, + "step": 1438 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1439 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1440 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 3.7018, + "step": 1441 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1442 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1443 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1444 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.9443, + "step": 1445 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1446 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1447 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 8.4794, + "step": 1448 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1449 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1450 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.9337, + "step": 1451 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 8.3651, + "step": 1452 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1453 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 7.2965, + "step": 1454 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1455 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1456 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 0.0, + "step": 1457 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 1.6543, + "step": 1458 + }, + { + "epoch": 0.12, + "learning_rate": 0.0, + "loss": 4.2454, + "step": 1459 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 4.0574, + "step": 1460 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1461 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.3218, + "step": 1462 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1463 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 4.8892, + "step": 1464 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1465 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 6.0283, + "step": 1466 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 4.8051, + "step": 1467 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1468 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1469 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 4.9304, + "step": 1470 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 3.8502, + "step": 1471 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.9266, + "step": 1472 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.961, + "step": 1473 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1474 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1475 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1476 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1477 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.075, + "step": 1478 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1479 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1480 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1481 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1482 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1483 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1484 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1485 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1486 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 4.5503, + "step": 1487 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.8559, + "step": 1488 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 9.8207, + "step": 1489 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1490 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1491 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 4.6777, + "step": 1492 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1493 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1031, + "step": 1494 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 8.4074, + "step": 1495 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1496 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1497 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 3.7996, + "step": 1498 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 3.8571, + "step": 1499 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1500 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1501 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 5.7037, + "step": 1502 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 4.0379, + "step": 1503 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 7.6189, + "step": 1504 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1505 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1506 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1507 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1508 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1509 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1510 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.8707, + "step": 1511 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.9858, + "step": 1512 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 8.8587, + "step": 1513 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 8.0279, + "step": 1514 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.1587, + "step": 1515 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.58, + "step": 1516 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.9997, + "step": 1517 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1518 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.9928, + "step": 1519 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 8.2781, + "step": 1520 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1521 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.0271, + "step": 1522 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1523 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1524 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1525 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.5378, + "step": 1526 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1527 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1528 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1529 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 8.0703, + "step": 1530 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 9.3978, + "step": 1531 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 4.8568, + "step": 1532 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1533 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1534 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1535 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1536 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.861, + "step": 1537 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 3.2328, + "step": 1538 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 6.6883, + "step": 1539 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3201, + "step": 1540 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1541 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1542 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1543 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1544 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 7.3011, + "step": 1545 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1546 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1547 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1548 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1549 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.9895, + "step": 1550 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1551 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.7497, + "step": 1552 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1553 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 5.6844, + "step": 1554 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 7.2863, + "step": 1555 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.0167, + "step": 1556 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.212, + "step": 1557 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 4.5155, + "step": 1558 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 3.7363, + "step": 1559 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 4.2264, + "step": 1560 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.9783, + "step": 1561 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 5.5095, + "step": 1562 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 5.8664, + "step": 1563 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 9.1433, + "step": 1564 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 6.8911, + "step": 1565 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1566 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.7464, + "step": 1567 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 10.283, + "step": 1568 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1569 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 4.3775, + "step": 1570 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1571 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.0561, + "step": 1572 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.8503, + "step": 1573 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1574 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2595, + "step": 1575 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1576 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.1293, + "step": 1577 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1578 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1579 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1580 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1581 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 3.8165, + "step": 1582 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 8.7623, + "step": 1583 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1584 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.9242, + "step": 1585 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1586 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1587 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 8.7379, + "step": 1588 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1589 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1590 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.0197, + "step": 1591 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.9791, + "step": 1592 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1593 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.187, + "step": 1594 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.6714, + "step": 1595 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 10.5386, + "step": 1596 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1597 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1598 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1599 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 6.3835, + "step": 1600 + }, + { + "epoch": 0.13, + "eval_loss": NaN, + "eval_runtime": 40.1112, + "eval_samples_per_second": 62.327, + "eval_steps_per_second": 7.803, + "step": 1600 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1601 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1602 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 7.0786, + "step": 1603 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1604 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 4.5455, + "step": 1605 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.5727, + "step": 1606 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1607 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1608 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1609 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.087, + "step": 1610 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1611 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 8.8549, + "step": 1612 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 4.1051, + "step": 1613 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1614 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1615 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.8271, + "step": 1616 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.9627, + "step": 1617 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.8557, + "step": 1618 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.9951, + "step": 1619 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.3235, + "step": 1620 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.2541, + "step": 1621 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 8.0934, + "step": 1622 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1623 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.3555, + "step": 1624 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1625 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.0001, + "step": 1626 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.841, + "step": 1627 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 6.8558, + "step": 1628 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.0008, + "step": 1629 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.4195, + "step": 1630 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1631 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1632 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1633 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.0312, + "step": 1634 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1635 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.1683, + "step": 1636 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.9762, + "step": 1637 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.0329, + "step": 1638 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1639 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.0032, + "step": 1640 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.9611, + "step": 1641 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 9.4417, + "step": 1642 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1643 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.9953, + "step": 1644 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 4.7023, + "step": 1645 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1646 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1647 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.4302, + "step": 1648 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.0613, + "step": 1649 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 5.2404, + "step": 1650 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1651 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1652 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1653 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.0498, + "step": 1654 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.8485, + "step": 1655 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2704, + "step": 1656 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1657 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1658 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 8.5877, + "step": 1659 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1660 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1661 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.861, + "step": 1662 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1663 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1664 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1665 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 4.662, + "step": 1666 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.8453, + "step": 1667 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1668 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1669 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1670 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1671 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.0665, + "step": 1672 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1673 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.9394, + "step": 1674 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1675 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.057, + "step": 1676 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 4.4447, + "step": 1677 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 8.369, + "step": 1678 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.6342, + "step": 1679 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1680 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.0905, + "step": 1681 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.3589, + "step": 1682 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 4.1283, + "step": 1683 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.7191, + "step": 1684 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1685 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.4999, + "step": 1686 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 4.0252, + "step": 1687 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 4.4345, + "step": 1688 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1689 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1690 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.7948, + "step": 1691 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.8638, + "step": 1692 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1693 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 5.0637, + "step": 1694 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.7847, + "step": 1695 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1696 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.7352, + "step": 1697 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.3134, + "step": 1698 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.8921, + "step": 1699 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1700 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1701 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 4.3084, + "step": 1702 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.8755, + "step": 1703 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.2793, + "step": 1704 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.548, + "step": 1705 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1706 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1707 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 8.3701, + "step": 1708 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1709 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.5562, + "step": 1710 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1711 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1712 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 5.173, + "step": 1713 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.388, + "step": 1714 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1715 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 10.2159, + "step": 1716 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.7795, + "step": 1717 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1718 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1719 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 8.8195, + "step": 1720 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1721 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1722 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.8659, + "step": 1723 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 8.3746, + "step": 1724 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1725 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.5877, + "step": 1726 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1727 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.6338, + "step": 1728 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1729 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.1333, + "step": 1730 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.9133, + "step": 1731 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1732 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2122, + "step": 1733 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 7.6463, + "step": 1734 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1735 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1736 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.9654, + "step": 1737 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1738 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.157, + "step": 1739 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1740 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.2214, + "step": 1741 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.7989, + "step": 1742 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 8.9004, + "step": 1743 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1744 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.1091, + "step": 1745 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.9518, + "step": 1746 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1747 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1748 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.5371, + "step": 1749 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.784, + "step": 1750 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.391, + "step": 1751 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.9685, + "step": 1752 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.8416, + "step": 1753 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1754 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1755 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1756 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1757 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.9628, + "step": 1758 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.0829, + "step": 1759 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1760 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.9961, + "step": 1761 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1762 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1763 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.8731, + "step": 1764 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1765 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1766 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1767 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1768 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1769 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1770 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.7915, + "step": 1771 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.8521, + "step": 1772 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.7161, + "step": 1773 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 8.122, + "step": 1774 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1553, + "step": 1775 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1776 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.6028, + "step": 1777 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1778 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.0527, + "step": 1779 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.5964, + "step": 1780 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 6.5944, + "step": 1781 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1782 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1783 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1784 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.1699, + "step": 1785 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.0385, + "step": 1786 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1787 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.4697, + "step": 1788 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1789 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1790 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.0817, + "step": 1791 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.7409, + "step": 1792 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.7403, + "step": 1793 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1794 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1795 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1796 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.2171, + "step": 1797 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.0876, + "step": 1798 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.9667, + "step": 1799 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.0341, + "step": 1800 + }, + { + "epoch": 0.15, + "eval_loss": NaN, + "eval_runtime": 40.537, + "eval_samples_per_second": 61.672, + "eval_steps_per_second": 7.721, + "step": 1800 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1801 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.8069, + "step": 1802 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.8669, + "step": 1803 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1804 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 5.3575, + "step": 1805 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1806 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1807 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.2747, + "step": 1808 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1809 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 6.1007, + "step": 1810 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 6.8699, + "step": 1811 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.9786, + "step": 1812 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.8645, + "step": 1813 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.6997, + "step": 1814 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1815 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.9946, + "step": 1816 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.5711, + "step": 1817 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.9733, + "step": 1818 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.8821, + "step": 1819 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.1312, + "step": 1820 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 6.5441, + "step": 1821 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.9795, + "step": 1822 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1823 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1824 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.2906, + "step": 1825 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.5827, + "step": 1826 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.5492, + "step": 1827 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.2948, + "step": 1828 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.9056, + "step": 1829 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.7594, + "step": 1830 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1831 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.8359, + "step": 1832 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 6.031, + "step": 1833 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.9002, + "step": 1834 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.1035, + "step": 1835 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.8436, + "step": 1836 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.5846, + "step": 1837 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.8566, + "step": 1838 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 6.8985, + "step": 1839 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.9894, + "step": 1840 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 9.2911, + "step": 1841 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 5.6001, + "step": 1842 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 9.4974, + "step": 1843 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 4.1058, + "step": 1844 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.9022, + "step": 1845 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.7838, + "step": 1846 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.8696, + "step": 1847 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.4088, + "step": 1848 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1849 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1850 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1851 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 7.5251, + "step": 1852 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1853 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1854 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1855 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1856 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 3.5562, + "step": 1857 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 6.57, + "step": 1858 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1859 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1860 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.1332, + "step": 1861 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 4.0688, + "step": 1862 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 8.1113, + "step": 1863 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.9791, + "step": 1864 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 8.7321, + "step": 1865 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 3.3053, + "step": 1866 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1867 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1868 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1869 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1870 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1871 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1872 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1873 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1874 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3289, + "step": 1875 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1876 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1877 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 4.2686, + "step": 1878 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 3.8808, + "step": 1879 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1880 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 9.8471, + "step": 1881 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1882 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.8947, + "step": 1883 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1884 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 7.7841, + "step": 1885 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 8.5268, + "step": 1886 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.2312, + "step": 1887 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.991, + "step": 1888 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1889 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 4.1965, + "step": 1890 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1891 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1892 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8115, + "step": 1893 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8435, + "step": 1894 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1895 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 4.7103, + "step": 1896 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 0.0, + "step": 1897 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 4.0945, + "step": 1898 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 2.0819, + "step": 1899 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1900 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1901 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 3.8191, + "step": 1902 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1903 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.5763, + "step": 1904 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 3.5704, + "step": 1905 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 9.0099, + "step": 1906 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.9589, + "step": 1907 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1908 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 5.0876, + "step": 1909 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1910 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1911 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 9.0176, + "step": 1912 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 5.222, + "step": 1913 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.9708, + "step": 1914 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.9362, + "step": 1915 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 8.2756, + "step": 1916 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1917 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1918 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1919 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1920 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.9726, + "step": 1921 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1922 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1923 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1924 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1925 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.6924, + "step": 1926 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.9711, + "step": 1927 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 2.2163, + "step": 1928 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.8312, + "step": 1929 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1930 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 4.1714, + "step": 1931 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1932 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 2.772, + "step": 1933 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 9.125, + "step": 1934 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1935 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1936 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1937 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 2.0261, + "step": 1938 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1939 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1940 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.8717, + "step": 1941 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1942 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 7.3458, + "step": 1943 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1944 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1945 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.0197, + "step": 1946 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1947 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 11.7447, + "step": 1948 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.0022, + "step": 1949 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.5595, + "step": 1950 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1951 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 6.4861, + "step": 1952 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 3.1559, + "step": 1953 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1954 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 4.7276, + "step": 1955 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1956 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.8296, + "step": 1957 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 2.2173, + "step": 1958 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1959 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1960 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1961 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1962 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1963 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1964 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 3.6105, + "step": 1965 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1966 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 3.7538, + "step": 1967 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.9301, + "step": 1968 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 6.8095, + "step": 1969 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 1.6151, + "step": 1970 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 3.5543, + "step": 1971 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1972 + }, + { + "epoch": 0.16, + "learning_rate": 0.0002, + "loss": 0.9428, + "step": 1973 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1974 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1975 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.16, + "step": 1976 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1977 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.3631, + "step": 1978 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1979 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1980 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.4648, + "step": 1981 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1982 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1983 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1984 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.7075, + "step": 1985 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 5.1815, + "step": 1986 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.1704, + "step": 1987 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1988 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.6638, + "step": 1989 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1990 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.5126, + "step": 1991 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1992 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1993 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 5.9859, + "step": 1994 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 6.9905, + "step": 1995 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1996 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.0736, + "step": 1997 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 1998 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.2359, + "step": 1999 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2000 + }, + { + "epoch": 0.17, + "eval_loss": NaN, + "eval_runtime": 40.539, + "eval_samples_per_second": 61.669, + "eval_steps_per_second": 7.721, + "step": 2000 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.9418, + "step": 2001 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.8394, + "step": 2002 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2003 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 7.8507, + "step": 2004 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2005 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.7224, + "step": 2006 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.9422, + "step": 2007 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2008 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 9.7727, + "step": 2009 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.5401, + "step": 2010 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 8.7443, + "step": 2011 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2012 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2013 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 5.926, + "step": 2014 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2015 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.5064, + "step": 2016 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2017 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 9.4897, + "step": 2018 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2019 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.108, + "step": 2020 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 7.9232, + "step": 2021 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2022 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2023 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 7.8767, + "step": 2024 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.6759, + "step": 2025 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.386, + "step": 2026 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.4356, + "step": 2027 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2028 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2029 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.021, + "step": 2030 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.3758, + "step": 2031 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.2341, + "step": 2032 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.1842, + "step": 2033 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2034 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.3286, + "step": 2035 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.2179, + "step": 2036 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 5.5673, + "step": 2037 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2038 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.2001, + "step": 2039 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2040 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.0576, + "step": 2041 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.3177, + "step": 2042 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2043 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2044 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.0142, + "step": 2045 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 6.9621, + "step": 2046 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2047 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2048 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.0141, + "step": 2049 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2050 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2051 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2052 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.163, + "step": 2053 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.5526, + "step": 2054 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.5034, + "step": 2055 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.1837, + "step": 2056 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2057 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.6334, + "step": 2058 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2059 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.3672, + "step": 2060 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.4739, + "step": 2061 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.3353, + "step": 2062 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2063 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2064 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 3.3419, + "step": 2065 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.6996, + "step": 2066 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.1727, + "step": 2067 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.233, + "step": 2068 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.0011, + "step": 2069 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2070 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.8805, + "step": 2071 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2072 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.1102, + "step": 2073 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.4722, + "step": 2074 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2075 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2076 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 2.7206, + "step": 2077 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2078 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2079 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2080 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.8169, + "step": 2081 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2082 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2083 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2084 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2085 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 10.894, + "step": 2086 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.7109, + "step": 2087 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 5.2367, + "step": 2088 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2089 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2090 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2091 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 4.4708, + "step": 2092 + }, + { + "epoch": 0.17, + "learning_rate": 0.0002, + "loss": 1.9425, + "step": 2093 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 7.5408, + "step": 2094 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.3328, + "step": 2095 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 2.0791, + "step": 2096 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.2353, + "step": 2097 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 2.4411, + "step": 2098 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.071, + "step": 2099 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2100 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2101 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2102 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.739, + "step": 2103 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2104 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2105 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.9213, + "step": 2106 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.0095, + "step": 2107 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2108 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.5195, + "step": 2109 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2110 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2111 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.1195, + "step": 2112 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2113 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2114 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.0512, + "step": 2115 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.188, + "step": 2116 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.1846, + "step": 2117 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.8046, + "step": 2118 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2119 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.8417, + "step": 2120 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.1611, + "step": 2121 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 5.732, + "step": 2122 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.4547, + "step": 2123 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.9718, + "step": 2124 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 5.5802, + "step": 2125 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2126 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2127 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 9.0782, + "step": 2128 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.3314, + "step": 2129 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2130 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2131 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2132 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.6401, + "step": 2133 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 8.2666, + "step": 2134 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.9472, + "step": 2135 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2136 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.0474, + "step": 2137 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.9097, + "step": 2138 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.5566, + "step": 2139 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.0597, + "step": 2140 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2141 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 2.6885, + "step": 2142 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 6.9899, + "step": 2143 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2144 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.9948, + "step": 2145 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2146 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 7.3626, + "step": 2147 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 6.3249, + "step": 2148 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2149 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.4048, + "step": 2150 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 4.3873, + "step": 2151 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2152 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2153 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2154 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.0248, + "step": 2155 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2156 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 7.5979, + "step": 2157 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2158 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.0563, + "step": 2159 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2160 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2161 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 7.8736, + "step": 2162 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2163 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.8478, + "step": 2164 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2165 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2166 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.721, + "step": 2167 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.9816, + "step": 2168 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.4013, + "step": 2169 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2170 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2171 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2172 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 2.0435, + "step": 2173 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.4208, + "step": 2174 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2175 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2176 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2177 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 7.5866, + "step": 2178 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2179 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.9424, + "step": 2180 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 2.3725, + "step": 2181 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 7.1582, + "step": 2182 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2183 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.8765, + "step": 2184 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2185 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2186 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 7.956, + "step": 2187 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 8.9578, + "step": 2188 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2189 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2190 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2191 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2192 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2193 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2194 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2195 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2196 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2197 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 8.3059, + "step": 2198 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 5.8832, + "step": 2199 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 9.3721, + "step": 2200 + }, + { + "epoch": 0.18, + "eval_loss": NaN, + "eval_runtime": 40.5566, + "eval_samples_per_second": 61.642, + "eval_steps_per_second": 7.718, + "step": 2200 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 1.8958, + "step": 2201 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.8278, + "step": 2202 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.8256, + "step": 2203 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2204 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2205 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2206 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 5.8118, + "step": 2207 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.419, + "step": 2208 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2209 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 3.6253, + "step": 2210 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 6.1069, + "step": 2211 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 5.8024, + "step": 2212 + }, + { + "epoch": 0.18, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2213 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.4576, + "step": 2214 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2215 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2216 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.4693, + "step": 2217 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2218 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2219 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 7.5614, + "step": 2220 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2221 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2222 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.4373, + "step": 2223 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2224 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.4625, + "step": 2225 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.1101, + "step": 2226 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2227 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2228 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.5125, + "step": 2229 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2230 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2231 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.3168, + "step": 2232 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.5373, + "step": 2233 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.9301, + "step": 2234 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.0241, + "step": 2235 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.9019, + "step": 2236 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.8784, + "step": 2237 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.4654, + "step": 2238 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2239 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2240 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 3.3726, + "step": 2241 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.9182, + "step": 2242 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.0418, + "step": 2243 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.8585, + "step": 2244 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 3.5727, + "step": 2245 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 13.3402, + "step": 2246 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2247 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2248 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2249 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 9.8827, + "step": 2250 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.9707, + "step": 2251 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.9059, + "step": 2252 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2253 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 3.2547, + "step": 2254 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 7.708, + "step": 2255 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 11.4131, + "step": 2256 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2257 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2258 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2259 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.9589, + "step": 2260 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.8529, + "step": 2261 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.7707, + "step": 2262 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2263 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.5355, + "step": 2264 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2265 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2266 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2267 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.441, + "step": 2268 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2269 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2270 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2271 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.1421, + "step": 2272 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2273 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.7311, + "step": 2274 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2275 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2276 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.1228, + "step": 2277 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 7.0421, + "step": 2278 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2279 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.0384, + "step": 2280 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2281 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 5.5248, + "step": 2282 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 8.8509, + "step": 2283 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 4.1379, + "step": 2284 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2285 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2286 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2287 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 7.4505, + "step": 2288 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2289 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.1903, + "step": 2290 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2291 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2292 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2293 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 7.6748, + "step": 2294 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2295 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 3.3294, + "step": 2296 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.7128, + "step": 2297 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2298 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2299 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2300 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2301 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 7.7046, + "step": 2302 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 3.534, + "step": 2303 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2304 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2305 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2306 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2307 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 3.6957, + "step": 2308 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2309 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 8.8196, + "step": 2310 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2311 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.1228, + "step": 2312 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.9658, + "step": 2313 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 9.1198, + "step": 2314 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.0228, + "step": 2315 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2316 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2317 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.8441, + "step": 2318 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.9394, + "step": 2319 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2320 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.0318, + "step": 2321 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 7.8186, + "step": 2322 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.9209, + "step": 2323 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.1612, + "step": 2324 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2325 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 2.3382, + "step": 2326 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 5.8183, + "step": 2327 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 1.8966, + "step": 2328 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2329 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2330 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2331 + }, + { + "epoch": 0.19, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2332 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2333 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2334 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2335 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2336 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 6.1441, + "step": 2337 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.9675, + "step": 2338 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.2072, + "step": 2339 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.8852, + "step": 2340 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.2452, + "step": 2341 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2342 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.2979, + "step": 2343 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.0999, + "step": 2344 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2345 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.6074, + "step": 2346 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2347 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2348 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.098, + "step": 2349 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2350 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 4.6627, + "step": 2351 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 9.6794, + "step": 2352 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2353 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.5295, + "step": 2354 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.9978, + "step": 2355 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.1335, + "step": 2356 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 5.2715, + "step": 2357 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.4174, + "step": 2358 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2359 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2360 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.9282, + "step": 2361 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 8.8612, + "step": 2362 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2363 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2364 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.0204, + "step": 2365 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.6662, + "step": 2366 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 6.844, + "step": 2367 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.6647, + "step": 2368 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2369 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.6873, + "step": 2370 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 4.1384, + "step": 2371 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 6.2732, + "step": 2372 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2373 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2374 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2375 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.8132, + "step": 2376 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2377 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2378 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2379 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 6.4261, + "step": 2380 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 5.9811, + "step": 2381 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2382 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 5.9362, + "step": 2383 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 8.5372, + "step": 2384 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.0226, + "step": 2385 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.5818, + "step": 2386 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 4.0953, + "step": 2387 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2388 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.7708, + "step": 2389 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2390 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 8.8161, + "step": 2391 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 11.2816, + "step": 2392 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2393 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.9308, + "step": 2394 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.6273, + "step": 2395 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2396 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.8652, + "step": 2397 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.1744, + "step": 2398 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 5.4529, + "step": 2399 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 4.9674, + "step": 2400 + }, + { + "epoch": 0.2, + "eval_loss": NaN, + "eval_runtime": 40.4247, + "eval_samples_per_second": 61.843, + "eval_steps_per_second": 7.743, + "step": 2400 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 6.4038, + "step": 2401 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2402 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.8799, + "step": 2403 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.8484, + "step": 2404 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2405 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.0907, + "step": 2406 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2407 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2408 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.9481, + "step": 2409 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2410 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2411 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.6937, + "step": 2412 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2413 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2414 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2415 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2416 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.0498, + "step": 2417 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.0064, + "step": 2418 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2419 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2420 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2421 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2422 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 5.0656, + "step": 2423 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2424 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.9786, + "step": 2425 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 5.374, + "step": 2426 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.8357, + "step": 2427 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2428 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.9484, + "step": 2429 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 8.4699, + "step": 2430 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.4053, + "step": 2431 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2432 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2433 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.9256, + "step": 2434 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2435 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2436 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.8755, + "step": 2437 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.6275, + "step": 2438 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2439 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2440 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 2.4455, + "step": 2441 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.7025, + "step": 2442 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.9542, + "step": 2443 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2444 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.9866, + "step": 2445 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 4.2599, + "step": 2446 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 3.2227, + "step": 2447 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 9.7959, + "step": 2448 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 1.7523, + "step": 2449 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 7.4486, + "step": 2450 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2451 + }, + { + "epoch": 0.2, + "learning_rate": 0.0002, + "loss": 0.9262, + "step": 2452 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.6725, + "step": 2453 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.708, + "step": 2454 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.7774, + "step": 2455 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.7481, + "step": 2456 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 4.2318, + "step": 2457 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.0553, + "step": 2458 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 4.2331, + "step": 2459 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 7.413, + "step": 2460 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.9271, + "step": 2461 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2462 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2463 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.0081, + "step": 2464 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2465 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2466 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2467 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.9584, + "step": 2468 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 7.422, + "step": 2469 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2470 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2471 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 5.8385, + "step": 2472 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2473 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2474 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 2.0285, + "step": 2475 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2476 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2477 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.7242, + "step": 2478 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.8033, + "step": 2479 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.8727, + "step": 2480 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2481 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2482 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.5117, + "step": 2483 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2484 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 5.1812, + "step": 2485 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.4066, + "step": 2486 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2487 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.9874, + "step": 2488 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.1083, + "step": 2489 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 4.1749, + "step": 2490 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.4568, + "step": 2491 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 7.015, + "step": 2492 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.7583, + "step": 2493 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 6.1934, + "step": 2494 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.0875, + "step": 2495 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 4.5579, + "step": 2496 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.8978, + "step": 2497 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2498 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2499 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2500 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 2.1529, + "step": 2501 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2502 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2503 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2504 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.9208, + "step": 2505 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 7.9981, + "step": 2506 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2507 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2508 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.7182, + "step": 2509 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2510 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2511 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2512 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2513 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 7.1927, + "step": 2514 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2515 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.9084, + "step": 2516 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2517 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 6.4378, + "step": 2518 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2519 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 7.2032, + "step": 2520 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.8313, + "step": 2521 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2522 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2523 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.919, + "step": 2524 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.898, + "step": 2525 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2526 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2527 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2528 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 11.1573, + "step": 2529 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2530 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.9724, + "step": 2531 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2532 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2533 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2534 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 4.6064, + "step": 2535 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2536 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 8.3287, + "step": 2537 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.244, + "step": 2538 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2539 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 2.1925, + "step": 2540 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.2108, + "step": 2541 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 9.6359, + "step": 2542 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.9023, + "step": 2543 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2544 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.0953, + "step": 2545 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2546 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2547 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 3.7975, + "step": 2548 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2549 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2550 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.8035, + "step": 2551 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 4.168, + "step": 2552 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2553 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.6792, + "step": 2554 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2555 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2556 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2557 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 5.0372, + "step": 2558 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2559 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 4.1409, + "step": 2560 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2561 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 2.4112, + "step": 2562 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2563 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 5.8417, + "step": 2564 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 8.5365, + "step": 2565 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2566 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2567 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 1.1437, + "step": 2568 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2569 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2570 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 8.2271, + "step": 2571 + }, + { + "epoch": 0.21, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2572 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 4.0178, + "step": 2573 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.0117, + "step": 2574 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2575 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2576 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.102, + "step": 2577 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.1872, + "step": 2578 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 5.869, + "step": 2579 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.6289, + "step": 2580 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.8258, + "step": 2581 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 7.6271, + "step": 2582 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.731, + "step": 2583 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2584 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2585 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2586 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.5419, + "step": 2587 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 5.4358, + "step": 2588 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.5963, + "step": 2589 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 10.3862, + "step": 2590 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2591 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 11.2001, + "step": 2592 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2593 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2594 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2595 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2596 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.491, + "step": 2597 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2598 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2599 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2600 + }, + { + "epoch": 0.22, + "eval_loss": NaN, + "eval_runtime": 40.6577, + "eval_samples_per_second": 61.489, + "eval_steps_per_second": 7.698, + "step": 2600 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2601 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.6913, + "step": 2602 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.803, + "step": 2603 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.7523, + "step": 2604 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.0252, + "step": 2605 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.8348, + "step": 2606 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2607 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.5145, + "step": 2608 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2609 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2610 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2611 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2612 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2613 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2614 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.7752, + "step": 2615 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.8678, + "step": 2616 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2617 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2618 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 4.038, + "step": 2619 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 7.8594, + "step": 2620 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.4358, + "step": 2621 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2622 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2623 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.9148, + "step": 2624 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2625 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 8.1582, + "step": 2626 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 6.4487, + "step": 2627 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2628 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2629 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2630 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.4455, + "step": 2631 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.9423, + "step": 2632 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2633 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 4.7241, + "step": 2634 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.7777, + "step": 2635 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 7.3335, + "step": 2636 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2637 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.094, + "step": 2638 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.3603, + "step": 2639 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 4.6802, + "step": 2640 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.0767, + "step": 2641 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 7.1912, + "step": 2642 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.1427, + "step": 2643 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2644 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 4.3373, + "step": 2645 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 3.3629, + "step": 2646 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2647 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 7.9094, + "step": 2648 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2649 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.8491, + "step": 2650 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2651 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2652 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2653 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2654 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 7.5013, + "step": 2655 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 7.5407, + "step": 2656 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2657 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2658 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2659 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2660 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.4204, + "step": 2661 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.5769, + "step": 2662 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2663 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2664 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.7565, + "step": 2665 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 8.6103, + "step": 2666 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.8357, + "step": 2667 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 9.3247, + "step": 2668 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2669 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2670 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2671 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 9.9135, + "step": 2672 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 2.0227, + "step": 2673 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2674 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.8794, + "step": 2675 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2676 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2677 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2678 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 1.8009, + "step": 2679 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2680 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2681 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2682 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2683 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 8.8786, + "step": 2684 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 10.3885, + "step": 2685 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2686 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2687 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2688 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2689 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2690 + }, + { + "epoch": 0.22, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2691 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.8706, + "step": 2692 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.8361, + "step": 2693 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.0737, + "step": 2694 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 8.2046, + "step": 2695 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.0706, + "step": 2696 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 5.7817, + "step": 2697 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2698 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.3935, + "step": 2699 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.8585, + "step": 2700 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.6376, + "step": 2701 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2702 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.8372, + "step": 2703 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.0747, + "step": 2704 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2705 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 6.1521, + "step": 2706 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2707 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.195, + "step": 2708 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.9549, + "step": 2709 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2710 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 9.3119, + "step": 2711 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 6.7614, + "step": 2712 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2713 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.0581, + "step": 2714 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2715 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2716 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2717 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.0408, + "step": 2718 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.5021, + "step": 2719 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2720 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2721 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2722 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 6.2398, + "step": 2723 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2724 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2725 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.6075, + "step": 2726 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.6661, + "step": 2727 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.015, + "step": 2728 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.1265, + "step": 2729 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2730 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2731 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.5972, + "step": 2732 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.1977, + "step": 2733 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2734 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2735 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 6.1039, + "step": 2736 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.3993, + "step": 2737 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2738 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 5.3397, + "step": 2739 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2740 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.0013, + "step": 2741 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2742 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.758, + "step": 2743 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2744 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.3765, + "step": 2745 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 5.8886, + "step": 2746 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2747 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.4292, + "step": 2748 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2749 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.5838, + "step": 2750 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.5964, + "step": 2751 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2752 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2753 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.6542, + "step": 2754 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2755 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.4352, + "step": 2756 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.2296, + "step": 2757 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2758 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.5741, + "step": 2759 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 8.8605, + "step": 2760 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 5.2518, + "step": 2761 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.8612, + "step": 2762 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.0333, + "step": 2763 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2764 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2765 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 8.1212, + "step": 2766 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.4345, + "step": 2767 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.2517, + "step": 2768 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.8264, + "step": 2769 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 5.3464, + "step": 2770 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.4944, + "step": 2771 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 5.9824, + "step": 2772 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2773 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.7, + "step": 2774 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2775 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.8109, + "step": 2776 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2777 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.0584, + "step": 2778 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2779 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2780 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.093, + "step": 2781 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.2437, + "step": 2782 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 8.0899, + "step": 2783 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2784 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2785 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.9202, + "step": 2786 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2787 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.8846, + "step": 2788 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2789 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2790 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.0103, + "step": 2791 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2792 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.5775, + "step": 2793 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.3984, + "step": 2794 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 1.1043, + "step": 2795 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2796 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 3.562, + "step": 2797 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2798 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.1613, + "step": 2799 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.3376, + "step": 2800 + }, + { + "epoch": 0.23, + "eval_loss": NaN, + "eval_runtime": 40.6257, + "eval_samples_per_second": 61.537, + "eval_steps_per_second": 7.704, + "step": 2800 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2801 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2802 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 7.87, + "step": 2803 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2804 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 2.3784, + "step": 2805 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.7962, + "step": 2806 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2807 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 0.9341, + "step": 2808 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.8966, + "step": 2809 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 11.8505, + "step": 2810 + }, + { + "epoch": 0.23, + "learning_rate": 0.0002, + "loss": 4.3227, + "step": 2811 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2812 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2813 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.8752, + "step": 2814 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.2878, + "step": 2815 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2816 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2817 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.1257, + "step": 2818 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2819 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2820 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.0982, + "step": 2821 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2822 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.5294, + "step": 2823 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2824 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.9476, + "step": 2825 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2826 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.1207, + "step": 2827 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2828 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2829 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 7.7884, + "step": 2830 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2831 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.7867, + "step": 2832 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2833 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 3.1663, + "step": 2834 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2835 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.9921, + "step": 2836 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2837 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.378, + "step": 2838 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.2746, + "step": 2839 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.1268, + "step": 2840 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2841 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2842 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2843 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2844 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.3648, + "step": 2845 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.3952, + "step": 2846 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2847 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 8.4811, + "step": 2848 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2849 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2850 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.0542, + "step": 2851 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 7.4905, + "step": 2852 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.7086, + "step": 2853 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2854 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.1558, + "step": 2855 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.3422, + "step": 2856 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2857 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 6.7713, + "step": 2858 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 7.0969, + "step": 2859 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.5118, + "step": 2860 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 7.3612, + "step": 2861 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2862 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.8953, + "step": 2863 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2864 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2865 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2866 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2867 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.1626, + "step": 2868 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 6.561, + "step": 2869 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 8.1548, + "step": 2870 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.2882, + "step": 2871 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.0975, + "step": 2872 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2873 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2874 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.3251, + "step": 2875 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2876 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 7.9258, + "step": 2877 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2878 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2879 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.0583, + "step": 2880 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.1771, + "step": 2881 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 6.3242, + "step": 2882 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 3.8587, + "step": 2883 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 6.2813, + "step": 2884 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.0015, + "step": 2885 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2886 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2887 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2888 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2889 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2890 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2891 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2892 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2893 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.7523, + "step": 2894 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2895 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 3.5483, + "step": 2896 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.9509, + "step": 2897 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.9968, + "step": 2898 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 6.3589, + "step": 2899 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 9.1624, + "step": 2900 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2901 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.7031, + "step": 2902 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2903 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.955, + "step": 2904 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2905 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 3.3997, + "step": 2906 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2907 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 2.1396, + "step": 2908 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.951, + "step": 2909 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.4144, + "step": 2910 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2911 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 8.8179, + "step": 2912 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2913 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 6.1996, + "step": 2914 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2915 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 7.7194, + "step": 2916 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 3.7272, + "step": 2917 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.1121, + "step": 2918 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.8043, + "step": 2919 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 8.1941, + "step": 2920 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.8971, + "step": 2921 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2922 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 5.8175, + "step": 2923 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2924 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2925 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2926 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 1.6108, + "step": 2927 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2928 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 3.7733, + "step": 2929 + }, + { + "epoch": 0.24, + "learning_rate": 0.0002, + "loss": 4.0127, + "step": 2930 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.7077, + "step": 2931 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2932 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2933 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2934 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.8122, + "step": 2935 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2936 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 9.6803, + "step": 2937 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2938 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.6482, + "step": 2939 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.9213, + "step": 2940 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2941 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2942 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.3731, + "step": 2943 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2944 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.8694, + "step": 2945 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2946 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2947 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.97, + "step": 2948 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 7.27, + "step": 2949 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2950 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.9638, + "step": 2951 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.7033, + "step": 2952 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.9179, + "step": 2953 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2954 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2955 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.9863, + "step": 2956 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2957 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.8758, + "step": 2958 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.048, + "step": 2959 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2960 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 5.1058, + "step": 2961 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 6.2794, + "step": 2962 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.5079, + "step": 2963 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.5687, + "step": 2964 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.9975, + "step": 2965 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2966 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.298, + "step": 2967 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.0723, + "step": 2968 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2969 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2970 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2971 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 6.6388, + "step": 2972 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2973 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2974 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.1096, + "step": 2975 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.6381, + "step": 2976 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2977 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2978 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2979 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.1973, + "step": 2980 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2981 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.3643, + "step": 2982 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 5.1721, + "step": 2983 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.2123, + "step": 2984 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2985 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2986 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2987 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.748, + "step": 2988 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.7093, + "step": 2989 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2990 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2991 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2992 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.851, + "step": 2993 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2994 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2995 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.4522, + "step": 2996 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2997 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 2998 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.1996, + "step": 2999 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.0192, + "step": 3000 + }, + { + "epoch": 0.25, + "eval_loss": NaN, + "eval_runtime": 40.2656, + "eval_samples_per_second": 62.088, + "eval_steps_per_second": 7.773, + "step": 3000 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.9794, + "step": 3001 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.9861, + "step": 3002 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 7.7348, + "step": 3003 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3004 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 7.6608, + "step": 3005 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3006 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.341, + "step": 3007 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.958, + "step": 3008 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.7489, + "step": 3009 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3010 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.049, + "step": 3011 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.2465, + "step": 3012 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 11.7686, + "step": 3013 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3014 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3015 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 5.6222, + "step": 3016 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3017 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 6.8129, + "step": 3018 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.2546, + "step": 3019 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.9979, + "step": 3020 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 9.5126, + "step": 3021 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.1383, + "step": 3022 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3023 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.6223, + "step": 3024 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3025 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3026 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3027 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 4.5463, + "step": 3028 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3029 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.8482, + "step": 3030 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.0748, + "step": 3031 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3032 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 2.016, + "step": 3033 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.0744, + "step": 3034 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3035 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 1.8062, + "step": 3036 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3037 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3038 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3039 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3040 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3041 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3042 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3043 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3044 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3045 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.8464, + "step": 3046 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3047 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3048 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 3.5565, + "step": 3049 + }, + { + "epoch": 0.25, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3050 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3051 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3052 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3053 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3054 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 7.9417, + "step": 3055 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.8193, + "step": 3056 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 4.1043, + "step": 3057 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 4.0546, + "step": 3058 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3059 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 8.1027, + "step": 3060 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3061 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3062 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 5.8355, + "step": 3063 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 3.4701, + "step": 3064 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3065 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 3.2914, + "step": 3066 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 5.6619, + "step": 3067 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.8956, + "step": 3068 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3069 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3070 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.8051, + "step": 3071 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3072 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 2.4279, + "step": 3073 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.0723, + "step": 3074 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 7.1402, + "step": 3075 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 8.5823, + "step": 3076 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 3.587, + "step": 3077 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3078 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3079 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3080 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3081 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 3.4281, + "step": 3082 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.7726, + "step": 3083 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3084 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 8.1562, + "step": 3085 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3086 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 9.6003, + "step": 3087 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3088 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 2.2855, + "step": 3089 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.6221, + "step": 3090 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 5.4474, + "step": 3091 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 5.0483, + "step": 3092 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 2.3853, + "step": 3093 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3094 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3095 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3096 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.3538, + "step": 3097 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.9828, + "step": 3098 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3099 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.902, + "step": 3100 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 7.1081, + "step": 3101 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3102 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3103 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3104 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3105 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3106 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3107 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 7.7653, + "step": 3108 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3109 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.8011, + "step": 3110 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3111 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 4.21, + "step": 3112 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 4.2024, + "step": 3113 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.0741, + "step": 3114 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3115 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.9862, + "step": 3116 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 4.2128, + "step": 3117 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.8204, + "step": 3118 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 3.6152, + "step": 3119 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3120 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 4.2347, + "step": 3121 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 3.6396, + "step": 3122 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.8555, + "step": 3123 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3124 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3125 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3126 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3127 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 8.0978, + "step": 3128 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.7515, + "step": 3129 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 6.8497, + "step": 3130 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3131 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3132 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3133 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3134 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 2.6837, + "step": 3135 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 5.006, + "step": 3136 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3137 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3138 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3139 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3140 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.9509, + "step": 3141 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 4.3639, + "step": 3142 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3143 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.001, + "step": 3144 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3145 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3146 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3147 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.0489, + "step": 3148 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.7395, + "step": 3149 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 9.3945, + "step": 3150 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3151 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3152 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3153 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.6796, + "step": 3154 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3155 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.1011, + "step": 3156 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3157 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 7.0474, + "step": 3158 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3159 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3160 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3161 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3162 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3163 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 5.2225, + "step": 3164 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3165 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.8361, + "step": 3166 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3167 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3168 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 1.1624, + "step": 3169 + }, + { + "epoch": 0.26, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3170 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3171 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 2.2142, + "step": 3172 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 2.3485, + "step": 3173 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3174 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 5.7756, + "step": 3175 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 7.1204, + "step": 3176 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3177 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3178 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3179 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 6.6894, + "step": 3180 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.5424, + "step": 3181 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.8102, + "step": 3182 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.9355, + "step": 3183 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3184 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3185 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3186 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.7483, + "step": 3187 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3188 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 6.5004, + "step": 3189 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.9152, + "step": 3190 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.8493, + "step": 3191 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.0135, + "step": 3192 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.9181, + "step": 3193 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3194 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.4914, + "step": 3195 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3196 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 9.1608, + "step": 3197 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.8671, + "step": 3198 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.3228, + "step": 3199 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3200 + }, + { + "epoch": 0.27, + "eval_loss": NaN, + "eval_runtime": 40.5366, + "eval_samples_per_second": 61.673, + "eval_steps_per_second": 7.721, + "step": 3200 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3201 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3202 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3203 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.9418, + "step": 3204 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.5376, + "step": 3205 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 8.7518, + "step": 3206 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3207 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.9102, + "step": 3208 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3209 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3210 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 6.1783, + "step": 3211 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3212 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.0544, + "step": 3213 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.9945, + "step": 3214 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 5.5622, + "step": 3215 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.645, + "step": 3216 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.8808, + "step": 3217 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3218 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 5.6448, + "step": 3219 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.754, + "step": 3220 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 7.1268, + "step": 3221 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3222 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.1064, + "step": 3223 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3224 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3225 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 7.3858, + "step": 3226 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3227 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.8886, + "step": 3228 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3229 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3230 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.0533, + "step": 3231 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.7967, + "step": 3232 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3233 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3234 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.9231, + "step": 3235 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.078, + "step": 3236 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.8714, + "step": 3237 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3238 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 2.8482, + "step": 3239 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3240 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3241 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 7.3685, + "step": 3242 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.8902, + "step": 3243 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 4.2472, + "step": 3244 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3245 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.9301, + "step": 3246 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.9081, + "step": 3247 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 6.4631, + "step": 3248 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3249 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 2.0449, + "step": 3250 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.1283, + "step": 3251 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3252 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3253 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.9513, + "step": 3254 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3255 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3256 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 9.0395, + "step": 3257 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3258 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 6.5533, + "step": 3259 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.1523, + "step": 3260 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3261 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.9341, + "step": 3262 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 7.2839, + "step": 3263 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3264 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3265 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3266 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 9.5936, + "step": 3267 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3268 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.3137, + "step": 3269 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3270 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3271 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3272 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 8.4758, + "step": 3273 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3274 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 2.2908, + "step": 3275 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3276 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3277 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.9998, + "step": 3278 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.8858, + "step": 3279 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3280 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3281 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3282 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3283 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3284 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 3.8323, + "step": 3285 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3286 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 1.039, + "step": 3287 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3288 + }, + { + "epoch": 0.27, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3289 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 5.9844, + "step": 3290 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3291 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3292 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.1358, + "step": 3293 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.8401, + "step": 3294 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 5.7743, + "step": 3295 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3296 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.8621, + "step": 3297 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 5.2579, + "step": 3298 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3299 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.6627, + "step": 3300 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3301 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.2131, + "step": 3302 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.7066, + "step": 3303 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 6.8875, + "step": 3304 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3305 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.8612, + "step": 3306 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.5656, + "step": 3307 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.8595, + "step": 3308 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.2352, + "step": 3309 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.8525, + "step": 3310 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 5.2575, + "step": 3311 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3312 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.5786, + "step": 3313 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.0435, + "step": 3314 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3315 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3316 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3317 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.2254, + "step": 3318 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3319 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3320 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 6.849, + "step": 3321 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 7.2595, + "step": 3322 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3323 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3324 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3325 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 11.2648, + "step": 3326 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 8.7223, + "step": 3327 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 4.349, + "step": 3328 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.3578, + "step": 3329 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3330 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3331 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.407, + "step": 3332 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3333 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.0008, + "step": 3334 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3335 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3336 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.0186, + "step": 3337 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.279, + "step": 3338 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3339 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 7.9527, + "step": 3340 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3341 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3342 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.1755, + "step": 3343 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3344 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3345 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.8138, + "step": 3346 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3347 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.7684, + "step": 3348 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.9347, + "step": 3349 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3350 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3351 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.7469, + "step": 3352 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 4.2655, + "step": 3353 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.7135, + "step": 3354 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3355 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3356 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 9.8816, + "step": 3357 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3358 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3359 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 4.6036, + "step": 3360 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.1882, + "step": 3361 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.7127, + "step": 3362 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3363 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3364 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3365 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3366 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3367 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3368 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3369 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.9135, + "step": 3370 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.2027, + "step": 3371 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3372 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3373 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3374 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.2555, + "step": 3375 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.5389, + "step": 3376 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 5.4486, + "step": 3377 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.882, + "step": 3378 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3379 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3380 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.286, + "step": 3381 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3382 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3383 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3384 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 7.468, + "step": 3385 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.0501, + "step": 3386 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 4.582, + "step": 3387 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3388 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.6676, + "step": 3389 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3390 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3391 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3392 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3393 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 8.5212, + "step": 3394 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.9269, + "step": 3395 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 1.3919, + "step": 3396 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.5774, + "step": 3397 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3398 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 9.312, + "step": 3399 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.9381, + "step": 3400 + }, + { + "epoch": 0.28, + "eval_loss": NaN, + "eval_runtime": 40.3029, + "eval_samples_per_second": 62.03, + "eval_steps_per_second": 7.766, + "step": 3400 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.4252, + "step": 3401 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3402 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 4.6181, + "step": 3403 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 3.9353, + "step": 3404 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 2.1189, + "step": 3405 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3406 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3407 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 5.852, + "step": 3408 + }, + { + "epoch": 0.28, + "learning_rate": 0.0002, + "loss": 0.9672, + "step": 3409 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3410 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 5.4328, + "step": 3411 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3412 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3413 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3414 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 2.0155, + "step": 3415 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 1.9281, + "step": 3416 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 2.8115, + "step": 3417 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3418 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3419 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3420 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 3.0399, + "step": 3421 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3422 + }, + { + "epoch": 0.29, + "learning_rate": 0.0002, + "loss": 0.0, + "step": 3423 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.7252, + "step": 3424 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3425 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3426 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3427 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 8.1271, + "step": 3428 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.9564, + "step": 3429 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 5.2747, + "step": 3430 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.0489, + "step": 3431 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 7.5811, + "step": 3432 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.1785, + "step": 3433 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.2517, + "step": 3434 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 3.5945, + "step": 3435 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 8.7147, + "step": 3436 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 7.7128, + "step": 3437 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3438 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 8.4043, + "step": 3439 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.8927, + "step": 3440 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.6778, + "step": 3441 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.9006, + "step": 3442 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.9283, + "step": 3443 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 7.7694, + "step": 3444 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.8328, + "step": 3445 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3446 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3447 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.9932, + "step": 3448 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3449 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.9979, + "step": 3450 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 2.2667, + "step": 3451 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 7.8484, + "step": 3452 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3453 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3454 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3455 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3456 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 2.7836, + "step": 3457 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3458 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 7.4761, + "step": 3459 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3460 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.1591, + "step": 3461 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 5.9251, + "step": 3462 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 3.7306, + "step": 3463 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3464 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3465 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3466 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.9593, + "step": 3467 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3468 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3469 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.9817, + "step": 3470 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.8053, + "step": 3471 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3472 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3473 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 10.3354, + "step": 3474 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.459, + "step": 3475 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.0281, + "step": 3476 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3477 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 2.0665, + "step": 3478 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.5374, + "step": 3479 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3480 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.1926, + "step": 3481 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 3.6464, + "step": 3482 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3483 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3484 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3485 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 9.9536, + "step": 3486 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 2.1352, + "step": 3487 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.0294, + "step": 3488 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3489 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.8207, + "step": 3490 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 9.6174, + "step": 3491 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3492 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 7.4924, + "step": 3493 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3494 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 7.5259, + "step": 3495 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3496 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3497 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.2182, + "step": 3498 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 2.1827, + "step": 3499 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3500 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.0221, + "step": 3501 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3502 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.7986, + "step": 3503 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 2.0929, + "step": 3504 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3505 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3506 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 3.643, + "step": 3507 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3508 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3509 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3510 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3511 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3512 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.9393, + "step": 3513 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3514 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3515 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 3.5622, + "step": 3516 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 1.1413, + "step": 3517 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 8.5247, + "step": 3518 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3519 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3520 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.2493, + "step": 3521 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3522 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 9.174, + "step": 3523 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3524 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3525 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.996, + "step": 3526 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 2.1666, + "step": 3527 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3528 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001999944269512637, + "loss": 4.1896, + "step": 3529 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3530 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3531 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3532 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 5.1456, + "step": 3533 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 5.5647, + "step": 3534 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3535 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3536 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.3891, + "step": 3537 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3538 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.0263, + "step": 3539 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 8.8189, + "step": 3540 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3541 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.2281, + "step": 3542 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3543 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3544 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 6.4216, + "step": 3545 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.1089, + "step": 3546 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.9854, + "step": 3547 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.0804, + "step": 3548 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3549 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 8.4601, + "step": 3550 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3551 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3552 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.1242, + "step": 3553 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.5081, + "step": 3554 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 9.1416, + "step": 3555 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3556 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.0224, + "step": 3557 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 5.5793, + "step": 3558 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.3404, + "step": 3559 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.3061, + "step": 3560 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3561 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.2258, + "step": 3562 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3563 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.3189, + "step": 3564 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3565 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3566 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.926, + "step": 3567 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 7.75, + "step": 3568 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 7.8106, + "step": 3569 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.0975, + "step": 3570 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.6651, + "step": 3571 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.1599, + "step": 3572 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.8459, + "step": 3573 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.8985, + "step": 3574 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3575 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.435, + "step": 3576 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3577 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.273, + "step": 3578 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 5.7811, + "step": 3579 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3580 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 8.6478, + "step": 3581 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.0275, + "step": 3582 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.9068, + "step": 3583 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3584 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.0871, + "step": 3585 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.0027, + "step": 3586 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3587 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.6775, + "step": 3588 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3589 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.9142, + "step": 3590 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.0995, + "step": 3591 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 11.7301, + "step": 3592 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.5631, + "step": 3593 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3594 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3595 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.8991, + "step": 3596 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3597 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.9183, + "step": 3598 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 8.3794, + "step": 3599 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.5741, + "step": 3600 + }, + { + "epoch": 0.3, + "eval_loss": NaN, + "eval_runtime": 40.5314, + "eval_samples_per_second": 61.681, + "eval_steps_per_second": 7.722, + "step": 3600 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.9487, + "step": 3601 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3602 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3603 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3604 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3605 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 7.4446, + "step": 3606 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3607 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 8.8191, + "step": 3608 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.76, + "step": 3609 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3610 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.2858, + "step": 3611 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 7.2901, + "step": 3612 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3613 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.0884, + "step": 3614 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.9502, + "step": 3615 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3616 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.8152, + "step": 3617 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3618 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3619 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3620 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3621 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3622 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3623 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 5.5169, + "step": 3624 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3625 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3626 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3627 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 3.392, + "step": 3628 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3629 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3630 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3631 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3632 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.0432, + "step": 3633 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3634 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3635 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3636 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3637 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 8.3224, + "step": 3638 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3639 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3640 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3641 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 2.3504, + "step": 3642 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.112, + "step": 3643 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3644 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.2873, + "step": 3645 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 4.3, + "step": 3646 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3647 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001999944269512637, + "loss": 1.9262, + "step": 3648 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3649 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 7.2117, + "step": 3650 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3651 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 2.0566, + "step": 3652 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3653 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3654 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3655 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 3.3123, + "step": 3656 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 5.8027, + "step": 3657 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3658 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.804, + "step": 3659 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.0845, + "step": 3660 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3661 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3662 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3663 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 9.0904, + "step": 3664 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3665 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3666 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3667 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 3.6369, + "step": 3668 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 8.2121, + "step": 3669 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3670 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3671 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.2386, + "step": 3672 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3673 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 3.8701, + "step": 3674 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3675 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3676 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3677 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.115, + "step": 3678 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3679 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.7665, + "step": 3680 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.0305, + "step": 3681 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 5.6492, + "step": 3682 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.8425, + "step": 3683 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3684 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 6.6016, + "step": 3685 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3686 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3687 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3688 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3689 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3690 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3691 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 5.5028, + "step": 3692 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.0887, + "step": 3693 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3694 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.6216, + "step": 3695 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3696 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3697 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3698 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3699 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.9473, + "step": 3700 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.9741, + "step": 3701 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 7.8511, + "step": 3702 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3703 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3704 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3705 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3706 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 7.4423, + "step": 3707 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3708 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3709 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3710 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 2.0791, + "step": 3711 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3712 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3713 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3714 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.9085, + "step": 3715 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3716 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3717 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3718 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.49, + "step": 3719 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.9619, + "step": 3720 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3721 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3722 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.2649, + "step": 3723 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3724 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.0366, + "step": 3725 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.9452, + "step": 3726 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3727 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 2.088, + "step": 3728 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3729 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 6.2874, + "step": 3730 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3731 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 6.0155, + "step": 3732 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.7768, + "step": 3733 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3734 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3735 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.8478, + "step": 3736 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 3.6496, + "step": 3737 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3738 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3739 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.8128, + "step": 3740 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.3734, + "step": 3741 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.1389, + "step": 3742 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3743 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3744 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 5.6606, + "step": 3745 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 8.6413, + "step": 3746 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 2.8916, + "step": 3747 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3748 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.2019, + "step": 3749 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 5.0741, + "step": 3750 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.9239, + "step": 3751 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3752 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 6.9434, + "step": 3753 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 3.9492, + "step": 3754 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3755 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 7.8425, + "step": 3756 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.8377, + "step": 3757 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.9527, + "step": 3758 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.9582, + "step": 3759 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.8098, + "step": 3760 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 4.6772, + "step": 3761 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3762 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 1.904, + "step": 3763 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3764 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3765 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3766 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 8.9331, + "step": 3767 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3768 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3769 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3770 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 9.686, + "step": 3771 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3772 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3773 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3774 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.0365, + "step": 3775 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3776 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 6.6873, + "step": 3777 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.0444, + "step": 3778 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3779 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3780 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 6.9463, + "step": 3781 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.3288, + "step": 3782 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3783 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3784 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.2635, + "step": 3785 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3786 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3787 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.8884, + "step": 3788 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3789 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3790 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.0787, + "step": 3791 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3792 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3793 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.0431, + "step": 3794 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 6.4756, + "step": 3795 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3796 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.2578, + "step": 3797 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3798 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.0616, + "step": 3799 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3800 + }, + { + "epoch": 0.32, + "eval_loss": NaN, + "eval_runtime": 40.8054, + "eval_samples_per_second": 61.266, + "eval_steps_per_second": 7.671, + "step": 3800 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 9.0586, + "step": 3801 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3802 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.1605, + "step": 3803 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.6307, + "step": 3804 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 3.7127, + "step": 3805 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3806 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3807 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3808 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3809 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 3.1952, + "step": 3810 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.3244, + "step": 3811 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3812 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.7178, + "step": 3813 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.846, + "step": 3814 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 7.2241, + "step": 3815 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.8769, + "step": 3816 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3817 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 3.9671, + "step": 3818 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 3.5565, + "step": 3819 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 7.3086, + "step": 3820 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.5791, + "step": 3821 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.9555, + "step": 3822 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3823 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3824 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3825 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.6867, + "step": 3826 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.9751, + "step": 3827 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.8508, + "step": 3828 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.3062, + "step": 3829 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.8794, + "step": 3830 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.1897, + "step": 3831 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 9.7595, + "step": 3832 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.8866, + "step": 3833 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.0025, + "step": 3834 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.346, + "step": 3835 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.3663, + "step": 3836 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 7.8214, + "step": 3837 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 13.937, + "step": 3838 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.9182, + "step": 3839 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3840 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 8.8916, + "step": 3841 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3842 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.0794, + "step": 3843 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3844 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.1539, + "step": 3845 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3846 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.9754, + "step": 3847 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.2543, + "step": 3848 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.0772, + "step": 3849 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3850 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.9198, + "step": 3851 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 8.7337, + "step": 3852 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.9703, + "step": 3853 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.1085, + "step": 3854 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3855 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.2346, + "step": 3856 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.9797, + "step": 3857 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3858 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 4.6654, + "step": 3859 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 7.4903, + "step": 3860 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3861 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 6.4773, + "step": 3862 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3863 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.0852, + "step": 3864 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3865 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 5.8414, + "step": 3866 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3867 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 2.533, + "step": 3868 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 10.7621, + "step": 3869 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 6.6782, + "step": 3870 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 6.5977, + "step": 3871 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 7.2208, + "step": 3872 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 8.9974, + "step": 3873 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 7.0978, + "step": 3874 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3875 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3876 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3877 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3878 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3879 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.9362, + "step": 3880 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 3.7746, + "step": 3881 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3882 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3883 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 6.9983, + "step": 3884 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 8.3742, + "step": 3885 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 3.6063, + "step": 3886 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001999944269512637, + "loss": 1.1037, + "step": 3887 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3888 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3889 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3890 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 9.0236, + "step": 3891 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.8414, + "step": 3892 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3893 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.6047, + "step": 3894 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3895 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.1252, + "step": 3896 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3897 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.8188, + "step": 3898 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.1472, + "step": 3899 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 8.118, + "step": 3900 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.2012, + "step": 3901 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3902 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.1579, + "step": 3903 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3904 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 5.0518, + "step": 3905 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3906 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.2508, + "step": 3907 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3908 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3909 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.9624, + "step": 3910 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3911 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 5.7856, + "step": 3912 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.7073, + "step": 3913 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.0698, + "step": 3914 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.1907, + "step": 3915 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 2.9845, + "step": 3916 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3917 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.9564, + "step": 3918 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3919 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3920 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3921 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3922 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.6052, + "step": 3923 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.6512, + "step": 3924 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.9223, + "step": 3925 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3926 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 8.0766, + "step": 3927 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 5.0875, + "step": 3928 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3929 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3930 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.847, + "step": 3931 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.8444, + "step": 3932 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 2.3355, + "step": 3933 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.8717, + "step": 3934 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.7535, + "step": 3935 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3936 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3937 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3938 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3939 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.9637, + "step": 3940 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.4794, + "step": 3941 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3942 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3943 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 8.8335, + "step": 3944 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3945 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 2.28, + "step": 3946 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3947 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.8047, + "step": 3948 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 5.1984, + "step": 3949 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.9645, + "step": 3950 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.1758, + "step": 3951 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 6.0737, + "step": 3952 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3953 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3954 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 8.1112, + "step": 3955 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.9219, + "step": 3956 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3957 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.8557, + "step": 3958 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.8955, + "step": 3959 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3960 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 3.8936, + "step": 3961 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.5108, + "step": 3962 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3963 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.042, + "step": 3964 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 3.0503, + "step": 3965 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 5.2058, + "step": 3966 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3967 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.018, + "step": 3968 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 2.1701, + "step": 3969 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3970 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.7046, + "step": 3971 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 2.2071, + "step": 3972 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.0525, + "step": 3973 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 8.378, + "step": 3974 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 2.8837, + "step": 3975 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.9556, + "step": 3976 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3977 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3978 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 9.3965, + "step": 3979 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3980 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3981 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3982 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3983 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 4.8205, + "step": 3984 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3985 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3986 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3987 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 8.7562, + "step": 3988 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3989 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.9865, + "step": 3990 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3991 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3992 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.2552, + "step": 3993 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 7.6012, + "step": 3994 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.8483, + "step": 3995 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.9023, + "step": 3996 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3997 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.8682, + "step": 3998 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 3999 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.1343, + "step": 4000 + }, + { + "epoch": 0.33, + "eval_loss": NaN, + "eval_runtime": 40.528, + "eval_samples_per_second": 61.686, + "eval_steps_per_second": 7.723, + "step": 4000 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4001 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4002 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 1.5904, + "step": 4003 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4004 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 3.3693, + "step": 4005 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4006 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001999944269512637, + "loss": 12.6352, + "step": 4007 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 6.6557, + "step": 4008 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.0262, + "step": 4009 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4010 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4011 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4012 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4013 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4014 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.2143, + "step": 4015 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4016 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4017 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4018 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4019 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.2229, + "step": 4020 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4021 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 4.1986, + "step": 4022 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.5261, + "step": 4023 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.9264, + "step": 4024 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4025 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4026 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 5.718, + "step": 4027 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 7.9252, + "step": 4028 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.6114, + "step": 4029 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 7.9863, + "step": 4030 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.2768, + "step": 4031 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4032 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4033 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.9001, + "step": 4034 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 2.0889, + "step": 4035 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.9462, + "step": 4036 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 8.5354, + "step": 4037 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.9757, + "step": 4038 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4039 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4040 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4041 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 5.5367, + "step": 4042 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4043 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4044 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.1202, + "step": 4045 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.2619, + "step": 4046 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4047 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.9477, + "step": 4048 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.6275, + "step": 4049 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.5908, + "step": 4050 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4051 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4052 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.226, + "step": 4053 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 7.0025, + "step": 4054 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4055 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 8.9711, + "step": 4056 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.7448, + "step": 4057 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 7.5651, + "step": 4058 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4059 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4060 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4061 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4062 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.7352, + "step": 4063 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.9633, + "step": 4064 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 2.7585, + "step": 4065 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4066 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4067 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 2.0017, + "step": 4068 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 5.9215, + "step": 4069 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 2.2832, + "step": 4070 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.075, + "step": 4071 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4072 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.8116, + "step": 4073 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 5.871, + "step": 4074 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4075 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.953, + "step": 4076 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.874, + "step": 4077 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.539, + "step": 4078 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4079 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4080 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4081 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4082 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 8.2702, + "step": 4083 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4084 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4085 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.4962, + "step": 4086 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4087 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4088 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.3512, + "step": 4089 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 8.635, + "step": 4090 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.7174, + "step": 4091 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 2.9093, + "step": 4092 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 11.1589, + "step": 4093 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4094 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4095 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.8881, + "step": 4096 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4097 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.8945, + "step": 4098 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 4.3487, + "step": 4099 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 4.0, + "step": 4100 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4101 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.9106, + "step": 4102 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4103 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4104 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4105 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 1.9125, + "step": 4106 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 6.2511, + "step": 4107 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 3.7398, + "step": 4108 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 4.1571, + "step": 4109 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 5.1311, + "step": 4110 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4111 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 4.2526, + "step": 4112 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4113 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4114 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4115 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4116 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 7.7724, + "step": 4117 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4118 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4119 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 2.2654, + "step": 4120 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4121 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4122 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 2.1991, + "step": 4123 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 5.9311, + "step": 4124 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 4.6242, + "step": 4125 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4126 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001999944269512637, + "loss": 0.9792, + "step": 4127 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 7.7565, + "step": 4128 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.7826, + "step": 4129 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4130 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.0482, + "step": 4131 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 6.6568, + "step": 4132 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4133 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4134 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4135 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.1625, + "step": 4136 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4137 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.9005, + "step": 4138 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 7.8954, + "step": 4139 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.3652, + "step": 4140 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4141 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4142 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 6.1353, + "step": 4143 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4144 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4145 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4146 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.3082, + "step": 4147 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4148 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.6613, + "step": 4149 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4150 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4151 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4152 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4153 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.3213, + "step": 4154 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.0886, + "step": 4155 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4156 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4157 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.0031, + "step": 4158 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4159 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4160 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 8.4849, + "step": 4161 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4162 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.0386, + "step": 4163 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.2202, + "step": 4164 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.5983, + "step": 4165 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.2529, + "step": 4166 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.5377, + "step": 4167 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 6.4775, + "step": 4168 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.0506, + "step": 4169 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 8.1124, + "step": 4170 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4171 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4172 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.0681, + "step": 4173 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.3068, + "step": 4174 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 8.0692, + "step": 4175 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.2598, + "step": 4176 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.0436, + "step": 4177 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4178 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4179 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.8912, + "step": 4180 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.6612, + "step": 4181 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4182 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4183 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4184 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.9021, + "step": 4185 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.9676, + "step": 4186 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4187 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.0047, + "step": 4188 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4189 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.8383, + "step": 4190 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.7605, + "step": 4191 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4192 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4193 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.1685, + "step": 4194 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4195 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.0561, + "step": 4196 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.7203, + "step": 4197 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4198 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.9182, + "step": 4199 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.244, + "step": 4200 + }, + { + "epoch": 0.35, + "eval_loss": NaN, + "eval_runtime": 40.6418, + "eval_samples_per_second": 61.513, + "eval_steps_per_second": 7.701, + "step": 4200 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.9751, + "step": 4201 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.0548, + "step": 4202 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 6.9036, + "step": 4203 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4204 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.9272, + "step": 4205 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 6.5804, + "step": 4206 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.7809, + "step": 4207 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.7871, + "step": 4208 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4209 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4210 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.0716, + "step": 4211 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 7.5109, + "step": 4212 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.8018, + "step": 4213 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.9591, + "step": 4214 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4215 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.2448, + "step": 4216 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.5332, + "step": 4217 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4218 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4219 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4220 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4221 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 10.1275, + "step": 4222 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4223 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.0431, + "step": 4224 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4225 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4226 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4227 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.1813, + "step": 4228 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.1153, + "step": 4229 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 2.0878, + "step": 4230 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.4025, + "step": 4231 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4232 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.9378, + "step": 4233 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 1.0872, + "step": 4234 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 5.6795, + "step": 4235 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.8339, + "step": 4236 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 6.0213, + "step": 4237 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4238 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4239 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 7.9027, + "step": 4240 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4241 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 4.9827, + "step": 4242 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 3.3402, + "step": 4243 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.8302, + "step": 4244 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4245 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4246 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 7.5063, + "step": 4247 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4248 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4249 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4250 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.8848, + "step": 4251 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.5452, + "step": 4252 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.0767, + "step": 4253 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.873, + "step": 4254 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4255 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4256 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 8.156, + "step": 4257 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4258 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4259 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.9804, + "step": 4260 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.3509, + "step": 4261 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.8441, + "step": 4262 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.8729, + "step": 4263 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 7.8144, + "step": 4264 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4265 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4266 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.5836, + "step": 4267 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 4.4446, + "step": 4268 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4269 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4270 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.5381, + "step": 4271 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4272 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4273 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 8.0151, + "step": 4274 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 7.7336, + "step": 4275 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.8726, + "step": 4276 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 7.5989, + "step": 4277 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4278 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4279 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.7174, + "step": 4280 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 5.7934, + "step": 4281 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.955, + "step": 4282 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 6.2183, + "step": 4283 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4284 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4285 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4286 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 8.2463, + "step": 4287 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 6.9663, + "step": 4288 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4289 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 4.2473, + "step": 4290 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 2.4481, + "step": 4291 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 9.7948, + "step": 4292 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.7029, + "step": 4293 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4294 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4295 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4296 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4297 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.836, + "step": 4298 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4299 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4300 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.0347, + "step": 4301 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.971, + "step": 4302 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4303 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 2.0843, + "step": 4304 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4305 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 7.148, + "step": 4306 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4307 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 4.6689, + "step": 4308 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 7.5345, + "step": 4309 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.7403, + "step": 4310 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.6691, + "step": 4311 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.8348, + "step": 4312 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.7668, + "step": 4313 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4314 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.9954, + "step": 4315 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4316 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4317 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.5721, + "step": 4318 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 5.2966, + "step": 4319 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 6.6498, + "step": 4320 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4321 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4322 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4323 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4324 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4325 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 4.3593, + "step": 4326 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4327 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 2.0198, + "step": 4328 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.9333, + "step": 4329 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.8951, + "step": 4330 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.5315, + "step": 4331 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4332 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4333 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4334 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.1517, + "step": 4335 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4336 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4337 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.2665, + "step": 4338 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4339 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 2.3957, + "step": 4340 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.9691, + "step": 4341 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4342 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 5.031, + "step": 4343 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4344 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 10.2832, + "step": 4345 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4346 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.7788, + "step": 4347 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.0194, + "step": 4348 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4349 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4350 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4351 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 9.4636, + "step": 4352 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4353 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4354 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.897, + "step": 4355 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 4.2934, + "step": 4356 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4357 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4358 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 1.5182, + "step": 4359 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4360 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4361 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 8.9023, + "step": 4362 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4363 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 3.3384, + "step": 4364 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4365 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001999944269512637, + "loss": 2.1051, + "step": 4366 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.9895, + "step": 4367 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.235, + "step": 4368 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4369 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4370 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 1.6688, + "step": 4371 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 1.4078, + "step": 4372 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.143, + "step": 4373 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.7254, + "step": 4374 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.6211, + "step": 4375 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4376 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 6.1533, + "step": 4377 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.7167, + "step": 4378 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4379 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4380 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4381 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4382 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.3532, + "step": 4383 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 5.7949, + "step": 4384 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 6.0345, + "step": 4385 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4386 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4387 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.1442, + "step": 4388 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4389 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4390 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4391 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.1038, + "step": 4392 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 8.4568, + "step": 4393 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4394 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.066, + "step": 4395 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.2763, + "step": 4396 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.9329, + "step": 4397 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4398 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4399 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4400 + }, + { + "epoch": 0.37, + "eval_loss": NaN, + "eval_runtime": 40.9589, + "eval_samples_per_second": 61.037, + "eval_steps_per_second": 7.642, + "step": 4400 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4401 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4402 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4403 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.8865, + "step": 4404 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.0661, + "step": 4405 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.9968, + "step": 4406 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 10.4144, + "step": 4407 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4408 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.5344, + "step": 4409 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.7948, + "step": 4410 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.9388, + "step": 4411 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4412 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.6846, + "step": 4413 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.105, + "step": 4414 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.4104, + "step": 4415 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.0398, + "step": 4416 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.9958, + "step": 4417 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4418 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 5.4571, + "step": 4419 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.8605, + "step": 4420 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4421 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.1168, + "step": 4422 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4423 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 5.7584, + "step": 4424 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4425 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.9852, + "step": 4426 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4427 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 8.4549, + "step": 4428 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.3864, + "step": 4429 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 1.1641, + "step": 4430 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.2905, + "step": 4431 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4432 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4433 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.8708, + "step": 4434 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.9518, + "step": 4435 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4436 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.587, + "step": 4437 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 1.07, + "step": 4438 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.8057, + "step": 4439 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.3462, + "step": 4440 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.1713, + "step": 4441 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4442 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.84, + "step": 4443 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4444 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4445 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4446 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4447 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4448 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4449 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.9176, + "step": 4450 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4451 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4452 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4453 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4454 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.9824, + "step": 4455 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4456 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 1.2536, + "step": 4457 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4458 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4459 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4460 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.3571, + "step": 4461 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 1.0795, + "step": 4462 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4463 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.2971, + "step": 4464 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.5844, + "step": 4465 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4466 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4467 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4468 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4469 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4470 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 6.1159, + "step": 4471 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 7.3768, + "step": 4472 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.19, + "step": 4473 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4474 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.9334, + "step": 4475 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.1259, + "step": 4476 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4477 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 3.8616, + "step": 4478 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4479 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4480 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4481 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4482 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.1066, + "step": 4483 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 1.8826, + "step": 4484 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 2.1722, + "step": 4485 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001999944269512637, + "loss": 4.309, + "step": 4486 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 7.7379, + "step": 4487 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.8532, + "step": 4488 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.6764, + "step": 4489 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4490 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4491 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4492 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.9546, + "step": 4493 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.5149, + "step": 4494 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.4848, + "step": 4495 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4496 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4497 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4498 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4499 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4500 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4501 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4502 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4503 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.9563, + "step": 4504 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4505 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.8403, + "step": 4506 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.8514, + "step": 4507 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.286, + "step": 4508 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.6812, + "step": 4509 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.0101, + "step": 4510 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.8956, + "step": 4511 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 7.6312, + "step": 4512 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 8.3148, + "step": 4513 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4514 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4515 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4516 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.863, + "step": 4517 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4518 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 9.5496, + "step": 4519 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.9445, + "step": 4520 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4521 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.586, + "step": 4522 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4523 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4524 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.2339, + "step": 4525 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 8.5051, + "step": 4526 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4527 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4528 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4529 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.138, + "step": 4530 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4531 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4532 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4533 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.2574, + "step": 4534 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.0714, + "step": 4535 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 4.6823, + "step": 4536 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.8283, + "step": 4537 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4538 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4539 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.2457, + "step": 4540 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 4.0105, + "step": 4541 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.8839, + "step": 4542 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4543 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.0567, + "step": 4544 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4545 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4546 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4547 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 7.6839, + "step": 4548 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.6607, + "step": 4549 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4550 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 4.3351, + "step": 4551 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.5946, + "step": 4552 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4553 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4554 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4555 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.4864, + "step": 4556 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.5884, + "step": 4557 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.4586, + "step": 4558 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.2788, + "step": 4559 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4560 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.0867, + "step": 4561 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 5.1577, + "step": 4562 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4563 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4564 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.9644, + "step": 4565 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4566 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.9162, + "step": 4567 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.9324, + "step": 4568 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4569 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.6422, + "step": 4570 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4571 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.4579, + "step": 4572 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.3462, + "step": 4573 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4574 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.0921, + "step": 4575 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.3995, + "step": 4576 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4577 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4578 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.0125, + "step": 4579 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 3.4512, + "step": 4580 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4581 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.8169, + "step": 4582 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 7.8384, + "step": 4583 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.0636, + "step": 4584 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 5.7123, + "step": 4585 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.8482, + "step": 4586 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 2.3829, + "step": 4587 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4588 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.8829, + "step": 4589 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 6.9663, + "step": 4590 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4591 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 9.2381, + "step": 4592 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4593 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 6.3808, + "step": 4594 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4595 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4596 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.5677, + "step": 4597 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 1.415, + "step": 4598 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4599 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 7.7438, + "step": 4600 + }, + { + "epoch": 0.38, + "eval_loss": NaN, + "eval_runtime": 41.1741, + "eval_samples_per_second": 60.718, + "eval_steps_per_second": 7.602, + "step": 4600 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4601 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4602 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 4.112, + "step": 4603 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4604 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4605 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.9809, + "step": 4606 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4607 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4608 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 7.3668, + "step": 4609 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4610 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4611 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 1.5387, + "step": 4612 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 5.3491, + "step": 4613 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.3615, + "step": 4614 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4615 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.6417, + "step": 4616 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4617 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4618 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4619 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4620 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4621 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.9813, + "step": 4622 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 6.8033, + "step": 4623 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4624 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4625 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 5.8787, + "step": 4626 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4627 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 8.7811, + "step": 4628 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4629 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.3431, + "step": 4630 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.543, + "step": 4631 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4632 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 4.268, + "step": 4633 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.4181, + "step": 4634 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4635 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4636 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.8579, + "step": 4637 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4638 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.9934, + "step": 4639 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.7518, + "step": 4640 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 10.9265, + "step": 4641 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.2145, + "step": 4642 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4643 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.9117, + "step": 4644 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 1.7495, + "step": 4645 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 1.4639, + "step": 4646 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4647 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4648 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.8533, + "step": 4649 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.1039, + "step": 4650 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 4.567, + "step": 4651 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4652 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.617, + "step": 4653 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4654 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4655 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 4.4319, + "step": 4656 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4657 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4658 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4659 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 1.7794, + "step": 4660 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.9196, + "step": 4661 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 9.5207, + "step": 4662 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4663 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4664 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4665 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 9.4236, + "step": 4666 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 6.9596, + "step": 4667 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4668 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 6.0181, + "step": 4669 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4670 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.4462, + "step": 4671 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.8118, + "step": 4672 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4673 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.721, + "step": 4674 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4675 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 5.0271, + "step": 4676 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.8687, + "step": 4677 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4678 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4679 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 7.2338, + "step": 4680 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4681 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.535, + "step": 4682 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4683 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4684 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4685 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.991, + "step": 4686 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4687 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4688 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.9988, + "step": 4689 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4690 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4691 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 5.4754, + "step": 4692 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4693 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4694 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4695 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.8336, + "step": 4696 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4697 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4698 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 1.9906, + "step": 4699 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4700 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 4.4334, + "step": 4701 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4702 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4703 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 8.8113, + "step": 4704 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 4.2702, + "step": 4705 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4706 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4707 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 8.1583, + "step": 4708 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4709 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.9649, + "step": 4710 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4711 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.1896, + "step": 4712 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.2306, + "step": 4713 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 1.0092, + "step": 4714 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4715 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.0531, + "step": 4716 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 3.9169, + "step": 4717 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 2.1048, + "step": 4718 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 4.2577, + "step": 4719 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4720 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 4.6472, + "step": 4721 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.9133, + "step": 4722 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4723 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 14.7549, + "step": 4724 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001999944269512637, + "loss": 0.8822, + "step": 4725 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.0606, + "step": 4726 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4727 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.8088, + "step": 4728 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 3.6435, + "step": 4729 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4730 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4731 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4732 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4733 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 2.291, + "step": 4734 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4735 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4736 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 7.8673, + "step": 4737 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 5.2309, + "step": 4738 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4739 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.6296, + "step": 4740 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 3.8614, + "step": 4741 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.9629, + "step": 4742 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4743 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 3.6215, + "step": 4744 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4745 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.0314, + "step": 4746 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 2.0005, + "step": 4747 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 9.4879, + "step": 4748 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.0861, + "step": 4749 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.9129, + "step": 4750 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4751 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.9115, + "step": 4752 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.9794, + "step": 4753 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 2.1153, + "step": 4754 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4755 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4756 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 2.3824, + "step": 4757 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4758 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4759 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4760 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.8596, + "step": 4761 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4762 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 3.7762, + "step": 4763 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 3.1644, + "step": 4764 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.9146, + "step": 4765 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.1446, + "step": 4766 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 2.2315, + "step": 4767 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.8267, + "step": 4768 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4769 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4770 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4771 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4772 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 2.2006, + "step": 4773 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.8578, + "step": 4774 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 12.4335, + "step": 4775 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4776 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.7993, + "step": 4777 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4778 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4779 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4780 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.8854, + "step": 4781 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.9791, + "step": 4782 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4783 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.5308, + "step": 4784 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4785 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4786 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 3.7972, + "step": 4787 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4788 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 5.5023, + "step": 4789 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 8.0567, + "step": 4790 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.0363, + "step": 4791 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 6.9657, + "step": 4792 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.2463, + "step": 4793 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.9434, + "step": 4794 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4795 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 7.5022, + "step": 4796 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 8.3325, + "step": 4797 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4798 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4799 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.2829, + "step": 4800 + }, + { + "epoch": 0.4, + "eval_loss": NaN, + "eval_runtime": 40.1297, + "eval_samples_per_second": 62.298, + "eval_steps_per_second": 7.8, + "step": 4800 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4801 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4802 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.1412, + "step": 4803 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 7.8585, + "step": 4804 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 8.0239, + "step": 4805 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4806 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.5397, + "step": 4807 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 8.4512, + "step": 4808 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.0042, + "step": 4809 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4810 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4811 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4812 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.8319, + "step": 4813 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4814 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.944, + "step": 4815 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 8.686, + "step": 4816 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4817 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4818 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4819 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 8.0752, + "step": 4820 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4821 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 4.4986, + "step": 4822 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 7.9226, + "step": 4823 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4824 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4825 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 9.9741, + "step": 4826 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 10.7434, + "step": 4827 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 8.2661, + "step": 4828 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 7.5527, + "step": 4829 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4830 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4831 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.1926, + "step": 4832 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 3.0228, + "step": 4833 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4834 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4835 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.9993, + "step": 4836 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 2.0633, + "step": 4837 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4838 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4839 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 1.9094, + "step": 4840 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4841 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 7.6388, + "step": 4842 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 7.1735, + "step": 4843 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4844 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4845 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4846 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4847 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4848 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4849 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.4707, + "step": 4850 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 4.1881, + "step": 4851 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4852 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4853 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.9505, + "step": 4854 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4855 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4856 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4857 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4858 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4859 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4860 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 6.4831, + "step": 4861 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.0604, + "step": 4862 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 4.1035, + "step": 4863 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4864 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 6.4102, + "step": 4865 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4866 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4867 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 3.6348, + "step": 4868 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4869 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.0346, + "step": 4870 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4871 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.1278, + "step": 4872 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 5.532, + "step": 4873 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4874 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.8083, + "step": 4875 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.8802, + "step": 4876 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.2168, + "step": 4877 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 6.8421, + "step": 4878 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4879 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 3.7116, + "step": 4880 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 3.4613, + "step": 4881 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4882 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 4.642, + "step": 4883 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.0023, + "step": 4884 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.8196, + "step": 4885 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.9584, + "step": 4886 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4887 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4888 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.965, + "step": 4889 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4890 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4891 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.1243, + "step": 4892 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.7546, + "step": 4893 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.1815, + "step": 4894 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4895 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 3.043, + "step": 4896 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 4.2986, + "step": 4897 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 5.8559, + "step": 4898 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 7.0516, + "step": 4899 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 9.344, + "step": 4900 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4901 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 8.6394, + "step": 4902 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.6881, + "step": 4903 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 3.5096, + "step": 4904 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4905 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.0256, + "step": 4906 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 6.5703, + "step": 4907 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4908 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4909 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4910 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 5.1077, + "step": 4911 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4912 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4913 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4914 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4915 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4916 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.5674, + "step": 4917 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4918 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 4.2448, + "step": 4919 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4920 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 10.8526, + "step": 4921 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 8.3761, + "step": 4922 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4923 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4924 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4925 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4926 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 3.8155, + "step": 4927 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.3811, + "step": 4928 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4929 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.5036, + "step": 4930 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4931 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4932 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 7.1385, + "step": 4933 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4934 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 8.6717, + "step": 4935 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4936 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4937 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 4.2382, + "step": 4938 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4939 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4940 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 8.4623, + "step": 4941 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4942 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4943 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4944 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4945 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.6917, + "step": 4946 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 3.6276, + "step": 4947 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4948 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.8416, + "step": 4949 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4950 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.6815, + "step": 4951 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4952 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4953 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4954 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4955 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.8347, + "step": 4956 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4957 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.1119, + "step": 4958 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.0408, + "step": 4959 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4960 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.7919, + "step": 4961 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.8243, + "step": 4962 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 2.2866, + "step": 4963 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001999944269512637, + "loss": 1.5979, + "step": 4964 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.8258, + "step": 4965 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.9925, + "step": 4966 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 9.2982, + "step": 4967 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4968 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4969 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.9007, + "step": 4970 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.2756, + "step": 4971 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4972 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4973 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4974 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.2723, + "step": 4975 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.0516, + "step": 4976 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4977 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4978 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4979 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4980 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4981 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4982 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.4002, + "step": 4983 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.6494, + "step": 4984 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4985 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.9073, + "step": 4986 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.1477, + "step": 4987 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4988 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 3.2145, + "step": 4989 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.9263, + "step": 4990 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4991 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 7.9136, + "step": 4992 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4993 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4994 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.7172, + "step": 4995 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.8725, + "step": 4996 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4997 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.8287, + "step": 4998 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 4999 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 3.341, + "step": 5000 + }, + { + "epoch": 0.42, + "eval_loss": NaN, + "eval_runtime": 40.0329, + "eval_samples_per_second": 62.449, + "eval_steps_per_second": 7.819, + "step": 5000 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5001 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 7.8591, + "step": 5002 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.6026, + "step": 5003 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 3.8053, + "step": 5004 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.2165, + "step": 5005 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 3.2614, + "step": 5006 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.7974, + "step": 5007 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.2407, + "step": 5008 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.1453, + "step": 5009 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 5.1291, + "step": 5010 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.2739, + "step": 5011 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5012 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 8.363, + "step": 5013 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5014 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5015 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.9634, + "step": 5016 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 6.5448, + "step": 5017 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 8.5133, + "step": 5018 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 9.2626, + "step": 5019 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.1523, + "step": 5020 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 3.3259, + "step": 5021 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5022 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.8977, + "step": 5023 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.9839, + "step": 5024 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5025 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 6.8405, + "step": 5026 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5027 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5028 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 12.8137, + "step": 5029 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.8789, + "step": 5030 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.0368, + "step": 5031 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5032 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5033 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 5.5106, + "step": 5034 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5035 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5036 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5037 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5038 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 10.6491, + "step": 5039 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 9.3801, + "step": 5040 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5041 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5042 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.9411, + "step": 5043 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5044 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5045 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5046 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.428, + "step": 5047 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5048 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 3.4092, + "step": 5049 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5050 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 5.7866, + "step": 5051 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5052 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.7651, + "step": 5053 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5054 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5055 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.4649, + "step": 5056 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5057 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.1494, + "step": 5058 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5059 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 5.1851, + "step": 5060 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5061 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.9861, + "step": 5062 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5063 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.4389, + "step": 5064 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 2.5534, + "step": 5065 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 5.5075, + "step": 5066 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5067 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5068 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5069 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 7.5214, + "step": 5070 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.6744, + "step": 5071 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5072 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5073 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5074 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 7.2259, + "step": 5075 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.9773, + "step": 5076 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5077 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.4918, + "step": 5078 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.1152, + "step": 5079 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.3564, + "step": 5080 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5081 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5082 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 4.4867, + "step": 5083 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001999944269512637, + "loss": 1.9661, + "step": 5084 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5085 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5086 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5087 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5088 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 8.8428, + "step": 5089 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 8.236, + "step": 5090 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5091 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 2.0085, + "step": 5092 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 6.695, + "step": 5093 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5094 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5095 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.6206, + "step": 5096 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5097 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5098 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5099 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 8.2339, + "step": 5100 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5101 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.835, + "step": 5102 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5103 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 6.9378, + "step": 5104 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.5401, + "step": 5105 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5106 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.7478, + "step": 5107 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.2874, + "step": 5108 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5109 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5110 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5111 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.9736, + "step": 5112 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 8.3938, + "step": 5113 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.4045, + "step": 5114 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.0595, + "step": 5115 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 2.1636, + "step": 5116 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5117 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.0916, + "step": 5118 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5119 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5120 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5121 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.9559, + "step": 5122 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5123 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5124 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.0383, + "step": 5125 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.6454, + "step": 5126 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5127 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.0671, + "step": 5128 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5129 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.4118, + "step": 5130 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5131 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5132 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 6.5865, + "step": 5133 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 2.1693, + "step": 5134 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5135 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5136 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.8165, + "step": 5137 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5138 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 6.9237, + "step": 5139 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.1518, + "step": 5140 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.1418, + "step": 5141 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5142 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5143 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5144 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5145 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5146 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5147 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 2.9318, + "step": 5148 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.7567, + "step": 5149 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5150 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 5.6625, + "step": 5151 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5152 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.881, + "step": 5153 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 8.0184, + "step": 5154 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5155 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 4.1821, + "step": 5156 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5157 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 4.5019, + "step": 5158 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5159 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.8495, + "step": 5160 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.2925, + "step": 5161 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.7645, + "step": 5162 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5163 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.0848, + "step": 5164 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5165 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.9075, + "step": 5166 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 4.4844, + "step": 5167 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.0202, + "step": 5168 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.459, + "step": 5169 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5170 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 4.0096, + "step": 5171 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5172 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5173 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 2.9897, + "step": 5174 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 2.1367, + "step": 5175 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.1449, + "step": 5176 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 4.2863, + "step": 5177 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5178 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5179 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5180 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 3.3039, + "step": 5181 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 1.6271, + "step": 5182 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5183 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5184 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5185 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 4.778, + "step": 5186 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.0401, + "step": 5187 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5188 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 9.0927, + "step": 5189 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.9069, + "step": 5190 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5191 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5192 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5193 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5194 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 2.291, + "step": 5195 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5196 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 7.6312, + "step": 5197 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 5.2315, + "step": 5198 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5199 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 5.355, + "step": 5200 + }, + { + "epoch": 0.43, + "eval_loss": NaN, + "eval_runtime": 40.0471, + "eval_samples_per_second": 62.427, + "eval_steps_per_second": 7.816, + "step": 5200 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5201 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 5.4077, + "step": 5202 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5203 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.1345, + "step": 5204 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.8494, + "step": 5205 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.9247, + "step": 5206 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.8927, + "step": 5207 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5208 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 3.6575, + "step": 5209 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.4363, + "step": 5210 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.0495, + "step": 5211 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 6.5794, + "step": 5212 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.589, + "step": 5213 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.8396, + "step": 5214 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5215 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5216 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 5.54, + "step": 5217 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5218 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.9053, + "step": 5219 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.5916, + "step": 5220 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5221 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 10.5049, + "step": 5222 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5223 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.4162, + "step": 5224 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5225 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.8671, + "step": 5226 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 5.1063, + "step": 5227 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5228 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 5.3431, + "step": 5229 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5230 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5231 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.2747, + "step": 5232 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5233 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.4944, + "step": 5234 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.057, + "step": 5235 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5236 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5237 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.2724, + "step": 5238 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.2996, + "step": 5239 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.0576, + "step": 5240 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 8.5208, + "step": 5241 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5242 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.1662, + "step": 5243 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.991, + "step": 5244 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.983, + "step": 5245 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 3.728, + "step": 5246 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5247 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.8072, + "step": 5248 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.0577, + "step": 5249 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.0958, + "step": 5250 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5251 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5252 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.0813, + "step": 5253 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 6.9468, + "step": 5254 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5255 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.343, + "step": 5256 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.8031, + "step": 5257 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5258 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5259 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.3472, + "step": 5260 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.9276, + "step": 5261 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.2706, + "step": 5262 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5263 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 6.6216, + "step": 5264 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.1273, + "step": 5265 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.0229, + "step": 5266 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 8.695, + "step": 5267 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 9.0335, + "step": 5268 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5269 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5270 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.922, + "step": 5271 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5272 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 9.4428, + "step": 5273 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.5378, + "step": 5274 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 3.118, + "step": 5275 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5276 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.7403, + "step": 5277 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 8.9404, + "step": 5278 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 3.2448, + "step": 5279 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.0186, + "step": 5280 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.0566, + "step": 5281 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.8429, + "step": 5282 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 6.6545, + "step": 5283 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5284 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5285 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 3.6172, + "step": 5286 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5287 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 5.0041, + "step": 5288 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5289 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 4.5257, + "step": 5290 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.7466, + "step": 5291 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5292 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5293 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5294 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.9151, + "step": 5295 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5296 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 8.7801, + "step": 5297 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5298 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5299 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.8333, + "step": 5300 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 8.7382, + "step": 5301 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5302 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.5677, + "step": 5303 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5304 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.9188, + "step": 5305 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.1734, + "step": 5306 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.0357, + "step": 5307 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.7598, + "step": 5308 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.9201, + "step": 5309 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5310 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.6504, + "step": 5311 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5312 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 7.9592, + "step": 5313 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.7711, + "step": 5314 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5315 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5316 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5317 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 3.2214, + "step": 5318 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 8.1282, + "step": 5319 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5320 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 1.924, + "step": 5321 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 2.7618, + "step": 5322 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5323 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5324 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 9.1152, + "step": 5325 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.0839, + "step": 5326 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.6924, + "step": 5327 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.0153, + "step": 5328 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.1154, + "step": 5329 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5330 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.3263, + "step": 5331 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.4859, + "step": 5332 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5333 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5334 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 8.1279, + "step": 5335 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5336 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 6.0894, + "step": 5337 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5338 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.154, + "step": 5339 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5340 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5341 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5342 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 6.134, + "step": 5343 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5344 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.156, + "step": 5345 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5346 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.652, + "step": 5347 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.105, + "step": 5348 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5349 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.8612, + "step": 5350 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.8176, + "step": 5351 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5352 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9281, + "step": 5353 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 9.0561, + "step": 5354 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5355 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5356 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5357 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.7794, + "step": 5358 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5359 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.8909, + "step": 5360 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5361 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5362 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5363 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5364 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5365 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5366 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5367 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5368 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9462, + "step": 5369 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.4059, + "step": 5370 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5371 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5372 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5373 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5374 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5375 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.786, + "step": 5376 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.6419, + "step": 5377 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.341, + "step": 5378 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 7.6519, + "step": 5379 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 8.1496, + "step": 5380 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.6361, + "step": 5381 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9301, + "step": 5382 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.3996, + "step": 5383 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5384 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.2308, + "step": 5385 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5386 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.6056, + "step": 5387 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.1055, + "step": 5388 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 6.0518, + "step": 5389 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.2176, + "step": 5390 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.875, + "step": 5391 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9556, + "step": 5392 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5393 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.3776, + "step": 5394 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 8.8804, + "step": 5395 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.123, + "step": 5396 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5397 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 7.0444, + "step": 5398 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5399 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 7.7009, + "step": 5400 + }, + { + "epoch": 0.45, + "eval_loss": NaN, + "eval_runtime": 40.2048, + "eval_samples_per_second": 62.182, + "eval_steps_per_second": 7.785, + "step": 5400 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5401 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 8.9664, + "step": 5402 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5403 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9511, + "step": 5404 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5405 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 8.385, + "step": 5406 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.8437, + "step": 5407 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.5988, + "step": 5408 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.8812, + "step": 5409 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.4716, + "step": 5410 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5411 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.4921, + "step": 5412 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9042, + "step": 5413 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5414 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.2524, + "step": 5415 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 7.4806, + "step": 5416 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.6708, + "step": 5417 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.6852, + "step": 5418 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.4433, + "step": 5419 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5420 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5421 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.8989, + "step": 5422 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 2.2025, + "step": 5423 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5424 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.6902, + "step": 5425 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5426 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5427 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.7248, + "step": 5428 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9226, + "step": 5429 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.9047, + "step": 5430 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5431 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5432 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 5.3621, + "step": 5433 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.0859, + "step": 5434 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 4.0815, + "step": 5435 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 5.1098, + "step": 5436 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 6.0242, + "step": 5437 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.4859, + "step": 5438 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 10.4101, + "step": 5439 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 1.1755, + "step": 5440 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 3.6526, + "step": 5441 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5442 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001999944269512637, + "loss": 5.5969, + "step": 5443 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.0436, + "step": 5444 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5445 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5446 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 5.2444, + "step": 5447 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5448 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 7.7401, + "step": 5449 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5450 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5451 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5452 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5453 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5454 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.4218, + "step": 5455 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5456 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5457 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.9544, + "step": 5458 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.3151, + "step": 5459 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5460 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5461 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.3477, + "step": 5462 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5463 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.7056, + "step": 5464 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.6757, + "step": 5465 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.129, + "step": 5466 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.027, + "step": 5467 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5468 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.2556, + "step": 5469 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5470 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.7469, + "step": 5471 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 7.1253, + "step": 5472 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5473 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.994, + "step": 5474 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.7097, + "step": 5475 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.9017, + "step": 5476 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.5379, + "step": 5477 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.7884, + "step": 5478 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5479 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 5.4544, + "step": 5480 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5481 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.2177, + "step": 5482 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 9.1409, + "step": 5483 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5484 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 2.1301, + "step": 5485 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5486 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.142, + "step": 5487 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5488 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.8134, + "step": 5489 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 7.1315, + "step": 5490 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.7888, + "step": 5491 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.9929, + "step": 5492 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.0347, + "step": 5493 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 2.9189, + "step": 5494 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5495 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.9598, + "step": 5496 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.0929, + "step": 5497 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5498 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.8908, + "step": 5499 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5500 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5501 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5502 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5503 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5504 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 5.8671, + "step": 5505 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.0062, + "step": 5506 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.1261, + "step": 5507 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5508 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.9067, + "step": 5509 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 3.0484, + "step": 5510 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5511 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.0242, + "step": 5512 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 7.9231, + "step": 5513 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5514 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5515 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5516 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5517 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5518 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5519 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5520 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5521 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5522 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5523 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.0431, + "step": 5524 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 2.0258, + "step": 5525 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.0191, + "step": 5526 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.2032, + "step": 5527 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5528 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.0398, + "step": 5529 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 7.1814, + "step": 5530 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5531 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 9.4119, + "step": 5532 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.0656, + "step": 5533 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5534 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5535 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 7.927, + "step": 5536 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5537 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5538 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.9475, + "step": 5539 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5540 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5541 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5542 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 5.2138, + "step": 5543 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.5955, + "step": 5544 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.0438, + "step": 5545 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5546 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5547 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5548 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.9042, + "step": 5549 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.9656, + "step": 5550 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5551 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5552 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 4.0624, + "step": 5553 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 2.6387, + "step": 5554 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5555 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5556 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.9615, + "step": 5557 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.8704, + "step": 5558 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5559 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 8.3905, + "step": 5560 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 1.9065, + "step": 5561 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5562 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.9738, + "step": 5563 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 6.2939, + "step": 5564 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5565 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.7966, + "step": 5566 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.9504, + "step": 5567 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5568 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5569 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5570 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5571 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5572 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 8.5027, + "step": 5573 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5574 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.0616, + "step": 5575 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.6496, + "step": 5576 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.3681, + "step": 5577 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5578 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5579 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5580 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.104, + "step": 5581 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5582 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5583 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.2037, + "step": 5584 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.5622, + "step": 5585 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5586 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5587 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.2861, + "step": 5588 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.1948, + "step": 5589 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 8.3288, + "step": 5590 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 7.5725, + "step": 5591 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.9191, + "step": 5592 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5593 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5594 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5595 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5596 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5597 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5598 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 6.2793, + "step": 5599 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.287, + "step": 5600 + }, + { + "epoch": 0.47, + "eval_loss": NaN, + "eval_runtime": 39.9421, + "eval_samples_per_second": 62.591, + "eval_steps_per_second": 7.836, + "step": 5600 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5601 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.0635, + "step": 5602 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5603 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.0827, + "step": 5604 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5605 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 9.1371, + "step": 5606 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5607 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.6186, + "step": 5608 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5609 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 7.3775, + "step": 5610 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.9158, + "step": 5611 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5612 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5613 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 8.2075, + "step": 5614 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5615 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5616 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5617 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.7377, + "step": 5618 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5619 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.8724, + "step": 5620 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.2433, + "step": 5621 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5622 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5623 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 7.0413, + "step": 5624 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5625 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.8848, + "step": 5626 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5627 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 9.451, + "step": 5628 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5629 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.4764, + "step": 5630 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 5.0169, + "step": 5631 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5632 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.2201, + "step": 5633 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.9388, + "step": 5634 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.2619, + "step": 5635 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 6.0171, + "step": 5636 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.0151, + "step": 5637 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 6.0822, + "step": 5638 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5639 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.2311, + "step": 5640 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5641 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5642 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.7936, + "step": 5643 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 9.238, + "step": 5644 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5645 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.3603, + "step": 5646 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.6871, + "step": 5647 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5648 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.0976, + "step": 5649 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.6458, + "step": 5650 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 8.8857, + "step": 5651 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5652 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5653 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 8.6126, + "step": 5654 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.8025, + "step": 5655 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 9.2735, + "step": 5656 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 7.0492, + "step": 5657 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5658 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.8365, + "step": 5659 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.823, + "step": 5660 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.4686, + "step": 5661 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5662 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.667, + "step": 5663 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 8.0943, + "step": 5664 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.5936, + "step": 5665 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.1544, + "step": 5666 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5667 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5668 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5669 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 3.9897, + "step": 5670 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 5.4574, + "step": 5671 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5672 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5673 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 1.4504, + "step": 5674 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5675 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 5.2419, + "step": 5676 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 4.2174, + "step": 5677 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 2.1081, + "step": 5678 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5679 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5680 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5681 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5682 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5683 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5684 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.0197, + "step": 5685 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5686 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.0332, + "step": 5687 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5688 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5689 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5690 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5691 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.8352, + "step": 5692 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5693 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 8.2053, + "step": 5694 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 9.5068, + "step": 5695 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5696 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5697 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.9055, + "step": 5698 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.0258, + "step": 5699 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5700 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5701 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5702 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5703 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5704 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 4.5006, + "step": 5705 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5706 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.6443, + "step": 5707 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5708 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 5.7939, + "step": 5709 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.9139, + "step": 5710 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5711 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5712 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5713 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5714 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.8349, + "step": 5715 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5716 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.8033, + "step": 5717 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.8567, + "step": 5718 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 4.4913, + "step": 5719 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5720 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.9222, + "step": 5721 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5722 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.8755, + "step": 5723 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.4831, + "step": 5724 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5725 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5726 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5727 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 2.5199, + "step": 5728 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.951, + "step": 5729 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5730 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5731 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5732 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5733 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.6996, + "step": 5734 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 2.0262, + "step": 5735 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 10.2554, + "step": 5736 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.664, + "step": 5737 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5738 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.8823, + "step": 5739 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5740 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5741 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5742 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 8.5644, + "step": 5743 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 10.0472, + "step": 5744 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5745 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 2.2373, + "step": 5746 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.2249, + "step": 5747 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5748 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.6605, + "step": 5749 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.7943, + "step": 5750 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5751 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5752 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5753 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5754 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 6.7104, + "step": 5755 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5756 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5757 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5758 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5759 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 2.6488, + "step": 5760 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5761 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5762 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5763 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 7.0474, + "step": 5764 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5765 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5766 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 4.0753, + "step": 5767 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.9628, + "step": 5768 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 7.9303, + "step": 5769 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.0606, + "step": 5770 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.55, + "step": 5771 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5772 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5773 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.9916, + "step": 5774 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 4.1454, + "step": 5775 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5776 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5777 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5778 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 2.3314, + "step": 5779 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5780 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 2.1186, + "step": 5781 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5782 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5783 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 9.2391, + "step": 5784 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.9902, + "step": 5785 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.0919, + "step": 5786 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.818, + "step": 5787 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.8843, + "step": 5788 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 7.586, + "step": 5789 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 7.911, + "step": 5790 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 1.1225, + "step": 5791 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.9566, + "step": 5792 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 4.3428, + "step": 5793 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 8.1662, + "step": 5794 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.184, + "step": 5795 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5796 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 6.1145, + "step": 5797 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 6.0491, + "step": 5798 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 5.6221, + "step": 5799 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 7.6782, + "step": 5800 + }, + { + "epoch": 0.48, + "eval_loss": NaN, + "eval_runtime": 41.3134, + "eval_samples_per_second": 60.513, + "eval_steps_per_second": 7.576, + "step": 5800 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 5.2503, + "step": 5801 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001999944269512637, + "loss": 3.1509, + "step": 5802 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5803 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5804 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5805 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.0374, + "step": 5806 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 6.2481, + "step": 5807 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5808 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5809 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5810 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.877, + "step": 5811 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5812 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.7933, + "step": 5813 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 7.6739, + "step": 5814 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 7.8761, + "step": 5815 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5816 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.5646, + "step": 5817 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5818 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5819 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5820 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.7731, + "step": 5821 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5822 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.8123, + "step": 5823 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5824 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5825 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5826 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5827 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5828 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5829 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.2263, + "step": 5830 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.101, + "step": 5831 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5832 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5833 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.677, + "step": 5834 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5835 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5836 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 8.7229, + "step": 5837 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 6.0873, + "step": 5838 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5839 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.0398, + "step": 5840 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5841 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5842 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.2335, + "step": 5843 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.9764, + "step": 5844 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.1075, + "step": 5845 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.2471, + "step": 5846 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 3.9959, + "step": 5847 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.2101, + "step": 5848 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.7904, + "step": 5849 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5850 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.227, + "step": 5851 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5852 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.0088, + "step": 5853 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.7293, + "step": 5854 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 7.9772, + "step": 5855 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.2675, + "step": 5856 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5857 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 9.0956, + "step": 5858 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 3.8009, + "step": 5859 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.9795, + "step": 5860 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.7458, + "step": 5861 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.2556, + "step": 5862 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.5096, + "step": 5863 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 3.0413, + "step": 5864 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5865 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5866 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.3494, + "step": 5867 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5868 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5869 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 3.4426, + "step": 5870 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 3.3619, + "step": 5871 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.3867, + "step": 5872 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 7.9619, + "step": 5873 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.5752, + "step": 5874 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5875 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5876 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5877 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.0463, + "step": 5878 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 3.8628, + "step": 5879 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5880 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5881 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.9655, + "step": 5882 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.612, + "step": 5883 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.5448, + "step": 5884 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 6.304, + "step": 5885 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5886 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5887 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5888 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5889 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5890 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5891 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 5.3507, + "step": 5892 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.6569, + "step": 5893 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 5.5781, + "step": 5894 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 3.8961, + "step": 5895 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5896 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.0455, + "step": 5897 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5898 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5899 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5900 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5901 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5902 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5903 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.8591, + "step": 5904 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 6.3689, + "step": 5905 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5906 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 8.4125, + "step": 5907 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5908 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5909 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5910 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.7154, + "step": 5911 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 1.3514, + "step": 5912 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5913 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5914 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 2.2228, + "step": 5915 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.5278, + "step": 5916 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 7.0507, + "step": 5917 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5918 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5919 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 6.9478, + "step": 5920 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001999944269512637, + "loss": 4.3688, + "step": 5921 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5922 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.0801, + "step": 5923 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5924 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 6.4974, + "step": 5925 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 8.1777, + "step": 5926 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5927 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.3376, + "step": 5928 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5929 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.1006, + "step": 5930 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.5741, + "step": 5931 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.7721, + "step": 5932 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5933 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5934 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 6.3714, + "step": 5935 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 9.3663, + "step": 5936 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.988, + "step": 5937 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.1506, + "step": 5938 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5939 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5940 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5941 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.6701, + "step": 5942 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5943 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.447, + "step": 5944 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5945 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 6.4868, + "step": 5946 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.194, + "step": 5947 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.6822, + "step": 5948 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.9489, + "step": 5949 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.818, + "step": 5950 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.8887, + "step": 5951 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.5656, + "step": 5952 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5953 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5954 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5955 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5956 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5957 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5958 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.3039, + "step": 5959 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5960 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5961 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.7399, + "step": 5962 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5963 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 9.0514, + "step": 5964 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.4597, + "step": 5965 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5966 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.2241, + "step": 5967 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5968 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.3498, + "step": 5969 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.615, + "step": 5970 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.3458, + "step": 5971 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.8183, + "step": 5972 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.1153, + "step": 5973 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.7175, + "step": 5974 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5975 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5976 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5977 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5978 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5979 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.8891, + "step": 5980 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5981 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.443, + "step": 5982 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.6258, + "step": 5983 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.8168, + "step": 5984 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.9169, + "step": 5985 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.2006, + "step": 5986 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5987 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5988 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5989 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.3514, + "step": 5990 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.8115, + "step": 5991 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.1212, + "step": 5992 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5993 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 6.4018, + "step": 5994 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5995 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 5.9393, + "step": 5996 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.2022, + "step": 5997 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 5998 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.1224, + "step": 5999 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 12.0547, + "step": 6000 + }, + { + "epoch": 0.5, + "eval_loss": NaN, + "eval_runtime": 41.2794, + "eval_samples_per_second": 60.563, + "eval_steps_per_second": 7.582, + "step": 6000 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6001 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.6078, + "step": 6002 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6003 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.9191, + "step": 6004 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 1.9308, + "step": 6005 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6006 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.2405, + "step": 6007 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6008 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6009 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.2287, + "step": 6010 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6011 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6012 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6013 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.0672, + "step": 6014 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 8.1747, + "step": 6015 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.1576, + "step": 6016 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 11.2015, + "step": 6017 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.9718, + "step": 6018 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.0533, + "step": 6019 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.109, + "step": 6020 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.9855, + "step": 6021 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6022 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6023 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6024 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 7.1317, + "step": 6025 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6026 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.3573, + "step": 6027 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.1909, + "step": 6028 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6029 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 8.0352, + "step": 6030 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 8.2576, + "step": 6031 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 2.159, + "step": 6032 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6033 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6034 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.7916, + "step": 6035 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6036 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6037 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 4.7322, + "step": 6038 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.0743, + "step": 6039 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 3.0124, + "step": 6040 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6041 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6042 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 11.3568, + "step": 6043 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.766, + "step": 6044 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.7121, + "step": 6045 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 8.5391, + "step": 6046 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 6.6968, + "step": 6047 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.9672, + "step": 6048 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6049 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.556, + "step": 6050 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6051 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6052 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6053 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6054 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6055 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.0495, + "step": 6056 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 8.4621, + "step": 6057 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 6.3241, + "step": 6058 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.865, + "step": 6059 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6060 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.6759, + "step": 6061 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6062 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.5341, + "step": 6063 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.7814, + "step": 6064 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.026, + "step": 6065 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 9.3084, + "step": 6066 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6067 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.456, + "step": 6068 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6069 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.9422, + "step": 6070 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 8.0665, + "step": 6071 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.1515, + "step": 6072 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6073 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 5.4737, + "step": 6074 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6075 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6076 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.9816, + "step": 6077 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.2621, + "step": 6078 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6079 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 5.5845, + "step": 6080 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.2932, + "step": 6081 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.1857, + "step": 6082 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.8257, + "step": 6083 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.7008, + "step": 6084 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6085 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6086 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 8.1598, + "step": 6087 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.2066, + "step": 6088 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6089 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.7134, + "step": 6090 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.0052, + "step": 6091 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.1096, + "step": 6092 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6093 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6094 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.8253, + "step": 6095 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.9476, + "step": 6096 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.4681, + "step": 6097 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6098 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 9.869, + "step": 6099 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.4605, + "step": 6100 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6101 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 5.3341, + "step": 6102 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6103 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 7.6963, + "step": 6104 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6105 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6106 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.747, + "step": 6107 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.1273, + "step": 6108 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6109 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.9161, + "step": 6110 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6111 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6112 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.2606, + "step": 6113 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6114 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.3992, + "step": 6115 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6116 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6117 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.0664, + "step": 6118 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.8117, + "step": 6119 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.2163, + "step": 6120 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.9367, + "step": 6121 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6122 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 5.8315, + "step": 6123 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6124 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6125 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6126 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6127 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.5427, + "step": 6128 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6129 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.1099, + "step": 6130 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.6703, + "step": 6131 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.7692, + "step": 6132 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.1615, + "step": 6133 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6134 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6135 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.2383, + "step": 6136 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 10.9645, + "step": 6137 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.4478, + "step": 6138 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.386, + "step": 6139 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.7045, + "step": 6140 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.0081, + "step": 6141 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6142 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6143 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.6581, + "step": 6144 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 9.532, + "step": 6145 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.3879, + "step": 6146 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6147 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 3.9421, + "step": 6148 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6149 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6150 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.438, + "step": 6151 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6152 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6153 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.0463, + "step": 6154 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 1.0332, + "step": 6155 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.0976, + "step": 6156 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6157 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 4.3654, + "step": 6158 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 2.8083, + "step": 6159 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001999944269512637, + "loss": 0.8902, + "step": 6160 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6161 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6162 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6163 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6164 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.7329, + "step": 6165 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6166 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 3.5793, + "step": 6167 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 7.953, + "step": 6168 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.5632, + "step": 6169 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6170 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 5.1453, + "step": 6171 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 3.866, + "step": 6172 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.7149, + "step": 6173 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.0248, + "step": 6174 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.0773, + "step": 6175 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 8.4077, + "step": 6176 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6177 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 8.0881, + "step": 6178 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6179 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6180 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 8.7204, + "step": 6181 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6182 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.8321, + "step": 6183 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6184 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.0008, + "step": 6185 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 8.0229, + "step": 6186 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.0325, + "step": 6187 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.9892, + "step": 6188 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.8435, + "step": 6189 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6190 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 8.9151, + "step": 6191 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6192 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.7908, + "step": 6193 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.17, + "step": 6194 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6195 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.2071, + "step": 6196 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6197 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6198 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6199 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 9.0119, + "step": 6200 + }, + { + "epoch": 0.52, + "eval_loss": NaN, + "eval_runtime": 40.9823, + "eval_samples_per_second": 61.002, + "eval_steps_per_second": 7.637, + "step": 6200 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6201 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6202 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6203 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6204 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 7.0606, + "step": 6205 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.5689, + "step": 6206 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.971, + "step": 6207 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6208 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.817, + "step": 6209 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 6.359, + "step": 6210 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.9571, + "step": 6211 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 6.635, + "step": 6212 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 7.4684, + "step": 6213 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.7489, + "step": 6214 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6215 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.0035, + "step": 6216 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6217 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.1516, + "step": 6218 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6219 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6220 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.8744, + "step": 6221 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6222 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.7712, + "step": 6223 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.9399, + "step": 6224 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.4859, + "step": 6225 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6226 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.0908, + "step": 6227 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6228 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 6.811, + "step": 6229 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6230 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.0766, + "step": 6231 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.7262, + "step": 6232 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 11.1338, + "step": 6233 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.8891, + "step": 6234 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6235 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6236 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6237 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6238 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6239 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.1375, + "step": 6240 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6241 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 5.2536, + "step": 6242 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6243 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6244 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 7.8467, + "step": 6245 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6246 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6247 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6248 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 5.1312, + "step": 6249 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.8138, + "step": 6250 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6251 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 8.1456, + "step": 6252 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.3515, + "step": 6253 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.8867, + "step": 6254 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.1892, + "step": 6255 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 6.0356, + "step": 6256 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6257 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6258 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 3.8837, + "step": 6259 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6260 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.909, + "step": 6261 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6262 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.1474, + "step": 6263 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6264 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 3.6641, + "step": 6265 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.7393, + "step": 6266 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6267 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 7.9967, + "step": 6268 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.1159, + "step": 6269 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6270 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 2.2501, + "step": 6271 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6272 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6273 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.62, + "step": 6274 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 1.7448, + "step": 6275 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 0.9233, + "step": 6276 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 5.0171, + "step": 6277 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 3.9205, + "step": 6278 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 4.1787, + "step": 6279 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001999944269512637, + "loss": 11.4405, + "step": 6280 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6281 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6282 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.2662, + "step": 6283 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6284 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.6972, + "step": 6285 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.9187, + "step": 6286 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6287 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6288 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.76, + "step": 6289 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.2374, + "step": 6290 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 7.4859, + "step": 6291 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6292 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.857, + "step": 6293 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.2973, + "step": 6294 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6295 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.9841, + "step": 6296 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6297 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.6011, + "step": 6298 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6299 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6300 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.6964, + "step": 6301 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.3066, + "step": 6302 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 6.4195, + "step": 6303 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.3946, + "step": 6304 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.0399, + "step": 6305 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.3693, + "step": 6306 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6307 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 5.3748, + "step": 6308 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6309 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 7.8746, + "step": 6310 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.0113, + "step": 6311 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.5545, + "step": 6312 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.1431, + "step": 6313 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 6.2991, + "step": 6314 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6315 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.924, + "step": 6316 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 5.0547, + "step": 6317 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6318 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 5.5448, + "step": 6319 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 9.9492, + "step": 6320 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.7552, + "step": 6321 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6322 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6323 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6324 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6325 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.9837, + "step": 6326 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.7273, + "step": 6327 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.7163, + "step": 6328 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6329 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6330 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.1223, + "step": 6331 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 9.7083, + "step": 6332 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6333 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.0832, + "step": 6334 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.1275, + "step": 6335 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.7913, + "step": 6336 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.4262, + "step": 6337 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6338 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.071, + "step": 6339 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6340 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.625, + "step": 6341 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6342 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.8314, + "step": 6343 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6344 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6345 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6346 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.7253, + "step": 6347 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 6.741, + "step": 6348 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.987, + "step": 6349 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 8.6264, + "step": 6350 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.4371, + "step": 6351 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.482, + "step": 6352 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6353 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6354 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6355 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.1632, + "step": 6356 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6357 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.8431, + "step": 6358 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6359 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6360 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.9452, + "step": 6361 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6362 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6363 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6364 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6365 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.9145, + "step": 6366 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.8923, + "step": 6367 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.1438, + "step": 6368 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6369 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 8.5719, + "step": 6370 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 8.1646, + "step": 6371 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.9239, + "step": 6372 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6373 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 2.0431, + "step": 6374 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 4.4345, + "step": 6375 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6376 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.9231, + "step": 6377 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6378 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6379 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.0558, + "step": 6380 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 7.2491, + "step": 6381 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6382 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6383 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 3.7004, + "step": 6384 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6385 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.5926, + "step": 6386 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 6.0088, + "step": 6387 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.9745, + "step": 6388 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6389 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6390 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 5.0338, + "step": 6391 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6392 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6393 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6394 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6395 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6396 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6397 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6398 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 1.0153, + "step": 6399 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001999944269512637, + "loss": 6.29, + "step": 6400 + }, + { + "epoch": 0.53, + "eval_loss": NaN, + "eval_runtime": 40.2955, + "eval_samples_per_second": 62.042, + "eval_steps_per_second": 7.768, + "step": 6400 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.0749, + "step": 6401 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 12.0866, + "step": 6402 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 5.0927, + "step": 6403 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 3.7196, + "step": 6404 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6405 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6406 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 4.8141, + "step": 6407 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.8137, + "step": 6408 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 8.287, + "step": 6409 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6410 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.1656, + "step": 6411 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 3.3141, + "step": 6412 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.9241, + "step": 6413 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6414 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.6574, + "step": 6415 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6416 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.1105, + "step": 6417 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.9039, + "step": 6418 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.7539, + "step": 6419 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 3.7582, + "step": 6420 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 6.0988, + "step": 6421 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6422 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6423 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 8.9297, + "step": 6424 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 5.0154, + "step": 6425 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6426 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 6.9768, + "step": 6427 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.0814, + "step": 6428 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 4.0596, + "step": 6429 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6430 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 8.4876, + "step": 6431 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.9635, + "step": 6432 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6433 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.8447, + "step": 6434 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6435 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.0828, + "step": 6436 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6437 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.3212, + "step": 6438 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 6.8518, + "step": 6439 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.8576, + "step": 6440 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6441 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6442 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6443 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.1479, + "step": 6444 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6445 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6446 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 4.7471, + "step": 6447 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 9.0172, + "step": 6448 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6449 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.4004, + "step": 6450 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 6.4422, + "step": 6451 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 6.6053, + "step": 6452 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6453 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6454 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 4.0406, + "step": 6455 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.8793, + "step": 6456 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.6064, + "step": 6457 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6458 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.836, + "step": 6459 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6460 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6461 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6462 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6463 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6464 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6465 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 4.9256, + "step": 6466 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6467 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.8173, + "step": 6468 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 3.6273, + "step": 6469 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.858, + "step": 6470 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.0129, + "step": 6471 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6472 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6473 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 7.7044, + "step": 6474 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.8115, + "step": 6475 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 5.644, + "step": 6476 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.9157, + "step": 6477 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6478 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 6.7794, + "step": 6479 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.9502, + "step": 6480 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6481 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 6.357, + "step": 6482 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6483 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6484 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.8894, + "step": 6485 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6486 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.4732, + "step": 6487 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6488 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6489 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6490 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6491 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.0685, + "step": 6492 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.9414, + "step": 6493 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.2342, + "step": 6494 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.1069, + "step": 6495 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.9724, + "step": 6496 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 5.1856, + "step": 6497 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6498 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 4.5357, + "step": 6499 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6500 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.9108, + "step": 6501 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.1513, + "step": 6502 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6503 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 3.1356, + "step": 6504 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.9612, + "step": 6505 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6506 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6507 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6508 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 4.7331, + "step": 6509 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6510 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6511 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.3291, + "step": 6512 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.5659, + "step": 6513 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6514 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6515 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.9271, + "step": 6516 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6517 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 1.0335, + "step": 6518 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001999944269512637, + "loss": 2.5406, + "step": 6519 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 6.2699, + "step": 6520 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6521 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 7.0255, + "step": 6522 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6523 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6524 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.061, + "step": 6525 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6526 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6527 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 8.3671, + "step": 6528 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 4.1017, + "step": 6529 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.9419, + "step": 6530 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.293, + "step": 6531 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.3177, + "step": 6532 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6533 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 5.1895, + "step": 6534 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 9.7647, + "step": 6535 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.7116, + "step": 6536 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 5.8372, + "step": 6537 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 7.0711, + "step": 6538 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.4676, + "step": 6539 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6540 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.6619, + "step": 6541 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6542 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6543 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6544 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.8911, + "step": 6545 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.923, + "step": 6546 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6547 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6548 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 7.1938, + "step": 6549 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6550 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6551 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6552 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.1036, + "step": 6553 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6554 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.498, + "step": 6555 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6556 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.8685, + "step": 6557 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6558 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.1798, + "step": 6559 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.0329, + "step": 6560 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6561 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6562 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.3548, + "step": 6563 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.0391, + "step": 6564 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6565 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6566 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 5.9445, + "step": 6567 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6568 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.9594, + "step": 6569 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6570 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 8.1905, + "step": 6571 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.7318, + "step": 6572 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.895, + "step": 6573 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6574 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6575 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 4.0443, + "step": 6576 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.9935, + "step": 6577 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 5.8634, + "step": 6578 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.8498, + "step": 6579 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6580 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 7.8781, + "step": 6581 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.6113, + "step": 6582 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.2096, + "step": 6583 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 8.2863, + "step": 6584 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.6869, + "step": 6585 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6586 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.8434, + "step": 6587 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 5.5774, + "step": 6588 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 5.3559, + "step": 6589 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.0072, + "step": 6590 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6591 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.9653, + "step": 6592 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.8922, + "step": 6593 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6594 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6595 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6596 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6597 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6598 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6599 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6600 + }, + { + "epoch": 0.55, + "eval_loss": NaN, + "eval_runtime": 40.2212, + "eval_samples_per_second": 62.156, + "eval_steps_per_second": 7.782, + "step": 6600 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 9.8312, + "step": 6601 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.082, + "step": 6602 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.8182, + "step": 6603 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6604 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.3962, + "step": 6605 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6606 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6607 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6608 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.3352, + "step": 6609 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 4.5328, + "step": 6610 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.0534, + "step": 6611 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6612 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6613 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6614 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6615 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.7531, + "step": 6616 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.7832, + "step": 6617 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6618 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6619 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.6576, + "step": 6620 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6621 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6622 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 7.7651, + "step": 6623 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.9364, + "step": 6624 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 2.5629, + "step": 6625 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 8.2845, + "step": 6626 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 7.861, + "step": 6627 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 5.8766, + "step": 6628 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.9344, + "step": 6629 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 9.6332, + "step": 6630 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 3.9874, + "step": 6631 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 8.5165, + "step": 6632 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 4.667, + "step": 6633 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.9073, + "step": 6634 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6635 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6636 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6637 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 1.4534, + "step": 6638 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6639 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.5044, + "step": 6640 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 5.0913, + "step": 6641 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 2.3734, + "step": 6642 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6643 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 2.96, + "step": 6644 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6645 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6646 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 8.2084, + "step": 6647 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6648 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.1119, + "step": 6649 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 4.7915, + "step": 6650 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.7416, + "step": 6651 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 2.5784, + "step": 6652 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6653 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6654 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.1761, + "step": 6655 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.1886, + "step": 6656 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.9872, + "step": 6657 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6658 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 7.4252, + "step": 6659 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.934, + "step": 6660 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6661 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6662 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6663 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 5.0301, + "step": 6664 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6665 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6666 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.9801, + "step": 6667 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6668 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 6.3177, + "step": 6669 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6670 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6671 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 6.0847, + "step": 6672 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 5.8098, + "step": 6673 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.4641, + "step": 6674 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6675 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6676 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.9065, + "step": 6677 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.9754, + "step": 6678 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.1361, + "step": 6679 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.9192, + "step": 6680 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.2519, + "step": 6681 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6682 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 6.8148, + "step": 6683 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 2.826, + "step": 6684 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6685 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6686 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6687 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6688 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6689 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6690 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.1153, + "step": 6691 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.8289, + "step": 6692 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6693 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.6785, + "step": 6694 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.962, + "step": 6695 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 2.7577, + "step": 6696 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 7.8732, + "step": 6697 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.8307, + "step": 6698 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 6.781, + "step": 6699 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6700 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6701 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6702 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.3345, + "step": 6703 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6704 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.92, + "step": 6705 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6706 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 4.1794, + "step": 6707 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 2.1031, + "step": 6708 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 10.7871, + "step": 6709 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 6.232, + "step": 6710 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.5745, + "step": 6711 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6712 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 7.3463, + "step": 6713 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6714 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6715 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6716 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 8.0177, + "step": 6717 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 4.0809, + "step": 6718 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6719 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6720 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.6372, + "step": 6721 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6722 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6723 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6724 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.0665, + "step": 6725 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 2.9418, + "step": 6726 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6727 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6728 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 4.6636, + "step": 6729 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.887, + "step": 6730 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.0882, + "step": 6731 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 7.8448, + "step": 6732 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6733 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 7.9232, + "step": 6734 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.512, + "step": 6735 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.5678, + "step": 6736 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.9459, + "step": 6737 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 6.5675, + "step": 6738 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 4.0961, + "step": 6739 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.0395, + "step": 6740 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6741 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6742 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6743 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6744 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6745 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 4.8743, + "step": 6746 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.9793, + "step": 6747 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.2656, + "step": 6748 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6749 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.9053, + "step": 6750 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.9716, + "step": 6751 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 3.0336, + "step": 6752 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6753 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.9312, + "step": 6754 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 7.6641, + "step": 6755 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 8.2109, + "step": 6756 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6757 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6758 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001999944269512637, + "loss": 1.4984, + "step": 6759 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6760 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6761 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6762 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6763 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6764 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6765 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.9418, + "step": 6766 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6767 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.8108, + "step": 6768 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6769 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 9.1863, + "step": 6770 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.9651, + "step": 6771 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6772 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6773 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 8.955, + "step": 6774 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.433, + "step": 6775 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 2.5902, + "step": 6776 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6777 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6778 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6779 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6780 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6781 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.9308, + "step": 6782 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6783 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.8314, + "step": 6784 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6785 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6786 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 2.3037, + "step": 6787 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 11.2791, + "step": 6788 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6789 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.7869, + "step": 6790 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6791 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6792 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.9482, + "step": 6793 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.0062, + "step": 6794 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 7.6444, + "step": 6795 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6796 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.6232, + "step": 6797 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 5.6158, + "step": 6798 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 7.4402, + "step": 6799 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.9828, + "step": 6800 + }, + { + "epoch": 0.57, + "eval_loss": NaN, + "eval_runtime": 40.3454, + "eval_samples_per_second": 61.965, + "eval_steps_per_second": 7.758, + "step": 6800 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6801 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.8824, + "step": 6802 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.4606, + "step": 6803 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6804 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 9.1473, + "step": 6805 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 8.3446, + "step": 6806 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.949, + "step": 6807 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6808 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6809 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6810 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 8.2061, + "step": 6811 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6812 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 6.0555, + "step": 6813 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 8.8896, + "step": 6814 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6815 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6816 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6817 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 5.1675, + "step": 6818 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6819 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 5.3297, + "step": 6820 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6821 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 6.1226, + "step": 6822 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.0675, + "step": 6823 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 5.6324, + "step": 6824 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 8.1058, + "step": 6825 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 2.8265, + "step": 6826 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.236, + "step": 6827 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6828 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.0979, + "step": 6829 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.7686, + "step": 6830 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.7579, + "step": 6831 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 5.3124, + "step": 6832 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6833 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6834 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.0256, + "step": 6835 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6836 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6837 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 6.3096, + "step": 6838 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6839 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 2.0282, + "step": 6840 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6841 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.0126, + "step": 6842 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6843 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6844 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.904, + "step": 6845 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6846 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 4.7088, + "step": 6847 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6848 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6849 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.6257, + "step": 6850 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6851 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 9.013, + "step": 6852 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 9.9323, + "step": 6853 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 2.3209, + "step": 6854 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6855 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 8.4209, + "step": 6856 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6857 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.5885, + "step": 6858 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6859 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 4.4318, + "step": 6860 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6861 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.9783, + "step": 6862 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 3.8258, + "step": 6863 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6864 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.622, + "step": 6865 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 4.2996, + "step": 6866 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 8.2706, + "step": 6867 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.0851, + "step": 6868 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6869 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 7.4337, + "step": 6870 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 4.4725, + "step": 6871 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6872 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6873 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6874 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 6.415, + "step": 6875 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 1.8702, + "step": 6876 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 7.3629, + "step": 6877 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6878 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6879 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.1526, + "step": 6880 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.6845, + "step": 6881 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.7682, + "step": 6882 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.3484, + "step": 6883 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6884 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.7963, + "step": 6885 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6886 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6887 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 5.1921, + "step": 6888 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6889 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 4.4045, + "step": 6890 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.2785, + "step": 6891 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 4.3722, + "step": 6892 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6893 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6894 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.1553, + "step": 6895 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6896 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6897 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6898 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.2876, + "step": 6899 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.3261, + "step": 6900 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6901 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6902 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.154, + "step": 6903 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 8.615, + "step": 6904 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.8835, + "step": 6905 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.9117, + "step": 6906 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6907 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6908 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.974, + "step": 6909 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.7649, + "step": 6910 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6911 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6912 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6913 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 11.2341, + "step": 6914 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6915 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6916 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.6562, + "step": 6917 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6918 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.9612, + "step": 6919 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 8.2817, + "step": 6920 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.802, + "step": 6921 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6922 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6923 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6924 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.971, + "step": 6925 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.1356, + "step": 6926 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 7.6905, + "step": 6927 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 9.1952, + "step": 6928 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.1212, + "step": 6929 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6930 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.9452, + "step": 6931 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.9104, + "step": 6932 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6933 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.9161, + "step": 6934 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.3787, + "step": 6935 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 8.3303, + "step": 6936 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.7039, + "step": 6937 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6938 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.8318, + "step": 6939 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.8711, + "step": 6940 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 5.3046, + "step": 6941 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.0652, + "step": 6942 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 4.261, + "step": 6943 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.3348, + "step": 6944 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.8038, + "step": 6945 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 8.2725, + "step": 6946 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6947 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6948 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6949 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 9.8043, + "step": 6950 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.6556, + "step": 6951 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6952 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6953 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6954 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.9513, + "step": 6955 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 7.4658, + "step": 6956 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.9001, + "step": 6957 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.4978, + "step": 6958 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6959 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 4.6191, + "step": 6960 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6961 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 9.0516, + "step": 6962 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6963 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6964 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 8.6473, + "step": 6965 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.9341, + "step": 6966 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.8693, + "step": 6967 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6968 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.707, + "step": 6969 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 3.0951, + "step": 6970 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.6906, + "step": 6971 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6972 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 2.166, + "step": 6973 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6974 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6975 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 5.1711, + "step": 6976 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 5.9629, + "step": 6977 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6978 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 4.8771, + "step": 6979 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6980 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6981 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 6.7308, + "step": 6982 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.88, + "step": 6983 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6984 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6985 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.8925, + "step": 6986 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.9787, + "step": 6987 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6988 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.0888, + "step": 6989 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6990 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6991 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6992 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6993 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 1.0462, + "step": 6994 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6995 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6996 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6997 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6998 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 6999 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7000 + }, + { + "epoch": 0.59, + "eval_loss": NaN, + "eval_runtime": 40.6103, + "eval_samples_per_second": 61.561, + "eval_steps_per_second": 7.707, + "step": 7000 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7001 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7002 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7003 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.9459, + "step": 7004 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7005 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7006 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7007 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 3.3966, + "step": 7008 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.6434, + "step": 7009 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 3.8483, + "step": 7010 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7011 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.0551, + "step": 7012 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.2387, + "step": 7013 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.2569, + "step": 7014 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7015 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 9.9032, + "step": 7016 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 8.5105, + "step": 7017 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.8856, + "step": 7018 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7019 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7020 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.4752, + "step": 7021 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 8.0929, + "step": 7022 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.2709, + "step": 7023 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.2048, + "step": 7024 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 8.005, + "step": 7025 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.9573, + "step": 7026 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7027 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7028 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7029 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7030 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.2519, + "step": 7031 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7032 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7033 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 5.1209, + "step": 7034 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 5.8264, + "step": 7035 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.4953, + "step": 7036 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.1954, + "step": 7037 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7038 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.5816, + "step": 7039 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.8014, + "step": 7040 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7041 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.0533, + "step": 7042 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.5113, + "step": 7043 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.971, + "step": 7044 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7045 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 6.1803, + "step": 7046 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 7.0187, + "step": 7047 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.4181, + "step": 7048 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7049 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7050 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.2484, + "step": 7051 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7052 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7053 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.5452, + "step": 7054 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7055 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7056 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7057 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.6399, + "step": 7058 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7059 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.7783, + "step": 7060 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.1414, + "step": 7061 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7062 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 5.4754, + "step": 7063 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7064 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7065 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.3288, + "step": 7066 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7067 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7068 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 7.7133, + "step": 7069 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7070 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 5.9819, + "step": 7071 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.4245, + "step": 7072 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7073 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7074 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 8.6498, + "step": 7075 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7076 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.1012, + "step": 7077 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7078 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7079 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 3.4708, + "step": 7080 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.8249, + "step": 7081 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 3.3357, + "step": 7082 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.7808, + "step": 7083 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.2273, + "step": 7084 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7085 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.4067, + "step": 7086 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.4538, + "step": 7087 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 6.3644, + "step": 7088 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.8481, + "step": 7089 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.2442, + "step": 7090 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 2.2784, + "step": 7091 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7092 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.6422, + "step": 7093 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 1.07, + "step": 7094 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7095 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.8961, + "step": 7096 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7097 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7098 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7099 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 3.7516, + "step": 7100 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7101 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7102 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 5.8589, + "step": 7103 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7104 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7105 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 3.4988, + "step": 7106 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7107 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7108 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7109 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7110 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.0846, + "step": 7111 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 5.6617, + "step": 7112 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7113 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7114 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7115 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 4.4608, + "step": 7116 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7117 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.0523, + "step": 7118 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.3647, + "step": 7119 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7120 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7121 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 3.7001, + "step": 7122 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7123 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7124 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7125 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7126 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7127 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7128 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7129 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.3924, + "step": 7130 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7131 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 3.9279, + "step": 7132 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.9478, + "step": 7133 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7134 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 6.9997, + "step": 7135 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 8.2957, + "step": 7136 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.3344, + "step": 7137 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7138 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7139 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7140 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.9242, + "step": 7141 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.0579, + "step": 7142 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7143 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.1909, + "step": 7144 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7145 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.9421, + "step": 7146 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.9457, + "step": 7147 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7148 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.8197, + "step": 7149 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.3523, + "step": 7150 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 3.0915, + "step": 7151 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7152 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 6.9244, + "step": 7153 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.7448, + "step": 7154 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7155 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7156 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7157 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.3974, + "step": 7158 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7159 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7160 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 8.2296, + "step": 7161 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7162 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7163 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7164 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.8661, + "step": 7165 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.9033, + "step": 7166 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7167 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7168 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.9836, + "step": 7169 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7170 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7171 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7172 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7173 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.0283, + "step": 7174 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.7694, + "step": 7175 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.6522, + "step": 7176 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7177 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7178 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 7.9943, + "step": 7179 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.1212, + "step": 7180 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 7.9895, + "step": 7181 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.8299, + "step": 7182 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7183 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 7.7716, + "step": 7184 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.8944, + "step": 7185 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.7098, + "step": 7186 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.44, + "step": 7187 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7188 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7189 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.0958, + "step": 7190 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7191 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7192 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7193 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7194 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7195 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 3.7841, + "step": 7196 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.0107, + "step": 7197 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7198 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 9.3276, + "step": 7199 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.2732, + "step": 7200 + }, + { + "epoch": 0.6, + "eval_loss": NaN, + "eval_runtime": 40.3952, + "eval_samples_per_second": 61.889, + "eval_steps_per_second": 7.748, + "step": 7200 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.0231, + "step": 7201 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7202 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 7.4976, + "step": 7203 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7204 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.8066, + "step": 7205 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7206 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7207 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7208 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 9.0212, + "step": 7209 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.9889, + "step": 7210 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.0026, + "step": 7211 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.7665, + "step": 7212 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7213 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7214 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7215 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7216 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7217 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 5.094, + "step": 7218 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7219 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7220 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7221 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7222 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 7.288, + "step": 7223 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7224 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7225 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 1.0888, + "step": 7226 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7227 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.2944, + "step": 7228 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7229 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 8.0355, + "step": 7230 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7231 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7232 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 4.3898, + "step": 7233 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.9874, + "step": 7234 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 2.2918, + "step": 7235 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7236 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7237 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 3.6708, + "step": 7238 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7239 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 5.3972, + "step": 7240 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7241 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7242 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7243 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7244 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7245 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.8621, + "step": 7246 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.8103, + "step": 7247 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7248 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7249 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7250 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7251 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.0294, + "step": 7252 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7253 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7254 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 7.7938, + "step": 7255 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7256 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7257 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 4.123, + "step": 7258 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7259 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.8941, + "step": 7260 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 7.6284, + "step": 7261 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.9406, + "step": 7262 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7263 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.9069, + "step": 7264 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.5712, + "step": 7265 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.7585, + "step": 7266 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7267 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.2163, + "step": 7268 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7269 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 3.8449, + "step": 7270 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7271 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7272 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7273 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7274 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7275 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7276 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.9492, + "step": 7277 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7278 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7279 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7280 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 7.901, + "step": 7281 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7282 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 9.0704, + "step": 7283 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7284 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7285 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.8007, + "step": 7286 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 4.2354, + "step": 7287 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7288 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7289 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 8.1035, + "step": 7290 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7291 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7292 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7293 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.0214, + "step": 7294 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.8374, + "step": 7295 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 4.2072, + "step": 7296 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7297 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7298 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7299 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 4.5, + "step": 7300 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.0393, + "step": 7301 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7302 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 7.0914, + "step": 7303 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 4.1095, + "step": 7304 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.0317, + "step": 7305 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 7.9101, + "step": 7306 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7307 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7308 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7309 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 5.9263, + "step": 7310 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 5.3797, + "step": 7311 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7312 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7313 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7314 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.073, + "step": 7315 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.0803, + "step": 7316 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 4.1657, + "step": 7317 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7318 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 6.1418, + "step": 7319 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 3.7337, + "step": 7320 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7321 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.9463, + "step": 7322 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7323 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7324 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 7.6804, + "step": 7325 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7326 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7327 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7328 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7329 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7330 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7331 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7332 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7333 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7334 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7335 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.848, + "step": 7336 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.968, + "step": 7337 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7338 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7339 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7340 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7341 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 7.1594, + "step": 7342 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.088, + "step": 7343 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 3.4839, + "step": 7344 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7345 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 6.6549, + "step": 7346 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.2006, + "step": 7347 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 4.9816, + "step": 7348 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 3.2691, + "step": 7349 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7350 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7351 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 1.1499, + "step": 7352 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 3.8166, + "step": 7353 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7354 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7355 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 9.7153, + "step": 7356 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001999944269512637, + "loss": 2.9832, + "step": 7357 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7358 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7359 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 2.7038, + "step": 7360 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 7.8313, + "step": 7361 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7362 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 2.0695, + "step": 7363 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.9516, + "step": 7364 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.6292, + "step": 7365 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.753, + "step": 7366 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 8.7587, + "step": 7367 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7368 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.6341, + "step": 7369 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7370 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.7667, + "step": 7371 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7372 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7373 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7374 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 8.0847, + "step": 7375 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7376 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7377 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.8523, + "step": 7378 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.0296, + "step": 7379 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 2.409, + "step": 7380 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7381 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 5.3572, + "step": 7382 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 8.1001, + "step": 7383 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 8.263, + "step": 7384 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.9288, + "step": 7385 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.5597, + "step": 7386 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.5519, + "step": 7387 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7388 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7389 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7390 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.7912, + "step": 7391 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.039, + "step": 7392 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7393 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 7.0966, + "step": 7394 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.8505, + "step": 7395 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.9829, + "step": 7396 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 5.5439, + "step": 7397 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7398 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7399 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7400 + }, + { + "epoch": 0.62, + "eval_loss": NaN, + "eval_runtime": 40.0879, + "eval_samples_per_second": 62.363, + "eval_steps_per_second": 7.808, + "step": 7400 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7401 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7402 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 5.733, + "step": 7403 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.9661, + "step": 7404 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.9629, + "step": 7405 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7406 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 2.15, + "step": 7407 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 5.925, + "step": 7408 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7409 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7410 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7411 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.9941, + "step": 7412 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.8848, + "step": 7413 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7414 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7415 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.923, + "step": 7416 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7417 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 5.229, + "step": 7418 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7419 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 2.0986, + "step": 7420 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7421 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7422 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7423 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.1323, + "step": 7424 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 6.5416, + "step": 7425 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 5.2438, + "step": 7426 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.0845, + "step": 7427 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.0142, + "step": 7428 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7429 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.2666, + "step": 7430 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.7395, + "step": 7431 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.7526, + "step": 7432 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 6.4531, + "step": 7433 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7434 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7435 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 8.0835, + "step": 7436 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7437 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7438 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7439 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7440 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.1358, + "step": 7441 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.5326, + "step": 7442 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7443 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7444 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7445 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7446 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7447 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7448 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.8702, + "step": 7449 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7450 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.0291, + "step": 7451 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7452 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 2.1925, + "step": 7453 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 2.1417, + "step": 7454 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.7783, + "step": 7455 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.872, + "step": 7456 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7457 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.9111, + "step": 7458 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7459 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 7.2296, + "step": 7460 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 3.932, + "step": 7461 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7462 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 7.984, + "step": 7463 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7464 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 7.4892, + "step": 7465 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 1.1897, + "step": 7466 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7467 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7468 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7469 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7470 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7471 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 4.7096, + "step": 7472 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7473 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7474 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7475 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001999944269512637, + "loss": 0.316, + "step": 7476 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.4536, + "step": 7477 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 7.9345, + "step": 7478 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7479 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7480 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7481 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7482 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7483 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 8.1178, + "step": 7484 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7485 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 6.4615, + "step": 7486 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7487 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.1663, + "step": 7488 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7489 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7490 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7491 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.5355, + "step": 7492 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7493 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 6.7279, + "step": 7494 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7495 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.943, + "step": 7496 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 5.4319, + "step": 7497 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 7.2735, + "step": 7498 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.8078, + "step": 7499 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 6.1512, + "step": 7500 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.55, + "step": 7501 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7502 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 5.5291, + "step": 7503 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 8.3938, + "step": 7504 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7505 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 8.3537, + "step": 7506 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 6.6978, + "step": 7507 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.5333, + "step": 7508 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7509 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 4.5907, + "step": 7510 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.445, + "step": 7511 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7512 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.1993, + "step": 7513 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.4222, + "step": 7514 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7515 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7516 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.4082, + "step": 7517 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.2551, + "step": 7518 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7519 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7520 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.7031, + "step": 7521 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7522 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7523 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7524 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 5.081, + "step": 7525 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 8.2495, + "step": 7526 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.2507, + "step": 7527 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.1377, + "step": 7528 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7529 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7530 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.0232, + "step": 7531 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.9067, + "step": 7532 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.0115, + "step": 7533 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7534 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7535 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7536 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7537 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 6.0602, + "step": 7538 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.8987, + "step": 7539 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 9.9945, + "step": 7540 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7541 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.7422, + "step": 7542 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7543 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7544 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.7117, + "step": 7545 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.5595, + "step": 7546 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7547 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 2.1409, + "step": 7548 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7549 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7550 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 8.4411, + "step": 7551 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7552 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.9155, + "step": 7553 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7554 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7555 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7556 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 5.2601, + "step": 7557 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 7.4088, + "step": 7558 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7559 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.715, + "step": 7560 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7561 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7562 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.8845, + "step": 7563 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 7.8226, + "step": 7564 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.702, + "step": 7565 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.027, + "step": 7566 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 6.8814, + "step": 7567 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7568 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7569 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 7.9735, + "step": 7570 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.9361, + "step": 7571 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.6863, + "step": 7572 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 8.3367, + "step": 7573 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.1204, + "step": 7574 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7575 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7576 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7577 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7578 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.691, + "step": 7579 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 6.5838, + "step": 7580 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7581 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7582 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7583 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 4.9692, + "step": 7584 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7585 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.3874, + "step": 7586 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7587 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 5.0479, + "step": 7588 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.9527, + "step": 7589 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 7.4417, + "step": 7590 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 1.983, + "step": 7591 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 3.6913, + "step": 7592 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.9702, + "step": 7593 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7594 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7595 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7596 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 9.6698, + "step": 7597 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.6277, + "step": 7598 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.0453, + "step": 7599 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7600 + }, + { + "epoch": 0.64, + "eval_loss": NaN, + "eval_runtime": 40.8278, + "eval_samples_per_second": 61.233, + "eval_steps_per_second": 7.666, + "step": 7600 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7601 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 9.8089, + "step": 7602 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 5.0238, + "step": 7603 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 5.2407, + "step": 7604 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 9.4224, + "step": 7605 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.8643, + "step": 7606 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.8954, + "step": 7607 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7608 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7609 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.8258, + "step": 7610 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.954, + "step": 7611 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7612 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7613 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7614 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7615 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.946, + "step": 7616 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.9664, + "step": 7617 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.9619, + "step": 7618 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7619 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 5.4319, + "step": 7620 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7621 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7622 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 9.1682, + "step": 7623 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.9245, + "step": 7624 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.2349, + "step": 7625 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7626 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 7.017, + "step": 7627 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.0005, + "step": 7628 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7629 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7630 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7631 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.3448, + "step": 7632 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7633 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.6138, + "step": 7634 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.903, + "step": 7635 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.604, + "step": 7636 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7637 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7638 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.9632, + "step": 7639 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 3.4983, + "step": 7640 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7641 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7642 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7643 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 9.3964, + "step": 7644 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7645 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.1133, + "step": 7646 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7647 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.9423, + "step": 7648 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.7942, + "step": 7649 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7650 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7651 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 3.4724, + "step": 7652 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7653 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7654 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.4165, + "step": 7655 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7656 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7657 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.3289, + "step": 7658 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 5.0998, + "step": 7659 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.2122, + "step": 7660 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.5552, + "step": 7661 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 7.9957, + "step": 7662 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7663 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.8245, + "step": 7664 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 6.8837, + "step": 7665 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.0407, + "step": 7666 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7667 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 8.1325, + "step": 7668 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.9429, + "step": 7669 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 3.6865, + "step": 7670 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7671 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.7083, + "step": 7672 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.6672, + "step": 7673 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.9743, + "step": 7674 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7675 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7676 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.9948, + "step": 7677 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 6.5671, + "step": 7678 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7679 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7680 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7681 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.5825, + "step": 7682 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7683 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.6134, + "step": 7684 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 5.073, + "step": 7685 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 7.5332, + "step": 7686 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.0, + "step": 7687 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7688 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.9431, + "step": 7689 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.6866, + "step": 7690 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7691 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 7.0827, + "step": 7692 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 6.2324, + "step": 7693 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.0442, + "step": 7694 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 1.7739, + "step": 7695 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 5.9355, + "step": 7696 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 3.6652, + "step": 7697 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7698 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7699 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 3.4083, + "step": 7700 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7701 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.1059, + "step": 7702 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 2.2558, + "step": 7703 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7704 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 5.3369, + "step": 7705 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7706 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7707 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.9489, + "step": 7708 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7709 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7710 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7711 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7712 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7713 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.5913, + "step": 7714 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 4.6516, + "step": 7715 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7716 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7717 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7718 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7719 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7720 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7721 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.3584, + "step": 7722 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.0231, + "step": 7723 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.0207, + "step": 7724 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 5.8526, + "step": 7725 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7726 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 4.5326, + "step": 7727 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 5.0403, + "step": 7728 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 3.4565, + "step": 7729 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7730 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7731 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 6.1202, + "step": 7732 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 5.9911, + "step": 7733 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 3.3539, + "step": 7734 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7735 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 12.1176, + "step": 7736 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7737 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7738 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 10.4028, + "step": 7739 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 10.7371, + "step": 7740 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7741 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7742 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.7443, + "step": 7743 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.9602, + "step": 7744 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7745 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7746 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 6.2142, + "step": 7747 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.13, + "step": 7748 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7749 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 9.8563, + "step": 7750 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.3232, + "step": 7751 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7752 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7753 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.9271, + "step": 7754 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7755 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 3.6152, + "step": 7756 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7757 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 3.3368, + "step": 7758 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7759 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 4.0767, + "step": 7760 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 6.6587, + "step": 7761 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 8.4026, + "step": 7762 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7763 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 7.6717, + "step": 7764 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7765 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7766 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 4.502, + "step": 7767 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.9909, + "step": 7768 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7769 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7770 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.8726, + "step": 7771 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.9661, + "step": 7772 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.0313, + "step": 7773 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.0671, + "step": 7774 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 8.4354, + "step": 7775 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7776 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 6.8777, + "step": 7777 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.7395, + "step": 7778 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 7.1583, + "step": 7779 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7780 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7781 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7782 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 4.0024, + "step": 7783 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.0488, + "step": 7784 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7785 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7786 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 8.5188, + "step": 7787 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.457, + "step": 7788 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7789 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7790 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.713, + "step": 7791 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.965, + "step": 7792 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.166, + "step": 7793 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.2435, + "step": 7794 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7795 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 5.235, + "step": 7796 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7797 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7798 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 8.3394, + "step": 7799 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 11.2587, + "step": 7800 + }, + { + "epoch": 0.65, + "eval_loss": NaN, + "eval_runtime": 40.8274, + "eval_samples_per_second": 61.233, + "eval_steps_per_second": 7.666, + "step": 7800 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 7.5332, + "step": 7801 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 3.9411, + "step": 7802 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7803 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7804 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7805 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7806 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7807 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7808 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7809 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7810 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.2321, + "step": 7811 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 3.8362, + "step": 7812 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.1372, + "step": 7813 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.4488, + "step": 7814 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7815 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7816 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7817 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 8.1787, + "step": 7818 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 5.8438, + "step": 7819 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 4.4749, + "step": 7820 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.2322, + "step": 7821 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 7.3902, + "step": 7822 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 9.5121, + "step": 7823 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.034, + "step": 7824 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.3408, + "step": 7825 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 9.7566, + "step": 7826 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 3.7206, + "step": 7827 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7828 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7829 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 2.2313, + "step": 7830 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7831 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 1.2212, + "step": 7832 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.9859, + "step": 7833 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.5961, + "step": 7834 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7835 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7836 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.0702, + "step": 7837 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 8.4124, + "step": 7838 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7839 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7840 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7841 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7842 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 10.0446, + "step": 7843 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7844 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 5.2905, + "step": 7845 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 6.2738, + "step": 7846 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7847 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7848 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 6.6464, + "step": 7849 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 10.835, + "step": 7850 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7851 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 2.615, + "step": 7852 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.7452, + "step": 7853 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7854 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 4.5972, + "step": 7855 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.1928, + "step": 7856 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7857 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7858 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 4.4056, + "step": 7859 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7860 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7861 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7862 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7863 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 4.7912, + "step": 7864 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7865 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7866 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 4.2644, + "step": 7867 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7868 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 2.8933, + "step": 7869 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.0, + "step": 7870 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7871 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7872 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 2.2883, + "step": 7873 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 2.1485, + "step": 7874 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 3.5057, + "step": 7875 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.9903, + "step": 7876 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7877 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 4.709, + "step": 7878 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 9.291, + "step": 7879 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7880 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.8453, + "step": 7881 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7882 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7883 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 2.814, + "step": 7884 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 2.9275, + "step": 7885 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7886 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 9.6967, + "step": 7887 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 10.3352, + "step": 7888 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7889 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 2.1045, + "step": 7890 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7891 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.1611, + "step": 7892 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 1.9655, + "step": 7893 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7894 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999944269512637, + "loss": 0.0, + "step": 7895 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 4.0098, + "step": 7896 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.5549, + "step": 7897 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 1.0549, + "step": 7898 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7899 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7900 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 2.2882, + "step": 7901 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7902 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7903 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 4.2481, + "step": 7904 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 7.9814, + "step": 7905 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7906 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 2.1449, + "step": 7907 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7908 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 5.5099, + "step": 7909 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 8.2479, + "step": 7910 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7911 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 3.9542, + "step": 7912 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7913 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7914 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 4.1167, + "step": 7915 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 2.0454, + "step": 7916 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.9509, + "step": 7917 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7918 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 3.9595, + "step": 7919 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 8.473, + "step": 7920 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7921 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.5866, + "step": 7922 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7923 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7924 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 2.1212, + "step": 7925 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7926 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 2.0379, + "step": 7927 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7928 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7929 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 8.4408, + "step": 7930 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 2.4201, + "step": 7931 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 3.9785, + "step": 7932 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 3.6826, + "step": 7933 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.853, + "step": 7934 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7935 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7936 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 1.8637, + "step": 7937 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 1.9053, + "step": 7938 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7939 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 7.5224, + "step": 7940 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7941 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7942 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7943 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 4.0953, + "step": 7944 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 6.7773, + "step": 7945 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7946 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7947 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 6.0911, + "step": 7948 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 9.2276, + "step": 7949 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 6.2108, + "step": 7950 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7951 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7952 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7953 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 3.9344, + "step": 7954 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001999888539025274, + "loss": 1.9398, + "step": 7955 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 6.4509, + "step": 7956 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 7.9892, + "step": 7957 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 3.3011, + "step": 7958 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 3.9681, + "step": 7959 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 5.1521, + "step": 7960 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.8615, + "step": 7961 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7962 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7963 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7964 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 4.6479, + "step": 7965 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.0442, + "step": 7966 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.0812, + "step": 7967 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 5.7476, + "step": 7968 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 6.9381, + "step": 7969 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.0918, + "step": 7970 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 7.4806, + "step": 7971 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 6.8877, + "step": 7972 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.8676, + "step": 7973 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 3.0088, + "step": 7974 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 3.4673, + "step": 7975 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7976 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 3.7944, + "step": 7977 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 8.3414, + "step": 7978 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7979 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7980 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7981 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 7.6127, + "step": 7982 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7983 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 6.3178, + "step": 7984 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7985 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 6.6582, + "step": 7986 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 5.3428, + "step": 7987 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7988 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 8.1267, + "step": 7989 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.2459, + "step": 7990 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.9579, + "step": 7991 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7992 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.9821, + "step": 7993 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 8.8981, + "step": 7994 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7995 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7996 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7997 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 7998 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 7.9213, + "step": 7999 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.8336, + "step": 8000 + }, + { + "epoch": 0.67, + "eval_loss": NaN, + "eval_runtime": 41.2837, + "eval_samples_per_second": 60.557, + "eval_steps_per_second": 7.582, + "step": 8000 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 9.828, + "step": 8001 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8002 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 7.843, + "step": 8003 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8004 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8005 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8006 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8007 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8008 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.8736, + "step": 8009 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.9311, + "step": 8010 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8011 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 8.2509, + "step": 8012 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 4.3029, + "step": 8013 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 4.3327, + "step": 8014 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.7537, + "step": 8015 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 5.6528, + "step": 8016 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8017 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8018 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8019 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.848, + "step": 8020 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 7.3242, + "step": 8021 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 2.0236, + "step": 8022 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 6.0533, + "step": 8023 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8024 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8025 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 1.9141, + "step": 8026 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8027 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.9126, + "step": 8028 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8029 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.8359, + "step": 8030 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.9141, + "step": 8031 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8032 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 0.0, + "step": 8033 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 5.6758, + "step": 8034 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001999888539025274, + "loss": 3.3508, + "step": 8035 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 3.5883, + "step": 8036 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 1.3429, + "step": 8037 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 8.0929, + "step": 8038 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8039 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 2.9264, + "step": 8040 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8041 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.9775, + "step": 8042 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 4.6115, + "step": 8043 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8044 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 7.3124, + "step": 8045 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8046 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8047 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 4.2106, + "step": 8048 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8049 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 2.3003, + "step": 8050 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 3.7419, + "step": 8051 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8052 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 7.2332, + "step": 8053 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8054 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 2.1636, + "step": 8055 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 1.9912, + "step": 8056 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.7515, + "step": 8057 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 1.8689, + "step": 8058 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8059 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 4.6532, + "step": 8060 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8061 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8062 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8063 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8064 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8065 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 7.475, + "step": 8066 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 4.3283, + "step": 8067 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8068 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8069 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8070 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8071 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.8845, + "step": 8072 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8073 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8074 + }, + { + "epoch": 0.67, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8075 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 13.1839, + "step": 8076 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 7.9938, + "step": 8077 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 8.8485, + "step": 8078 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.2559, + "step": 8079 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 4.6312, + "step": 8080 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8081 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 7.692, + "step": 8082 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8083 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8084 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8085 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8086 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 4.3316, + "step": 8087 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.1276, + "step": 8088 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8089 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.8488, + "step": 8090 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8091 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.786, + "step": 8092 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8093 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8094 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 5.0245, + "step": 8095 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 10.0357, + "step": 8096 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8097 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8098 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.6976, + "step": 8099 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 10.089, + "step": 8100 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.7703, + "step": 8101 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.9527, + "step": 8102 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.9672, + "step": 8103 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8104 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.6319, + "step": 8105 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.5503, + "step": 8106 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8107 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.0356, + "step": 8108 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8109 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8110 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8111 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 8.1593, + "step": 8112 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.9281, + "step": 8113 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8114 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8115 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 5.968, + "step": 8116 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 7.4105, + "step": 8117 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.8954, + "step": 8118 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8119 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.939, + "step": 8120 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.8452, + "step": 8121 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8122 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8123 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8124 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8125 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 5.1238, + "step": 8126 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8127 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 7.2114, + "step": 8128 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.0944, + "step": 8129 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8130 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8131 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.3837, + "step": 8132 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.0371, + "step": 8133 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 10.0842, + "step": 8134 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.5469, + "step": 8135 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.9322, + "step": 8136 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8137 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8138 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.8734, + "step": 8139 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 4.5081, + "step": 8140 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8141 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8142 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8143 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8144 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8145 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8146 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.9733, + "step": 8147 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 4.2327, + "step": 8148 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8149 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 4.1448, + "step": 8150 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8151 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.2564, + "step": 8152 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.9838, + "step": 8153 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 4.7801, + "step": 8154 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8155 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8156 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.9833, + "step": 8157 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.854, + "step": 8158 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 7.6554, + "step": 8159 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8160 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 8.0504, + "step": 8161 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.0138, + "step": 8162 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.6139, + "step": 8163 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8164 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8165 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8166 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8167 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.3786, + "step": 8168 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8169 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8170 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.1002, + "step": 8171 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.0479, + "step": 8172 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.318, + "step": 8173 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8174 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8175 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.8099, + "step": 8176 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 2.0813, + "step": 8177 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8178 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.0597, + "step": 8179 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 7.2624, + "step": 8180 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 7.9266, + "step": 8181 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.4736, + "step": 8182 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8183 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 1.591, + "step": 8184 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 5.7958, + "step": 8185 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8186 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8187 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.747, + "step": 8188 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 3.6518, + "step": 8189 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019998328085379108, + "loss": 0.0, + "step": 8190 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019997770780505475, + "loss": 4.2886, + "step": 8191 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8192 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8193 + }, + { + "epoch": 0.68, + "learning_rate": 0.00019997770780505475, + "loss": 2.4194, + "step": 8194 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8195 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.1102, + "step": 8196 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8197 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8198 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 8.5177, + "step": 8199 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.5997, + "step": 8200 + }, + { + "epoch": 0.69, + "eval_loss": NaN, + "eval_runtime": 40.5665, + "eval_samples_per_second": 61.627, + "eval_steps_per_second": 7.716, + "step": 8200 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.9324, + "step": 8201 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.5848, + "step": 8202 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 5.0798, + "step": 8203 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8204 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8205 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8206 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.0024, + "step": 8207 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8208 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.4532, + "step": 8209 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8210 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 8.3937, + "step": 8211 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.7435, + "step": 8212 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.3961, + "step": 8213 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.3544, + "step": 8214 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8215 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 5.8836, + "step": 8216 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.0676, + "step": 8217 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 8.8499, + "step": 8218 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8219 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 7.6308, + "step": 8220 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 8.1023, + "step": 8221 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.9496, + "step": 8222 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.1767, + "step": 8223 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8224 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.1903, + "step": 8225 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8226 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.0317, + "step": 8227 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 7.386, + "step": 8228 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.5559, + "step": 8229 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8230 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8231 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8232 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 6.6659, + "step": 8233 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8234 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8235 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8236 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8237 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.4302, + "step": 8238 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8239 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.0496, + "step": 8240 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8241 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 7.6912, + "step": 8242 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.338, + "step": 8243 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.0191, + "step": 8244 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 8.5745, + "step": 8245 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.5752, + "step": 8246 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8247 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8248 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8249 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8250 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8251 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.3266, + "step": 8252 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.8929, + "step": 8253 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 7.5208, + "step": 8254 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8255 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 5.9742, + "step": 8256 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.7364, + "step": 8257 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.0502, + "step": 8258 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8259 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8260 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8261 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 8.4172, + "step": 8262 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8263 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.0582, + "step": 8264 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.0187, + "step": 8265 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8266 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 7.7438, + "step": 8267 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8268 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 5.582, + "step": 8269 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8270 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.0288, + "step": 8271 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 9.2029, + "step": 8272 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 9.0033, + "step": 8273 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.2237, + "step": 8274 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.8524, + "step": 8275 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8276 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 5.4782, + "step": 8277 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8278 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8279 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.2072, + "step": 8280 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.037, + "step": 8281 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 5.6067, + "step": 8282 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8283 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.1539, + "step": 8284 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 13.1412, + "step": 8285 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.4067, + "step": 8286 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8287 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.8455, + "step": 8288 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.6288, + "step": 8289 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8290 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.0299, + "step": 8291 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 5.9891, + "step": 8292 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 9.4834, + "step": 8293 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8294 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.9925, + "step": 8295 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 6.4344, + "step": 8296 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.0494, + "step": 8297 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 7.6722, + "step": 8298 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.8333, + "step": 8299 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8300 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 6.3839, + "step": 8301 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.1951, + "step": 8302 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8303 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8304 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 2.3155, + "step": 8305 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8306 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8307 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 3.3335, + "step": 8308 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.3876, + "step": 8309 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 1.7462, + "step": 8310 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 6.9742, + "step": 8311 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 9.9559, + "step": 8312 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8313 + }, + { + "epoch": 0.69, + "learning_rate": 0.00019997770780505475, + "loss": 4.1446, + "step": 8314 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.1023, + "step": 8315 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.7937, + "step": 8316 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8317 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.0361, + "step": 8318 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8319 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.9843, + "step": 8320 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.1303, + "step": 8321 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8322 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8323 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.3702, + "step": 8324 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8325 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 4.9394, + "step": 8326 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 5.3923, + "step": 8327 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8328 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8329 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.9807, + "step": 8330 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.1745, + "step": 8331 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8332 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.0201, + "step": 8333 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 6.9342, + "step": 8334 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.4865, + "step": 8335 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.0143, + "step": 8336 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8337 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.6196, + "step": 8338 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.8706, + "step": 8339 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.134, + "step": 8340 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8341 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.884, + "step": 8342 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8343 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.5666, + "step": 8344 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8345 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.549, + "step": 8346 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8347 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 8.0758, + "step": 8348 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8349 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.9144, + "step": 8350 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.0845, + "step": 8351 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 4.768, + "step": 8352 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 8.0961, + "step": 8353 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.0271, + "step": 8354 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8355 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8356 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8357 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8358 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 4.0127, + "step": 8359 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.8902, + "step": 8360 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 12.2622, + "step": 8361 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.1176, + "step": 8362 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8363 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 3.4942, + "step": 8364 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8365 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8366 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8367 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8368 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8369 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8370 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 3.6945, + "step": 8371 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 8.1529, + "step": 8372 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 3.3711, + "step": 8373 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.1359, + "step": 8374 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 6.6, + "step": 8375 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.1974, + "step": 8376 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8377 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8378 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8379 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8380 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.3081, + "step": 8381 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 3.0091, + "step": 8382 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8383 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8384 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8385 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8386 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.7526, + "step": 8387 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8388 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8389 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8390 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.9511, + "step": 8391 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 5.2206, + "step": 8392 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 9.1032, + "step": 8393 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.3061, + "step": 8394 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8395 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 9.3628, + "step": 8396 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.3443, + "step": 8397 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8398 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 8.5251, + "step": 8399 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8400 + }, + { + "epoch": 0.7, + "eval_loss": NaN, + "eval_runtime": 40.5652, + "eval_samples_per_second": 61.629, + "eval_steps_per_second": 7.716, + "step": 8400 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8401 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8402 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8403 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 3.1978, + "step": 8404 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 4.3772, + "step": 8405 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8406 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.272, + "step": 8407 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8408 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 4.091, + "step": 8409 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8410 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.8282, + "step": 8411 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8412 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.0244, + "step": 8413 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.941, + "step": 8414 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.0357, + "step": 8415 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8416 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.8492, + "step": 8417 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8418 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8419 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8420 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.8675, + "step": 8421 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8422 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 6.9829, + "step": 8423 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 3.5101, + "step": 8424 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8425 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8426 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 1.9658, + "step": 8427 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 7.6521, + "step": 8428 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 8.7839, + "step": 8429 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8430 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 2.3756, + "step": 8431 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 3.7259, + "step": 8432 + }, + { + "epoch": 0.7, + "learning_rate": 0.00019997770780505475, + "loss": 4.9941, + "step": 8433 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8434 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.0174, + "step": 8435 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8436 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 7.5593, + "step": 8437 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8438 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.6243, + "step": 8439 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8440 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.9539, + "step": 8441 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.9193, + "step": 8442 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8443 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8444 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.027, + "step": 8445 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.0867, + "step": 8446 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.8798, + "step": 8447 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.6996, + "step": 8448 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 7.6477, + "step": 8449 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 2.2267, + "step": 8450 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 2.1545, + "step": 8451 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8452 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8453 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8454 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8455 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.6093, + "step": 8456 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.7744, + "step": 8457 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8458 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.5359, + "step": 8459 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 2.6646, + "step": 8460 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8461 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.0212, + "step": 8462 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 10.101, + "step": 8463 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.4221, + "step": 8464 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.1091, + "step": 8465 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8466 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.6269, + "step": 8467 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8468 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 7.718, + "step": 8469 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8470 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.8063, + "step": 8471 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8472 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 8.9192, + "step": 8473 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.8689, + "step": 8474 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.4109, + "step": 8475 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8476 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8477 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8478 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.0032, + "step": 8479 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8480 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8481 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 2.1758, + "step": 8482 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8483 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 2.5931, + "step": 8484 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.7911, + "step": 8485 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.9605, + "step": 8486 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8487 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.0696, + "step": 8488 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8489 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.0562, + "step": 8490 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 2.3948, + "step": 8491 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8492 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8493 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 8.9407, + "step": 8494 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8495 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8496 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.1833, + "step": 8497 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.728, + "step": 8498 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.542, + "step": 8499 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8500 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8501 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.613, + "step": 8502 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8503 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8504 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.7752, + "step": 8505 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8506 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8507 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.9952, + "step": 8508 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 2.1326, + "step": 8509 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.0748, + "step": 8510 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 6.6262, + "step": 8511 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8512 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8513 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.3873, + "step": 8514 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8515 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8516 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8517 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.094, + "step": 8518 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.9179, + "step": 8519 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 7.7244, + "step": 8520 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8521 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.5452, + "step": 8522 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.9477, + "step": 8523 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8524 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8525 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8526 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8527 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8528 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 6.6801, + "step": 8529 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8530 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 15.2559, + "step": 8531 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 11.4072, + "step": 8532 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 8.3047, + "step": 8533 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8534 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8535 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8536 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8537 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 5.1494, + "step": 8538 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8539 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.2328, + "step": 8540 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8541 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8542 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8543 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8544 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.963, + "step": 8545 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 4.9004, + "step": 8546 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 1.0971, + "step": 8547 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 3.7588, + "step": 8548 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8549 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 7.7802, + "step": 8550 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8551 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8552 + }, + { + "epoch": 0.71, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8553 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 3.8864, + "step": 8554 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.4106, + "step": 8555 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 9.8317, + "step": 8556 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 5.3836, + "step": 8557 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8558 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8559 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 7.1101, + "step": 8560 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 3.8591, + "step": 8561 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.6499, + "step": 8562 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 8.8406, + "step": 8563 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.4627, + "step": 8564 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.027, + "step": 8565 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8566 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8567 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.2538, + "step": 8568 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8569 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.5065, + "step": 8570 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.0793, + "step": 8571 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8572 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 8.4631, + "step": 8573 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.121, + "step": 8574 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8575 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.4599, + "step": 8576 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 10.3658, + "step": 8577 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.1224, + "step": 8578 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.7967, + "step": 8579 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.4765, + "step": 8580 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8581 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.7982, + "step": 8582 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.032, + "step": 8583 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8584 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 8.0718, + "step": 8585 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.8474, + "step": 8586 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.308, + "step": 8587 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.8767, + "step": 8588 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 5.9111, + "step": 8589 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.724, + "step": 8590 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.0122, + "step": 8591 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.361, + "step": 8592 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8593 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 7.5707, + "step": 8594 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8595 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 8.2157, + "step": 8596 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.1994, + "step": 8597 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.8795, + "step": 8598 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.3717, + "step": 8599 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.7144, + "step": 8600 + }, + { + "epoch": 0.72, + "eval_loss": NaN, + "eval_runtime": 41.2588, + "eval_samples_per_second": 60.593, + "eval_steps_per_second": 7.586, + "step": 8600 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.7025, + "step": 8601 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.115, + "step": 8602 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8603 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.9271, + "step": 8604 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8605 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8606 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8607 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8608 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8609 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8610 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8611 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.055, + "step": 8612 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 3.21, + "step": 8613 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8614 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8615 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8616 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8617 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 3.9959, + "step": 8618 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 10.8589, + "step": 8619 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8620 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.0095, + "step": 8621 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 9.8971, + "step": 8622 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.4258, + "step": 8623 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.3025, + "step": 8624 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 8.7961, + "step": 8625 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8626 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.9756, + "step": 8627 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.6098, + "step": 8628 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8629 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 5.3176, + "step": 8630 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8631 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.1671, + "step": 8632 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8633 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8634 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 8.3366, + "step": 8635 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8636 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 3.9268, + "step": 8637 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.8402, + "step": 8638 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8639 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 3.8983, + "step": 8640 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8641 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.3954, + "step": 8642 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.0152, + "step": 8643 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.2188, + "step": 8644 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.7137, + "step": 8645 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8646 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8647 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8648 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.0437, + "step": 8649 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.9882, + "step": 8650 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8651 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.5799, + "step": 8652 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.8949, + "step": 8653 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8654 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.9351, + "step": 8655 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8656 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8657 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 8.3061, + "step": 8658 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.9647, + "step": 8659 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 2.499, + "step": 8660 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 4.2057, + "step": 8661 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8662 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8663 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8664 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8665 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8666 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 6.0817, + "step": 8667 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.9089, + "step": 8668 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.0899, + "step": 8669 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 1.1064, + "step": 8670 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8671 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 11.5526, + "step": 8672 + }, + { + "epoch": 0.72, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8673 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8674 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8675 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 7.3686, + "step": 8676 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 6.6814, + "step": 8677 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.4711, + "step": 8678 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 7.1124, + "step": 8679 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8680 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8681 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8682 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8683 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8684 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.7075, + "step": 8685 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8686 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.6351, + "step": 8687 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8688 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8689 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.9729, + "step": 8690 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8691 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8692 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8693 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.8674, + "step": 8694 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.0527, + "step": 8695 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.1483, + "step": 8696 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8697 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8698 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 8.2875, + "step": 8699 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.2232, + "step": 8700 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8701 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8702 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 7.334, + "step": 8703 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8704 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.2943, + "step": 8705 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 5.3016, + "step": 8706 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8707 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.1126, + "step": 8708 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.1049, + "step": 8709 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8710 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.5864, + "step": 8711 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8712 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8713 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.8496, + "step": 8714 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.2574, + "step": 8715 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.9167, + "step": 8716 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.118, + "step": 8717 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 6.903, + "step": 8718 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8719 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8720 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8721 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8722 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8723 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 8.7324, + "step": 8724 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8725 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 8.1971, + "step": 8726 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8727 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8728 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 8.7103, + "step": 8729 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8730 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8731 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8732 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.155, + "step": 8733 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8734 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8735 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8736 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.6515, + "step": 8737 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 7.1672, + "step": 8738 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.1033, + "step": 8739 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.1453, + "step": 8740 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.1846, + "step": 8741 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.1193, + "step": 8742 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 7.4437, + "step": 8743 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.883, + "step": 8744 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.7976, + "step": 8745 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 7.2072, + "step": 8746 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.9668, + "step": 8747 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 9.5571, + "step": 8748 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8749 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8750 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8751 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8752 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8753 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.902, + "step": 8754 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8755 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 9.0328, + "step": 8756 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.1381, + "step": 8757 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8758 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.8656, + "step": 8759 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8760 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.3316, + "step": 8761 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 7.9678, + "step": 8762 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8763 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.2306, + "step": 8764 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8765 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.2998, + "step": 8766 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.5319, + "step": 8767 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8768 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.6186, + "step": 8769 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.5384, + "step": 8770 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8771 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.8614, + "step": 8772 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8773 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.0112, + "step": 8774 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.4809, + "step": 8775 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8776 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.842, + "step": 8777 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8778 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8779 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8780 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.4392, + "step": 8781 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.0269, + "step": 8782 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 4.707, + "step": 8783 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8784 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.9566, + "step": 8785 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.2193, + "step": 8786 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 2.6092, + "step": 8787 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8788 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 3.931, + "step": 8789 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8790 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 1.595, + "step": 8791 + }, + { + "epoch": 0.73, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8792 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.2148, + "step": 8793 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 4.0161, + "step": 8794 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8795 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.8939, + "step": 8796 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 8.8342, + "step": 8797 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.8837, + "step": 8798 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8799 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8800 + }, + { + "epoch": 0.74, + "eval_loss": NaN, + "eval_runtime": 40.4072, + "eval_samples_per_second": 61.87, + "eval_steps_per_second": 7.746, + "step": 8800 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8801 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.5771, + "step": 8802 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8803 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8804 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8805 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.9572, + "step": 8806 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.9845, + "step": 8807 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8808 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8809 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 5.7843, + "step": 8810 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.8048, + "step": 8811 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8812 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8813 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8814 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 6.9517, + "step": 8815 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.912, + "step": 8816 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8817 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.8501, + "step": 8818 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.739, + "step": 8819 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8820 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8821 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8822 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 6.4772, + "step": 8823 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8824 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 2.3804, + "step": 8825 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8826 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8827 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.566, + "step": 8828 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8829 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8830 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8831 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8832 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 6.2833, + "step": 8833 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.666, + "step": 8834 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.8426, + "step": 8835 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8836 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8837 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8838 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8839 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.9774, + "step": 8840 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 5.7247, + "step": 8841 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.3784, + "step": 8842 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8843 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8844 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.1122, + "step": 8845 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8846 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8847 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8848 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 4.3478, + "step": 8849 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 1.0114, + "step": 8850 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8851 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8852 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.359, + "step": 8853 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.8924, + "step": 8854 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 5.4725, + "step": 8855 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 1.2931, + "step": 8856 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.9437, + "step": 8857 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 1.0588, + "step": 8858 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 1.0705, + "step": 8859 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8860 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8861 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8862 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8863 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 8.1119, + "step": 8864 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.2057, + "step": 8865 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8866 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.9507, + "step": 8867 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 4.5028, + "step": 8868 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.2854, + "step": 8869 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8870 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8871 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8872 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8873 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8874 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8875 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 1.8387, + "step": 8876 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.6472, + "step": 8877 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.212, + "step": 8878 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8879 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.8605, + "step": 8880 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 6.4461, + "step": 8881 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8882 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.8067, + "step": 8883 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8884 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.966, + "step": 8885 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 8.5651, + "step": 8886 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 9.5186, + "step": 8887 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8888 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8889 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8890 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8891 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 4.6676, + "step": 8892 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 4.379, + "step": 8893 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.4678, + "step": 8894 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 5.7996, + "step": 8895 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8896 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8897 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8898 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8899 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.4209, + "step": 8900 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.3523, + "step": 8901 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.3193, + "step": 8902 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.6167, + "step": 8903 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8904 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 6.773, + "step": 8905 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 7.731, + "step": 8906 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 2.3582, + "step": 8907 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 1.5863, + "step": 8908 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 9.4027, + "step": 8909 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 3.0977, + "step": 8910 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 1.6022, + "step": 8911 + }, + { + "epoch": 0.74, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8912 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8913 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8914 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 7.7034, + "step": 8915 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 6.7568, + "step": 8916 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8917 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8918 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8919 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.0803, + "step": 8920 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.7232, + "step": 8921 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 9.5633, + "step": 8922 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.1763, + "step": 8923 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8924 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 10.246, + "step": 8925 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8926 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 4.0638, + "step": 8927 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8928 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8929 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8930 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8931 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8932 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8933 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8934 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 2.022, + "step": 8935 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8936 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8937 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8938 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8939 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8940 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.0636, + "step": 8941 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8942 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8943 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.9469, + "step": 8944 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8945 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 8.2682, + "step": 8946 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.2037, + "step": 8947 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8948 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 10.0876, + "step": 8949 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8950 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8951 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8952 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 5.5435, + "step": 8953 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8954 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 2.2947, + "step": 8955 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8956 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8957 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 7.5988, + "step": 8958 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.88, + "step": 8959 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.5475, + "step": 8960 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.9176, + "step": 8961 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.4534, + "step": 8962 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8963 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8964 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 12.9734, + "step": 8965 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8966 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8967 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.064, + "step": 8968 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8969 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 5.8307, + "step": 8970 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.2351, + "step": 8971 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8972 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8973 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 7.2775, + "step": 8974 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.1647, + "step": 8975 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8976 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8977 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8978 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.7847, + "step": 8979 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8980 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 6.3123, + "step": 8981 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.8164, + "step": 8982 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.6154, + "step": 8983 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 5.2655, + "step": 8984 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 4.8989, + "step": 8985 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 9.6707, + "step": 8986 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8987 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8988 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8989 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8990 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.823, + "step": 8991 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.6014, + "step": 8992 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8993 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8994 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.6768, + "step": 8995 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 8.877, + "step": 8996 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 2.2786, + "step": 8997 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 8998 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 4.0536, + "step": 8999 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.1961, + "step": 9000 + }, + { + "epoch": 0.75, + "eval_loss": NaN, + "eval_runtime": 40.5497, + "eval_samples_per_second": 61.653, + "eval_steps_per_second": 7.719, + "step": 9000 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9001 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9002 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.8323, + "step": 9003 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 8.1469, + "step": 9004 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9005 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9006 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9007 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 4.1537, + "step": 9008 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9009 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9010 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.2782, + "step": 9011 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.0465, + "step": 9012 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9013 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9014 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9015 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.8738, + "step": 9016 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9017 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 4.9733, + "step": 9018 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.5191, + "step": 9019 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 6.7051, + "step": 9020 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 7.7747, + "step": 9021 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.4967, + "step": 9022 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9023 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 1.1461, + "step": 9024 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9025 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9026 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9027 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 11.7269, + "step": 9028 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.3906, + "step": 9029 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 7.3945, + "step": 9030 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 3.176, + "step": 9031 + }, + { + "epoch": 0.75, + "learning_rate": 0.00019997770780505475, + "loss": 6.4855, + "step": 9032 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 6.942, + "step": 9033 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 6.4264, + "step": 9034 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9035 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 6.6154, + "step": 9036 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 9.5034, + "step": 9037 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.0958, + "step": 9038 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.0855, + "step": 9039 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.1488, + "step": 9040 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 12.2663, + "step": 9041 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.4787, + "step": 9042 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.131, + "step": 9043 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9044 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.0063, + "step": 9045 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9046 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.0201, + "step": 9047 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 8.2984, + "step": 9048 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 5.9639, + "step": 9049 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.5649, + "step": 9050 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9051 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.1161, + "step": 9052 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.8419, + "step": 9053 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9054 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 11.4706, + "step": 9055 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9056 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.7919, + "step": 9057 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.2991, + "step": 9058 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9059 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9060 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.075, + "step": 9061 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9062 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9063 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 8.7238, + "step": 9064 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 7.9439, + "step": 9065 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 6.7477, + "step": 9066 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.1491, + "step": 9067 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.7647, + "step": 9068 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9069 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9070 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9071 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.0664, + "step": 9072 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.0385, + "step": 9073 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9074 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.1064, + "step": 9075 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.6901, + "step": 9076 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9077 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9078 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.0311, + "step": 9079 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9080 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9081 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.653, + "step": 9082 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 6.199, + "step": 9083 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9084 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.1086, + "step": 9085 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9086 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9087 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9088 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9089 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.7356, + "step": 9090 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.1676, + "step": 9091 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.9802, + "step": 9092 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9093 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9094 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9095 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9096 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.3684, + "step": 9097 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9098 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9099 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.9184, + "step": 9100 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 7.9097, + "step": 9101 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9102 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9103 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9104 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.1639, + "step": 9105 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9106 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9107 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9108 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.6907, + "step": 9109 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 8.8259, + "step": 9110 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9111 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.9028, + "step": 9112 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9113 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9114 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.3997, + "step": 9115 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9116 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9117 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9118 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 7.935, + "step": 9119 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.0663, + "step": 9120 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.8639, + "step": 9121 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9122 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 8.5379, + "step": 9123 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.3261, + "step": 9124 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9125 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.8539, + "step": 9126 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9127 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.9665, + "step": 9128 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9129 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 9.8631, + "step": 9130 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9131 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 4.5775, + "step": 9132 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9133 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.968, + "step": 9134 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.8177, + "step": 9135 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.4058, + "step": 9136 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9137 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9138 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.7154, + "step": 9139 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.1252, + "step": 9140 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.852, + "step": 9141 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9142 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.792, + "step": 9143 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.7841, + "step": 9144 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9145 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 2.5301, + "step": 9146 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.6606, + "step": 9147 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 3.7373, + "step": 9148 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 1.8803, + "step": 9149 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9150 + }, + { + "epoch": 0.76, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9151 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.3351, + "step": 9152 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.656, + "step": 9153 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.9157, + "step": 9154 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.9972, + "step": 9155 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9156 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 5.2358, + "step": 9157 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.6767, + "step": 9158 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.3253, + "step": 9159 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9160 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9161 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9162 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.7918, + "step": 9163 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9164 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.7856, + "step": 9165 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 7.5355, + "step": 9166 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9167 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.998, + "step": 9168 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9169 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9170 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.2753, + "step": 9171 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9172 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9173 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9174 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9175 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9176 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9177 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.7949, + "step": 9178 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9179 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 7.5786, + "step": 9180 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9181 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.0969, + "step": 9182 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.1757, + "step": 9183 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.9543, + "step": 9184 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.8556, + "step": 9185 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9186 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.9675, + "step": 9187 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.6236, + "step": 9188 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9189 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 7.153, + "step": 9190 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9191 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9192 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.0717, + "step": 9193 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.7496, + "step": 9194 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9195 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.084, + "step": 9196 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.0124, + "step": 9197 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.395, + "step": 9198 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9199 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9200 + }, + { + "epoch": 0.77, + "eval_loss": NaN, + "eval_runtime": 40.51, + "eval_samples_per_second": 61.713, + "eval_steps_per_second": 7.726, + "step": 9200 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.754, + "step": 9201 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.9162, + "step": 9202 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.7071, + "step": 9203 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9204 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 8.5443, + "step": 9205 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.989, + "step": 9206 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.7256, + "step": 9207 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9208 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9209 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9210 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.1393, + "step": 9211 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.8934, + "step": 9212 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9213 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.9327, + "step": 9214 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9215 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9216 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.6916, + "step": 9217 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.9202, + "step": 9218 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 5.4939, + "step": 9219 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.1067, + "step": 9220 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.1264, + "step": 9221 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.1312, + "step": 9222 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 13.2689, + "step": 9223 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.8393, + "step": 9224 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 8.2531, + "step": 9225 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9226 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 7.0343, + "step": 9227 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9228 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.949, + "step": 9229 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.9442, + "step": 9230 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 5.9308, + "step": 9231 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9232 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9233 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.2309, + "step": 9234 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9235 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9236 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.1642, + "step": 9237 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9238 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 6.7858, + "step": 9239 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.917, + "step": 9240 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.2151, + "step": 9241 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9242 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9243 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.2129, + "step": 9244 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9245 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.3667, + "step": 9246 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9247 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9248 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.0497, + "step": 9249 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.0777, + "step": 9250 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.222, + "step": 9251 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9252 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.2127, + "step": 9253 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 8.1787, + "step": 9254 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 4.001, + "step": 9255 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.8787, + "step": 9256 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.1337, + "step": 9257 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9258 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 7.3706, + "step": 9259 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 8.4682, + "step": 9260 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 1.8194, + "step": 9261 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 2.1174, + "step": 9262 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.3659, + "step": 9263 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9264 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9265 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.9171, + "step": 9266 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 7.8728, + "step": 9267 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9268 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9269 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 8.1951, + "step": 9270 + }, + { + "epoch": 0.77, + "learning_rate": 0.00019997770780505475, + "loss": 3.4323, + "step": 9271 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9272 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.056, + "step": 9273 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9274 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.8888, + "step": 9275 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 6.6863, + "step": 9276 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9277 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.731, + "step": 9278 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.6247, + "step": 9279 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9280 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.965, + "step": 9281 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.2737, + "step": 9282 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.9759, + "step": 9283 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.911, + "step": 9284 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 6.6458, + "step": 9285 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9286 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9287 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.0061, + "step": 9288 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9289 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.9221, + "step": 9290 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9291 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9292 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.0679, + "step": 9293 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9294 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.9338, + "step": 9295 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.2936, + "step": 9296 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 5.5776, + "step": 9297 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.5534, + "step": 9298 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9299 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9300 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.3015, + "step": 9301 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9302 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9303 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9304 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.8786, + "step": 9305 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.5135, + "step": 9306 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.721, + "step": 9307 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 11.9171, + "step": 9308 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.0136, + "step": 9309 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.7744, + "step": 9310 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9311 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9312 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.5782, + "step": 9313 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9314 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.7341, + "step": 9315 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9316 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9317 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.5737, + "step": 9318 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9319 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9320 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 5.1204, + "step": 9321 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.8104, + "step": 9322 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.6315, + "step": 9323 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9324 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.3129, + "step": 9325 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.4512, + "step": 9326 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9327 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9328 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 10.726, + "step": 9329 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.0178, + "step": 9330 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.68, + "step": 9331 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9332 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9333 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.8251, + "step": 9334 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.5935, + "step": 9335 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9336 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9337 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.5158, + "step": 9338 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9339 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.0888, + "step": 9340 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9341 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.5676, + "step": 9342 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9343 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.8934, + "step": 9344 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9345 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9346 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9347 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.9951, + "step": 9348 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 8.6872, + "step": 9349 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.7839, + "step": 9350 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9351 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.4656, + "step": 9352 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.149, + "step": 9353 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.1767, + "step": 9354 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9355 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9356 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9357 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.7933, + "step": 9358 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.8673, + "step": 9359 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9360 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9361 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9362 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.1929, + "step": 9363 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.9669, + "step": 9364 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9365 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 6.5594, + "step": 9366 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9367 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.6812, + "step": 9368 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.7682, + "step": 9369 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9370 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 4.4563, + "step": 9371 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.1454, + "step": 9372 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.3254, + "step": 9373 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9374 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.9798, + "step": 9375 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9376 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.3573, + "step": 9377 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.386, + "step": 9378 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.0887, + "step": 9379 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9380 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.7286, + "step": 9381 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 1.905, + "step": 9382 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.3437, + "step": 9383 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9384 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9385 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9386 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9387 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 3.9305, + "step": 9388 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 7.9579, + "step": 9389 + }, + { + "epoch": 0.78, + "learning_rate": 0.00019997770780505475, + "loss": 2.8917, + "step": 9390 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.2129, + "step": 9391 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9392 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.9513, + "step": 9393 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.8511, + "step": 9394 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.9416, + "step": 9395 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 4.3605, + "step": 9396 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 5.0577, + "step": 9397 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9398 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0781, + "step": 9399 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.9955, + "step": 9400 + }, + { + "epoch": 0.79, + "eval_loss": NaN, + "eval_runtime": 40.1578, + "eval_samples_per_second": 62.254, + "eval_steps_per_second": 7.794, + "step": 9400 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.8691, + "step": 9401 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9402 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 4.0693, + "step": 9403 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9404 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.0894, + "step": 9405 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 8.5327, + "step": 9406 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0626, + "step": 9407 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 5.7196, + "step": 9408 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.1482, + "step": 9409 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.5881, + "step": 9410 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 8.3708, + "step": 9411 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9412 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.0142, + "step": 9413 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 8.5632, + "step": 9414 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.7075, + "step": 9415 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9416 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.9742, + "step": 9417 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9418 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9419 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.1535, + "step": 9420 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9421 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9422 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9423 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.3477, + "step": 9424 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9425 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9426 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9427 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.4972, + "step": 9428 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.8559, + "step": 9429 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9430 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.6282, + "step": 9431 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 6.0704, + "step": 9432 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9433 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.7681, + "step": 9434 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.8961, + "step": 9435 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9436 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9437 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9438 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.8764, + "step": 9439 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.3585, + "step": 9440 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.718, + "step": 9441 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 6.8439, + "step": 9442 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 5.6632, + "step": 9443 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.9817, + "step": 9444 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9445 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.8776, + "step": 9446 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.0526, + "step": 9447 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0651, + "step": 9448 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9449 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9450 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.7664, + "step": 9451 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9452 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9453 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9454 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9455 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9456 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.9393, + "step": 9457 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9458 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9459 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9460 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.4032, + "step": 9461 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9462 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.3515, + "step": 9463 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 4.184, + "step": 9464 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0847, + "step": 9465 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9466 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0052, + "step": 9467 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.1118, + "step": 9468 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 6.9236, + "step": 9469 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9470 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9471 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.9228, + "step": 9472 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9473 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.8992, + "step": 9474 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 5.1633, + "step": 9475 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.788, + "step": 9476 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9477 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.3881, + "step": 9478 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.9745, + "step": 9479 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9480 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9481 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.1175, + "step": 9482 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9483 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.8697, + "step": 9484 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.8008, + "step": 9485 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0024, + "step": 9486 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9487 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9488 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0229, + "step": 9489 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 4.4931, + "step": 9490 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 4.7773, + "step": 9491 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.7801, + "step": 9492 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 2.0377, + "step": 9493 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 4.5013, + "step": 9494 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 8.8226, + "step": 9495 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 9.5117, + "step": 9496 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.5191, + "step": 9497 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.7556, + "step": 9498 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9499 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.1342, + "step": 9500 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 1.3281, + "step": 9501 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.4562, + "step": 9502 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9503 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 3.5244, + "step": 9504 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9505 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 4.4564, + "step": 9506 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9507 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 6.8596, + "step": 9508 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9509 + }, + { + "epoch": 0.79, + "learning_rate": 0.00019997770780505475, + "loss": 7.4453, + "step": 9510 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9511 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9512 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 7.5799, + "step": 9513 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 4.5004, + "step": 9514 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 3.5407, + "step": 9515 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9516 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9517 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9518 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.9145, + "step": 9519 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9520 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9521 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9522 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 2.1958, + "step": 9523 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9524 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9525 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9526 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9527 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.7556, + "step": 9528 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.5675, + "step": 9529 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9530 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 7.8782, + "step": 9531 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9532 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9533 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.8421, + "step": 9534 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 4.6409, + "step": 9535 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9536 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9537 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 8.2438, + "step": 9538 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 2.0054, + "step": 9539 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 2.478, + "step": 9540 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 4.2432, + "step": 9541 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.7723, + "step": 9542 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9543 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9544 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9545 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9546 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9547 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 3.6271, + "step": 9548 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9549 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.1198, + "step": 9550 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 7.6731, + "step": 9551 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9552 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 9.041, + "step": 9553 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 3.6076, + "step": 9554 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9555 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9556 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 7.8078, + "step": 9557 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9558 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9559 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 3.994, + "step": 9560 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9561 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9562 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 4.5951, + "step": 9563 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 9.6494, + "step": 9564 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.0851, + "step": 9565 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 8.6509, + "step": 9566 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 10.5627, + "step": 9567 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9568 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9569 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9570 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9571 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9572 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9573 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9574 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 5.0709, + "step": 9575 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9576 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 2.2363, + "step": 9577 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.8383, + "step": 9578 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 3.9177, + "step": 9579 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9580 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 4.2564, + "step": 9581 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9582 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 1.3737, + "step": 9583 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 8.3745, + "step": 9584 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 8.0456, + "step": 9585 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 4.087, + "step": 9586 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.6912, + "step": 9587 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9588 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 8.1045, + "step": 9589 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 8.8856, + "step": 9590 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 2.0524, + "step": 9591 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 7.5523, + "step": 9592 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9593 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 7.2289, + "step": 9594 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9595 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9596 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 3.8619, + "step": 9597 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 3.1537, + "step": 9598 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997770780505475, + "loss": 0.0, + "step": 9599 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 4.4472, + "step": 9600 + }, + { + "epoch": 0.8, + "eval_loss": NaN, + "eval_runtime": 40.6467, + "eval_samples_per_second": 61.506, + "eval_steps_per_second": 7.7, + "step": 9600 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 1.8424, + "step": 9601 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 4.0376, + "step": 9602 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9603 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 4.1414, + "step": 9604 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9605 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 3.8757, + "step": 9606 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 5.0226, + "step": 9607 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9608 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 4.2408, + "step": 9609 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 6.7642, + "step": 9610 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9611 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.9895, + "step": 9612 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9613 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9614 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 1.1073, + "step": 9615 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 4.2482, + "step": 9616 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 3.0272, + "step": 9617 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9618 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 4.7531, + "step": 9619 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9620 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 2.1982, + "step": 9621 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9622 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9623 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9624 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9625 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9626 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 1.5735, + "step": 9627 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9628 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9629 + }, + { + "epoch": 0.8, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9630 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.7856, + "step": 9631 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 8.8295, + "step": 9632 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9633 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 6.6513, + "step": 9634 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.658, + "step": 9635 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.1158, + "step": 9636 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.6239, + "step": 9637 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.9499, + "step": 9638 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.202, + "step": 9639 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9640 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.899, + "step": 9641 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9642 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 8.6583, + "step": 9643 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.0279, + "step": 9644 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9645 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9646 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 3.0083, + "step": 9647 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 6.5835, + "step": 9648 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9649 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9650 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9651 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 8.3236, + "step": 9652 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.9549, + "step": 9653 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 7.5064, + "step": 9654 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9655 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9656 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9657 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9658 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.7172, + "step": 9659 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 6.1881, + "step": 9660 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 3.2198, + "step": 9661 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9662 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 3.5875, + "step": 9663 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9664 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 5.4371, + "step": 9665 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9666 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9667 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 3.7349, + "step": 9668 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.1826, + "step": 9669 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.6495, + "step": 9670 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9671 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 7.1568, + "step": 9672 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.2921, + "step": 9673 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.5545, + "step": 9674 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9675 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.0129, + "step": 9676 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9677 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9678 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9679 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9680 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.0953, + "step": 9681 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 6.5141, + "step": 9682 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9683 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.9606, + "step": 9684 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9685 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9686 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.127, + "step": 9687 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9688 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9689 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9690 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.5857, + "step": 9691 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9692 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9693 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9694 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9695 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.0588, + "step": 9696 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.8234, + "step": 9697 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9698 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9699 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 8.1622, + "step": 9700 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.1162, + "step": 9701 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.9718, + "step": 9702 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.6071, + "step": 9703 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 7.8287, + "step": 9704 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 7.1152, + "step": 9705 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 9.0918, + "step": 9706 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.6379, + "step": 9707 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.4915, + "step": 9708 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9709 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.1444, + "step": 9710 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.8038, + "step": 9711 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 5.9157, + "step": 9712 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.3021, + "step": 9713 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.0721, + "step": 9714 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9715 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 5.1332, + "step": 9716 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9717 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 8.359, + "step": 9718 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9719 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.7369, + "step": 9720 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 8.851, + "step": 9721 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.1034, + "step": 9722 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.8546, + "step": 9723 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9724 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.5405, + "step": 9725 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9726 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 5.4532, + "step": 9727 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 4.2446, + "step": 9728 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9729 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9730 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9731 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9732 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 3.6794, + "step": 9733 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9734 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 2.9606, + "step": 9735 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 10.7591, + "step": 9736 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 3.3714, + "step": 9737 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9738 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 1.0898, + "step": 9739 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9740 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9741 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9742 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9743 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9744 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9745 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 3.5694, + "step": 9746 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9747 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9748 + }, + { + "epoch": 0.81, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9749 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9750 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9751 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.4922, + "step": 9752 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9753 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 13.6456, + "step": 9754 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 8.9032, + "step": 9755 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9756 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.7523, + "step": 9757 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 1.4179, + "step": 9758 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9759 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9760 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.9205, + "step": 9761 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9762 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.4527, + "step": 9763 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.5375, + "step": 9764 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 7.4451, + "step": 9765 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9766 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 4.2054, + "step": 9767 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 1.9833, + "step": 9768 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.9621, + "step": 9769 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9770 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9771 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9772 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9773 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9774 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9775 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9776 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.7771, + "step": 9777 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 1.5571, + "step": 9778 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.6907, + "step": 9779 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 4.2326, + "step": 9780 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 6.8233, + "step": 9781 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9782 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9783 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9784 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9785 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.6732, + "step": 9786 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.9138, + "step": 9787 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9788 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 6.8011, + "step": 9789 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.0065, + "step": 9790 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9791 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9792 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9793 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.7711, + "step": 9794 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.0964, + "step": 9795 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 4.9458, + "step": 9796 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 8.0822, + "step": 9797 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 1.9228, + "step": 9798 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.1969, + "step": 9799 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9800 + }, + { + "epoch": 0.82, + "eval_loss": NaN, + "eval_runtime": 41.1926, + "eval_samples_per_second": 60.691, + "eval_steps_per_second": 7.598, + "step": 9800 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9801 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9802 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 9.2803, + "step": 9803 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.1878, + "step": 9804 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9805 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9806 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 9.9599, + "step": 9807 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.4315, + "step": 9808 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9809 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9810 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9811 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.916, + "step": 9812 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9813 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.8772, + "step": 9814 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.2508, + "step": 9815 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 10.2361, + "step": 9816 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 1.8347, + "step": 9817 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9818 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.592, + "step": 9819 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9820 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.113, + "step": 9821 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9822 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9823 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.6809, + "step": 9824 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 6.396, + "step": 9825 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9826 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.139, + "step": 9827 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 8.2779, + "step": 9828 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.16, + "step": 9829 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9830 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9831 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9832 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9833 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9834 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.1762, + "step": 9835 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 1.0647, + "step": 9836 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9837 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 6.0557, + "step": 9838 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.9695, + "step": 9839 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9840 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9841 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9842 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.0339, + "step": 9843 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9844 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9845 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.3974, + "step": 9846 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.7086, + "step": 9847 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9848 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 3.2721, + "step": 9849 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.1644, + "step": 9850 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9851 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9852 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.4679, + "step": 9853 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 5.257, + "step": 9854 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9855 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9856 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9857 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 6.6568, + "step": 9858 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9859 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9860 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9861 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.6242, + "step": 9862 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9863 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9864 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9865 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9866 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 6.806, + "step": 9867 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 6.2654, + "step": 9868 + }, + { + "epoch": 0.82, + "learning_rate": 0.00019997213475631847, + "loss": 2.1478, + "step": 9869 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9870 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9871 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9872 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9873 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9874 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9875 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.0677, + "step": 9876 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.0628, + "step": 9877 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 7.4064, + "step": 9878 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9879 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.9526, + "step": 9880 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.7985, + "step": 9881 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.294, + "step": 9882 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.3577, + "step": 9883 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9884 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.1949, + "step": 9885 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9886 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9887 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9888 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 8.8567, + "step": 9889 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 9.5023, + "step": 9890 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9891 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9892 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.8897, + "step": 9893 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9894 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9895 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 4.0492, + "step": 9896 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.02, + "step": 9897 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9898 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9899 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9900 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 2.2457, + "step": 9901 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9902 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9903 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9904 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 8.5656, + "step": 9905 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9906 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9907 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9908 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.8356, + "step": 9909 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 4.2844, + "step": 9910 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9911 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9912 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.8939, + "step": 9913 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9914 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9915 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 2.1501, + "step": 9916 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.864, + "step": 9917 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9918 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 4.5829, + "step": 9919 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9920 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9921 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9922 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9923 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9924 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9925 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 7.3291, + "step": 9926 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9927 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 9.0126, + "step": 9928 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9929 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9930 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 6.2413, + "step": 9931 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.7599, + "step": 9932 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.7375, + "step": 9933 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9934 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9935 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9936 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 4.0915, + "step": 9937 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 8.3743, + "step": 9938 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.1552, + "step": 9939 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9940 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 2.4106, + "step": 9941 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 8.6825, + "step": 9942 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 8.4385, + "step": 9943 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.935, + "step": 9944 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 10.8563, + "step": 9945 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9946 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9947 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.9784, + "step": 9948 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 2.7185, + "step": 9949 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9950 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 2.0556, + "step": 9951 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9952 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9953 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.9776, + "step": 9954 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 6.4129, + "step": 9955 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 4.6387, + "step": 9956 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.2806, + "step": 9957 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 7.5007, + "step": 9958 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9959 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9960 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 7.5336, + "step": 9961 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9962 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 5.7241, + "step": 9963 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 6.7489, + "step": 9964 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.8741, + "step": 9965 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.9781, + "step": 9966 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9967 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9968 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.8754, + "step": 9969 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 1.7291, + "step": 9970 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.1884, + "step": 9971 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9972 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9973 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9974 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 6.6008, + "step": 9975 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 5.2201, + "step": 9976 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.773, + "step": 9977 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9978 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 8.1985, + "step": 9979 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9980 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 4.2319, + "step": 9981 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.3703, + "step": 9982 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9983 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.8971, + "step": 9984 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9985 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 3.1014, + "step": 9986 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9987 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9988 + }, + { + "epoch": 0.83, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9989 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 2.2991, + "step": 9990 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9991 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.7369, + "step": 9992 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.2191, + "step": 9993 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 2.1278, + "step": 9994 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.9729, + "step": 9995 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.1721, + "step": 9996 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.2809, + "step": 9997 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 9998 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.1664, + "step": 9999 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.2062, + "step": 10000 + }, + { + "epoch": 0.84, + "eval_loss": NaN, + "eval_runtime": 40.3692, + "eval_samples_per_second": 61.928, + "eval_steps_per_second": 7.753, + "step": 10000 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 5.4089, + "step": 10001 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10002 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.0943, + "step": 10003 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 9.6029, + "step": 10004 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.7085, + "step": 10005 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.3448, + "step": 10006 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10007 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10008 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10009 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.096, + "step": 10010 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 9.9399, + "step": 10011 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10012 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10013 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 9.9362, + "step": 10014 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10015 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 5.7373, + "step": 10016 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 5.3018, + "step": 10017 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10018 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9415, + "step": 10019 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.8378, + "step": 10020 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10021 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10022 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.7726, + "step": 10023 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 2.9043, + "step": 10024 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.9793, + "step": 10025 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10026 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10027 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10028 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.8566, + "step": 10029 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.5321, + "step": 10030 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 2.7896, + "step": 10031 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10032 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 2.2842, + "step": 10033 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10034 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.1921, + "step": 10035 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9129, + "step": 10036 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.382, + "step": 10037 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10038 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10039 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10040 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10041 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10042 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10043 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.7945, + "step": 10044 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.9827, + "step": 10045 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 8.8265, + "step": 10046 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.138, + "step": 10047 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10048 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.961, + "step": 10049 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10050 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.5543, + "step": 10051 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10052 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 6.013, + "step": 10053 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.033, + "step": 10054 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10055 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10056 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.4862, + "step": 10057 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10058 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 5.6746, + "step": 10059 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10060 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10061 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.1795, + "step": 10062 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.1838, + "step": 10063 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 2.3099, + "step": 10064 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.6428, + "step": 10065 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9428, + "step": 10066 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10067 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 6.2781, + "step": 10068 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10069 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9498, + "step": 10070 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 5.4274, + "step": 10071 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.9973, + "step": 10072 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10073 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.2537, + "step": 10074 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.0603, + "step": 10075 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.4266, + "step": 10076 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9158, + "step": 10077 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10078 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10079 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.2682, + "step": 10080 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.1946, + "step": 10081 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10082 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.0556, + "step": 10083 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10084 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 2.9366, + "step": 10085 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.6015, + "step": 10086 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10087 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.4571, + "step": 10088 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.4917, + "step": 10089 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.4759, + "step": 10090 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.7617, + "step": 10091 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9739, + "step": 10092 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 7.2132, + "step": 10093 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10094 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.0024, + "step": 10095 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 9.24, + "step": 10096 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.1379, + "step": 10097 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9368, + "step": 10098 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 5.9666, + "step": 10099 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.9928, + "step": 10100 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 3.2564, + "step": 10101 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 1.0474, + "step": 10102 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 6.3898, + "step": 10103 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 4.4681, + "step": 10104 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 11.1195, + "step": 10105 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10106 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10107 + }, + { + "epoch": 0.84, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10108 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.3973, + "step": 10109 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10110 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 5.6271, + "step": 10111 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.1855, + "step": 10112 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.1109, + "step": 10113 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.0066, + "step": 10114 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.7878, + "step": 10115 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10116 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.1138, + "step": 10117 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10118 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10119 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.2419, + "step": 10120 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.7028, + "step": 10121 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.6492, + "step": 10122 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.5932, + "step": 10123 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 5.8533, + "step": 10124 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.0539, + "step": 10125 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.8343, + "step": 10126 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.9763, + "step": 10127 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.8893, + "step": 10128 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.0893, + "step": 10129 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.205, + "step": 10130 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.0873, + "step": 10131 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10132 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10133 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.9133, + "step": 10134 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10135 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10136 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.9006, + "step": 10137 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10138 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10139 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.0744, + "step": 10140 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 10.926, + "step": 10141 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10142 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.0841, + "step": 10143 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.6984, + "step": 10144 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.7897, + "step": 10145 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 8.0737, + "step": 10146 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10147 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.1048, + "step": 10148 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.911, + "step": 10149 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.8056, + "step": 10150 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 9.8918, + "step": 10151 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10152 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.5358, + "step": 10153 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.2567, + "step": 10154 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.6608, + "step": 10155 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10156 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10157 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 14.0691, + "step": 10158 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.504, + "step": 10159 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10160 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.0873, + "step": 10161 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10162 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.1919, + "step": 10163 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10164 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10165 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.9886, + "step": 10166 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 9.0877, + "step": 10167 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.0309, + "step": 10168 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10169 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10170 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.8173, + "step": 10171 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10172 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.8988, + "step": 10173 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10174 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.8601, + "step": 10175 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.6515, + "step": 10176 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 5.6198, + "step": 10177 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10178 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.1257, + "step": 10179 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10180 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.8544, + "step": 10181 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 8.5344, + "step": 10182 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.9614, + "step": 10183 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10184 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10185 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.6522, + "step": 10186 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.4672, + "step": 10187 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10188 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10189 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10190 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10191 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10192 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.6456, + "step": 10193 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10194 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10195 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.8808, + "step": 10196 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.8888, + "step": 10197 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.799, + "step": 10198 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10199 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10200 + }, + { + "epoch": 0.85, + "eval_loss": NaN, + "eval_runtime": 39.9063, + "eval_samples_per_second": 62.647, + "eval_steps_per_second": 7.843, + "step": 10200 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.7254, + "step": 10201 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 6.9525, + "step": 10202 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10203 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.7561, + "step": 10204 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10205 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10206 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.8925, + "step": 10207 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 2.0513, + "step": 10208 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.0151, + "step": 10209 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.9096, + "step": 10210 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10211 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10212 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10213 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.903, + "step": 10214 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.8591, + "step": 10215 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10216 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10217 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 7.5723, + "step": 10218 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.4622, + "step": 10219 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 3.519, + "step": 10220 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.7788, + "step": 10221 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10222 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 5.4711, + "step": 10223 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 1.0176, + "step": 10224 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10225 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10226 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 4.8084, + "step": 10227 + }, + { + "epoch": 0.85, + "learning_rate": 0.00019997213475631847, + "loss": 8.6777, + "step": 10228 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10229 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.5398, + "step": 10230 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10231 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.6237, + "step": 10232 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10233 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10234 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 7.5889, + "step": 10235 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10236 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10237 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.8258, + "step": 10238 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10239 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10240 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10241 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 5.373, + "step": 10242 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10243 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 4.6806, + "step": 10244 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10245 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 7.3477, + "step": 10246 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10247 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10248 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 2.3511, + "step": 10249 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10250 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10251 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10252 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10253 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.29, + "step": 10254 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.5583, + "step": 10255 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10256 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10257 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10258 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10259 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 5.2992, + "step": 10260 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.045, + "step": 10261 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10262 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.9338, + "step": 10263 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10264 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.1319, + "step": 10265 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10266 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.6929, + "step": 10267 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.9728, + "step": 10268 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.8622, + "step": 10269 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10270 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.8022, + "step": 10271 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.7501, + "step": 10272 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10273 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10274 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.9444, + "step": 10275 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 4.0923, + "step": 10276 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.9469, + "step": 10277 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10278 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10279 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10280 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 11.7431, + "step": 10281 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.157, + "step": 10282 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 7.3982, + "step": 10283 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 10.215, + "step": 10284 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10285 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.7175, + "step": 10286 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 2.8079, + "step": 10287 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.8624, + "step": 10288 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10289 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10290 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 4.2375, + "step": 10291 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10292 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.9952, + "step": 10293 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10294 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.1137, + "step": 10295 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10296 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10297 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.5509, + "step": 10298 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 9.7483, + "step": 10299 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.5451, + "step": 10300 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 4.765, + "step": 10301 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.7818, + "step": 10302 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10303 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10304 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 2.0993, + "step": 10305 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 2.3711, + "step": 10306 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10307 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10308 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.1002, + "step": 10309 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 7.1782, + "step": 10310 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 7.5807, + "step": 10311 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10312 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10313 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10314 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10315 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10316 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.8463, + "step": 10317 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.4102, + "step": 10318 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.0863, + "step": 10319 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10320 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10321 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 5.1492, + "step": 10322 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.8946, + "step": 10323 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.0422, + "step": 10324 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 5.0543, + "step": 10325 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10326 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 9.0074, + "step": 10327 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 4.1025, + "step": 10328 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 6.4973, + "step": 10329 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.9287, + "step": 10330 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 8.6449, + "step": 10331 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.8828, + "step": 10332 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.2553, + "step": 10333 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10334 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10335 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10336 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 6.7016, + "step": 10337 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10338 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10339 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.7569, + "step": 10340 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 4.8308, + "step": 10341 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 2.105, + "step": 10342 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10343 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10344 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.7688, + "step": 10345 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 3.8563, + "step": 10346 + }, + { + "epoch": 0.86, + "learning_rate": 0.00019997213475631847, + "loss": 1.2721, + "step": 10347 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10348 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 9.2872, + "step": 10349 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10350 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.9795, + "step": 10351 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10352 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.7108, + "step": 10353 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10354 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10355 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10356 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.1859, + "step": 10357 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 6.1156, + "step": 10358 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.813, + "step": 10359 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10360 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.3335, + "step": 10361 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.0764, + "step": 10362 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10363 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.0374, + "step": 10364 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10365 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.97, + "step": 10366 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.8424, + "step": 10367 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 8.9217, + "step": 10368 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 9.2681, + "step": 10369 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10370 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 9.2232, + "step": 10371 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.7895, + "step": 10372 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10373 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.2266, + "step": 10374 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 7.7524, + "step": 10375 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 2.0844, + "step": 10376 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.8154, + "step": 10377 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10378 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.3779, + "step": 10379 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10380 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10381 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10382 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10383 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.863, + "step": 10384 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10385 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 7.8469, + "step": 10386 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10387 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.5452, + "step": 10388 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.0228, + "step": 10389 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10390 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10391 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.2133, + "step": 10392 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 2.677, + "step": 10393 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10394 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 7.9889, + "step": 10395 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10396 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.8787, + "step": 10397 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10398 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 6.8951, + "step": 10399 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 7.0485, + "step": 10400 + }, + { + "epoch": 0.87, + "eval_loss": NaN, + "eval_runtime": 39.9921, + "eval_samples_per_second": 62.512, + "eval_steps_per_second": 7.827, + "step": 10400 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10401 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10402 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10403 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 8.4083, + "step": 10404 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 8.9133, + "step": 10405 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.2099, + "step": 10406 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10407 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10408 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.9467, + "step": 10409 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.2111, + "step": 10410 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.5924, + "step": 10411 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10412 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10413 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10414 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10415 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.0714, + "step": 10416 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.8071, + "step": 10417 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.2983, + "step": 10418 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 8.0698, + "step": 10419 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10420 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10421 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.4642, + "step": 10422 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10423 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.3581, + "step": 10424 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 7.7397, + "step": 10425 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.9765, + "step": 10426 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10427 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10428 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 7.8032, + "step": 10429 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.2547, + "step": 10430 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10431 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.9358, + "step": 10432 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10433 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10434 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10435 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10436 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10437 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10438 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10439 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.7231, + "step": 10440 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.8437, + "step": 10441 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.6017, + "step": 10442 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.2345, + "step": 10443 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10444 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.7284, + "step": 10445 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10446 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10447 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10448 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10449 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.8384, + "step": 10450 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10451 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 2.1745, + "step": 10452 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 2.7715, + "step": 10453 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10454 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 2.1137, + "step": 10455 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.5946, + "step": 10456 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10457 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 3.7235, + "step": 10458 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10459 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 5.8602, + "step": 10460 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10461 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 1.1451, + "step": 10462 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 4.1501, + "step": 10463 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10464 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10465 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10466 + }, + { + "epoch": 0.87, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10467 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.6049, + "step": 10468 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.0039, + "step": 10469 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 8.2658, + "step": 10470 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.6095, + "step": 10471 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10472 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10473 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 3.1406, + "step": 10474 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.8164, + "step": 10475 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 8.4289, + "step": 10476 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 8.9146, + "step": 10477 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10478 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10479 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 3.3182, + "step": 10480 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10481 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10482 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10483 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10484 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10485 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 5.4136, + "step": 10486 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.0925, + "step": 10487 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.3605, + "step": 10488 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 7.1926, + "step": 10489 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10490 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10491 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 5.7887, + "step": 10492 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 6.3164, + "step": 10493 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 4.1261, + "step": 10494 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10495 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10496 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10497 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 3.6987, + "step": 10498 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10499 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10500 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10501 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.7432, + "step": 10502 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 5.2507, + "step": 10503 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.5538, + "step": 10504 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10505 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10506 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.9068, + "step": 10507 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10508 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.9077, + "step": 10509 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10510 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 8.0987, + "step": 10511 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10512 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 7.5583, + "step": 10513 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10514 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10515 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10516 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10517 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10518 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10519 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 9.1151, + "step": 10520 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10521 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.0233, + "step": 10522 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.303, + "step": 10523 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10524 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10525 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10526 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.0364, + "step": 10527 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10528 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10529 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10530 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10531 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10532 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 6.0829, + "step": 10533 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10534 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.0339, + "step": 10535 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 7.4135, + "step": 10536 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 8.0732, + "step": 10537 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 7.8353, + "step": 10538 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.9207, + "step": 10539 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 5.5371, + "step": 10540 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 4.4004, + "step": 10541 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 4.8837, + "step": 10542 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10543 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10544 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10545 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10546 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10547 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10548 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 6.8548, + "step": 10549 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10550 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.9733, + "step": 10551 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10552 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10553 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.3857, + "step": 10554 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10555 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.8841, + "step": 10556 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 7.5168, + "step": 10557 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.5276, + "step": 10558 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.0178, + "step": 10559 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.0523, + "step": 10560 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.0359, + "step": 10561 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10562 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 3.5745, + "step": 10563 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 4.3186, + "step": 10564 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 8.1019, + "step": 10565 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.9212, + "step": 10566 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.0161, + "step": 10567 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10568 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10569 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10570 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10571 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10572 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10573 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 2.4647, + "step": 10574 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10575 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10576 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 3.9472, + "step": 10577 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10578 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 4.3137, + "step": 10579 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10580 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.7279, + "step": 10581 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 8.407, + "step": 10582 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 1.1911, + "step": 10583 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10584 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10585 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10586 + }, + { + "epoch": 0.88, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10587 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.7792, + "step": 10588 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.2738, + "step": 10589 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10590 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10591 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10592 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.3793, + "step": 10593 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10594 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 7.8473, + "step": 10595 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.2402, + "step": 10596 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10597 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 8.0682, + "step": 10598 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10599 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10600 + }, + { + "epoch": 0.89, + "eval_loss": NaN, + "eval_runtime": 40.0291, + "eval_samples_per_second": 62.455, + "eval_steps_per_second": 7.819, + "step": 10600 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.6267, + "step": 10601 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10602 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 6.8636, + "step": 10603 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.4148, + "step": 10604 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.9274, + "step": 10605 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10606 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10607 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10608 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.7795, + "step": 10609 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10610 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.3075, + "step": 10611 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10612 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.2006, + "step": 10613 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.3548, + "step": 10614 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.5707, + "step": 10615 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.9959, + "step": 10616 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.2346, + "step": 10617 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.0354, + "step": 10618 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10619 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.56, + "step": 10620 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10621 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.4231, + "step": 10622 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10623 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10624 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10625 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.4165, + "step": 10626 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 5.0081, + "step": 10627 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.4378, + "step": 10628 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.9548, + "step": 10629 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10630 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10631 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.9619, + "step": 10632 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10633 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10634 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.1802, + "step": 10635 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10636 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10637 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10638 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10639 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 7.5555, + "step": 10640 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10641 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.9803, + "step": 10642 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10643 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10644 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.1305, + "step": 10645 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10646 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 9.1374, + "step": 10647 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10648 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10649 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10650 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10651 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10652 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.6907, + "step": 10653 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10654 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10655 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10656 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.448, + "step": 10657 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.0252, + "step": 10658 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 13.9012, + "step": 10659 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.2541, + "step": 10660 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.5458, + "step": 10661 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.641, + "step": 10662 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.2707, + "step": 10663 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.9915, + "step": 10664 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.9361, + "step": 10665 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10666 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.0282, + "step": 10667 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.8675, + "step": 10668 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10669 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 9.3425, + "step": 10670 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 7.3459, + "step": 10671 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.2497, + "step": 10672 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10673 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10674 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.901, + "step": 10675 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.7077, + "step": 10676 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10677 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.6543, + "step": 10678 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10679 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 10.8782, + "step": 10680 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10681 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10682 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.7285, + "step": 10683 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.0692, + "step": 10684 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.839, + "step": 10685 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10686 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10687 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.9481, + "step": 10688 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10689 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10690 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10691 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10692 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 2.1822, + "step": 10693 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10694 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10695 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 1.8515, + "step": 10696 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10697 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10698 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.9872, + "step": 10699 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 7.0369, + "step": 10700 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 3.8995, + "step": 10701 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10702 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10703 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10704 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 5.9161, + "step": 10705 + }, + { + "epoch": 0.89, + "learning_rate": 0.00019997213475631847, + "loss": 4.2969, + "step": 10706 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.2517, + "step": 10707 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10708 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10709 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10710 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 4.4011, + "step": 10711 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10712 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10713 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10714 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10715 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 8.7437, + "step": 10716 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.887, + "step": 10717 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10718 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.8037, + "step": 10719 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 4.1785, + "step": 10720 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.6217, + "step": 10721 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10722 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.0393, + "step": 10723 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10724 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10725 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.8599, + "step": 10726 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.3646, + "step": 10727 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.736, + "step": 10728 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10729 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10730 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 5.2645, + "step": 10731 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 4.2678, + "step": 10732 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10733 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10734 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10735 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.459, + "step": 10736 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10737 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 2.687, + "step": 10738 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 9.2206, + "step": 10739 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.8776, + "step": 10740 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 8.1336, + "step": 10741 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.2809, + "step": 10742 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 8.1899, + "step": 10743 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.9887, + "step": 10744 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 7.2005, + "step": 10745 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.9622, + "step": 10746 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10747 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 7.8647, + "step": 10748 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10749 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10750 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10751 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10752 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10753 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10754 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.9229, + "step": 10755 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10756 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10757 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.9191, + "step": 10758 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10759 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10760 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.7512, + "step": 10761 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10762 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10763 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10764 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.7747, + "step": 10765 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.1949, + "step": 10766 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 10.3684, + "step": 10767 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10768 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10769 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.0794, + "step": 10770 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10771 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10772 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 9.2052, + "step": 10773 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 4.0542, + "step": 10774 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 5.181, + "step": 10775 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10776 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.9665, + "step": 10777 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 2.0408, + "step": 10778 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.855, + "step": 10779 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10780 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10781 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10782 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10783 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10784 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10785 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10786 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10787 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10788 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.8177, + "step": 10789 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.1781, + "step": 10790 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10791 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 9.5178, + "step": 10792 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.3668, + "step": 10793 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10794 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.8749, + "step": 10795 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.8007, + "step": 10796 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.8954, + "step": 10797 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.8847, + "step": 10798 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.6602, + "step": 10799 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 8.5299, + "step": 10800 + }, + { + "epoch": 0.9, + "eval_loss": NaN, + "eval_runtime": 40.0424, + "eval_samples_per_second": 62.434, + "eval_steps_per_second": 7.817, + "step": 10800 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10801 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.9397, + "step": 10802 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.2771, + "step": 10803 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.7444, + "step": 10804 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 4.2476, + "step": 10805 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10806 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.3723, + "step": 10807 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10808 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10809 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.2277, + "step": 10810 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 5.8718, + "step": 10811 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 2.0449, + "step": 10812 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10813 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.6738, + "step": 10814 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 6.1533, + "step": 10815 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10816 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 8.7928, + "step": 10817 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10818 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 7.1876, + "step": 10819 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 12.1784, + "step": 10820 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10821 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10822 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10823 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 1.6876, + "step": 10824 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 3.4846, + "step": 10825 + }, + { + "epoch": 0.9, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10826 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.5599, + "step": 10827 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.8108, + "step": 10828 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.9314, + "step": 10829 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10830 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10831 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 8.2133, + "step": 10832 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.017, + "step": 10833 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10834 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10835 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10836 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 4.1356, + "step": 10837 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.604, + "step": 10838 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10839 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 7.2969, + "step": 10840 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10841 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 8.7343, + "step": 10842 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 8.4177, + "step": 10843 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 7.5432, + "step": 10844 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 8.0403, + "step": 10845 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10846 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.1968, + "step": 10847 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.7686, + "step": 10848 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10849 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10850 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 7.8763, + "step": 10851 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.0498, + "step": 10852 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.8405, + "step": 10853 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.0575, + "step": 10854 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10855 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10856 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.7057, + "step": 10857 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10858 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 8.4753, + "step": 10859 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 5.5155, + "step": 10860 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 5.0266, + "step": 10861 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.0616, + "step": 10862 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10863 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.9244, + "step": 10864 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10865 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10866 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10867 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10868 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.346, + "step": 10869 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.9834, + "step": 10870 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 4.8163, + "step": 10871 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10872 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 11.3528, + "step": 10873 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 4.8401, + "step": 10874 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.9187, + "step": 10875 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.3211, + "step": 10876 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.2648, + "step": 10877 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10878 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 5.8871, + "step": 10879 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10880 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.6858, + "step": 10881 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10882 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10883 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10884 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 8.9582, + "step": 10885 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.186, + "step": 10886 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10887 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.6021, + "step": 10888 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10889 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.002, + "step": 10890 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10891 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.9754, + "step": 10892 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10893 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10894 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 5.6836, + "step": 10895 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10896 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10897 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.6091, + "step": 10898 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10899 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10900 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.0191, + "step": 10901 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.0973, + "step": 10902 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10903 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10904 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10905 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 9.0681, + "step": 10906 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.9726, + "step": 10907 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10908 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 8.5208, + "step": 10909 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.6194, + "step": 10910 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10911 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.9058, + "step": 10912 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 9.2272, + "step": 10913 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.1732, + "step": 10914 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10915 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.0454, + "step": 10916 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 7.8198, + "step": 10917 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10918 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10919 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10920 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 1.1054, + "step": 10921 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10922 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10923 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.9945, + "step": 10924 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.9242, + "step": 10925 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 6.0947, + "step": 10926 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 4.2063, + "step": 10927 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.7984, + "step": 10928 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10929 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10930 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10931 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10932 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10933 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.3934, + "step": 10934 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 12.9077, + "step": 10935 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10936 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10937 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10938 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 5.819, + "step": 10939 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10940 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.5667, + "step": 10941 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 4.6984, + "step": 10942 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 3.8847, + "step": 10943 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.1008, + "step": 10944 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10945 + }, + { + "epoch": 0.91, + "learning_rate": 0.00019997213475631847, + "loss": 2.7729, + "step": 10946 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10947 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10948 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10949 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10950 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.9365, + "step": 10951 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10952 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10953 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.8458, + "step": 10954 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.4145, + "step": 10955 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.8802, + "step": 10956 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.2048, + "step": 10957 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10958 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10959 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.8214, + "step": 10960 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.1915, + "step": 10961 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10962 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10963 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9761, + "step": 10964 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.7659, + "step": 10965 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10966 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 5.0519, + "step": 10967 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.7574, + "step": 10968 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10969 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.5647, + "step": 10970 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10971 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 9.3255, + "step": 10972 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.2763, + "step": 10973 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10974 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9228, + "step": 10975 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10976 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10977 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9154, + "step": 10978 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10979 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9084, + "step": 10980 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10981 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10982 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.6561, + "step": 10983 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.1228, + "step": 10984 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.1109, + "step": 10985 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10986 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.1188, + "step": 10987 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10988 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.8062, + "step": 10989 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10990 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10991 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.8719, + "step": 10992 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10993 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.6207, + "step": 10994 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 10995 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.1433, + "step": 10996 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9744, + "step": 10997 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.0898, + "step": 10998 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.7839, + "step": 10999 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.0461, + "step": 11000 + }, + { + "epoch": 0.92, + "eval_loss": NaN, + "eval_runtime": 40.0301, + "eval_samples_per_second": 62.453, + "eval_steps_per_second": 7.819, + "step": 11000 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11001 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.6703, + "step": 11002 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11003 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11004 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11005 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.4069, + "step": 11006 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11007 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.4462, + "step": 11008 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11009 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9, + "step": 11010 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11011 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11012 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11013 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11014 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.2159, + "step": 11015 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.0301, + "step": 11016 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.8718, + "step": 11017 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.8722, + "step": 11018 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.1376, + "step": 11019 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.8311, + "step": 11020 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.1432, + "step": 11021 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.433, + "step": 11022 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11023 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.8349, + "step": 11024 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11025 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.3625, + "step": 11026 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.1041, + "step": 11027 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 9.0443, + "step": 11028 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.1671, + "step": 11029 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.7356, + "step": 11030 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.5556, + "step": 11031 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9457, + "step": 11032 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.4293, + "step": 11033 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.627, + "step": 11034 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11035 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11036 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11037 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 6.8462, + "step": 11038 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 3.0221, + "step": 11039 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11040 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11041 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.7698, + "step": 11042 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11043 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.7358, + "step": 11044 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.1063, + "step": 11045 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.5497, + "step": 11046 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11047 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.6094, + "step": 11048 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11049 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11050 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11051 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11052 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11053 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11054 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 8.4128, + "step": 11055 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11056 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11057 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.9002, + "step": 11058 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 1.2362, + "step": 11059 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11060 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11061 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11062 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 7.514, + "step": 11063 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 4.2918, + "step": 11064 + }, + { + "epoch": 0.92, + "learning_rate": 0.00019997213475631847, + "loss": 2.0721, + "step": 11065 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11066 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.7474, + "step": 11067 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11068 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11069 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.9908, + "step": 11070 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.1012, + "step": 11071 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.9824, + "step": 11072 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.8493, + "step": 11073 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11074 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 6.1484, + "step": 11075 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.2387, + "step": 11076 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11077 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 3.8976, + "step": 11078 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.3618, + "step": 11079 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.7995, + "step": 11080 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 7.2432, + "step": 11081 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 6.0721, + "step": 11082 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 7.6733, + "step": 11083 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.6557, + "step": 11084 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11085 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11086 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11087 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11088 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 3.3877, + "step": 11089 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11090 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11091 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 10.3336, + "step": 11092 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.6804, + "step": 11093 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11094 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.0311, + "step": 11095 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.5369, + "step": 11096 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11097 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11098 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.069, + "step": 11099 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.6943, + "step": 11100 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 9.9594, + "step": 11101 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11102 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.9158, + "step": 11103 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.4022, + "step": 11104 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11105 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 6.6043, + "step": 11106 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.8007, + "step": 11107 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 10.034, + "step": 11108 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.3684, + "step": 11109 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.2214, + "step": 11110 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11111 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11112 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11113 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11114 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.0519, + "step": 11115 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11116 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.712, + "step": 11117 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.9198, + "step": 11118 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.5984, + "step": 11119 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11120 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11121 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11122 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 7.1609, + "step": 11123 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11124 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11125 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11126 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11127 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11128 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.9471, + "step": 11129 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.5519, + "step": 11130 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 10.4878, + "step": 11131 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11132 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11133 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.9932, + "step": 11134 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11135 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 6.9928, + "step": 11136 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11137 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.7858, + "step": 11138 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11139 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.1483, + "step": 11140 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.2581, + "step": 11141 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.7916, + "step": 11142 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 9.4862, + "step": 11143 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.7448, + "step": 11144 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.2941, + "step": 11145 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11146 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11147 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11148 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 3.6962, + "step": 11149 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.5504, + "step": 11150 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.4353, + "step": 11151 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11152 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 3.1814, + "step": 11153 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.5218, + "step": 11154 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11155 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11156 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11157 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11158 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.7026, + "step": 11159 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.1726, + "step": 11160 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.5862, + "step": 11161 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11162 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11163 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11164 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 11.4142, + "step": 11165 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11166 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 1.0303, + "step": 11167 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 6.6667, + "step": 11168 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11169 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11170 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 3.7636, + "step": 11171 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.6001, + "step": 11172 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11173 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 3.7623, + "step": 11174 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11175 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11176 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 4.1763, + "step": 11177 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.1061, + "step": 11178 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 5.0006, + "step": 11179 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.9568, + "step": 11180 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 9.3552, + "step": 11181 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 0.9987, + "step": 11182 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.0339, + "step": 11183 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 8.3991, + "step": 11184 + }, + { + "epoch": 0.93, + "learning_rate": 0.00019997213475631847, + "loss": 2.0016, + "step": 11185 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.1729, + "step": 11186 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.9555, + "step": 11187 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11188 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11189 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11190 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 6.687, + "step": 11191 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.146, + "step": 11192 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 5.0145, + "step": 11193 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.4575, + "step": 11194 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 6.9316, + "step": 11195 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 5.3814, + "step": 11196 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11197 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11198 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11199 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.9472, + "step": 11200 + }, + { + "epoch": 0.94, + "eval_loss": NaN, + "eval_runtime": 40.0362, + "eval_samples_per_second": 62.444, + "eval_steps_per_second": 7.818, + "step": 11200 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.13, + "step": 11201 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11202 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11203 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11204 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.7046, + "step": 11205 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.2596, + "step": 11206 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.8554, + "step": 11207 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.446, + "step": 11208 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11209 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11210 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11211 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11212 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11213 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.1287, + "step": 11214 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.0525, + "step": 11215 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.2109, + "step": 11216 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11217 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 6.2942, + "step": 11218 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11219 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.9993, + "step": 11220 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 7.886, + "step": 11221 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.2096, + "step": 11222 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 2.1569, + "step": 11223 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11224 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11225 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11226 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.9092, + "step": 11227 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.3016, + "step": 11228 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.1274, + "step": 11229 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.7895, + "step": 11230 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.2383, + "step": 11231 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11232 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11233 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 6.4827, + "step": 11234 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11235 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11236 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11237 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.7884, + "step": 11238 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11239 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.9894, + "step": 11240 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.9159, + "step": 11241 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11242 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11243 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11244 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.6362, + "step": 11245 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.7366, + "step": 11246 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.7908, + "step": 11247 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.7346, + "step": 11248 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.8438, + "step": 11249 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.4961, + "step": 11250 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11251 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 2.3563, + "step": 11252 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.4457, + "step": 11253 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11254 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11255 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.8515, + "step": 11256 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11257 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11258 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.914, + "step": 11259 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 2.8625, + "step": 11260 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 5.5278, + "step": 11261 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.8867, + "step": 11262 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11263 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 2.1118, + "step": 11264 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11265 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11266 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11267 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.7184, + "step": 11268 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11269 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11270 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11271 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11272 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.6525, + "step": 11273 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11274 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.263, + "step": 11275 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11276 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.3653, + "step": 11277 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11278 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 2.463, + "step": 11279 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 3.8223, + "step": 11280 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11281 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11282 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 2.0247, + "step": 11283 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11284 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.1076, + "step": 11285 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 2.154, + "step": 11286 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11287 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 8.3716, + "step": 11288 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 4.2652, + "step": 11289 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.7886, + "step": 11290 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11291 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11292 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.9623, + "step": 11293 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.978, + "step": 11294 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.8436, + "step": 11295 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.0692, + "step": 11296 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.0315, + "step": 11297 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11298 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 1.127, + "step": 11299 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11300 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11301 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11302 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11303 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11304 + }, + { + "epoch": 0.94, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11305 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 7.9504, + "step": 11306 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.9805, + "step": 11307 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 7.426, + "step": 11308 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.5386, + "step": 11309 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11310 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.051, + "step": 11311 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11312 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 5.7391, + "step": 11313 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11314 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.1504, + "step": 11315 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.3894, + "step": 11316 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11317 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11318 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.5411, + "step": 11319 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11320 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11321 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11322 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11323 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.6256, + "step": 11324 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.2557, + "step": 11325 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 8.4753, + "step": 11326 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11327 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11328 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.4337, + "step": 11329 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11330 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11331 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11332 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.1652, + "step": 11333 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11334 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11335 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11336 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 5.3624, + "step": 11337 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.2773, + "step": 11338 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.874, + "step": 11339 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11340 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.3816, + "step": 11341 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.704, + "step": 11342 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.5964, + "step": 11343 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.6169, + "step": 11344 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11345 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11346 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.9556, + "step": 11347 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.1089, + "step": 11348 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11349 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.316, + "step": 11350 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.0142, + "step": 11351 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.8807, + "step": 11352 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11353 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11354 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.6283, + "step": 11355 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.1782, + "step": 11356 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11357 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.9215, + "step": 11358 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.3499, + "step": 11359 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.3252, + "step": 11360 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11361 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.6957, + "step": 11362 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.1308, + "step": 11363 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11364 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.7154, + "step": 11365 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11366 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11367 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11368 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 5.8695, + "step": 11369 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11370 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11371 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11372 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11373 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11374 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 9.4908, + "step": 11375 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11376 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11377 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11378 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.5941, + "step": 11379 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.8494, + "step": 11380 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11381 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.6685, + "step": 11382 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.9531, + "step": 11383 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.0356, + "step": 11384 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11385 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.3059, + "step": 11386 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 6.331, + "step": 11387 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.1105, + "step": 11388 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.1918, + "step": 11389 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11390 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11391 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.8719, + "step": 11392 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.0798, + "step": 11393 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11394 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11395 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11396 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11397 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 3.3026, + "step": 11398 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11399 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 5.075, + "step": 11400 + }, + { + "epoch": 0.95, + "eval_loss": NaN, + "eval_runtime": 40.0138, + "eval_samples_per_second": 62.478, + "eval_steps_per_second": 7.822, + "step": 11400 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11401 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.741, + "step": 11402 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11403 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 5.1469, + "step": 11404 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11405 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11406 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.8727, + "step": 11407 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 4.8822, + "step": 11408 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 10.2262, + "step": 11409 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11410 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.4842, + "step": 11411 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.7828, + "step": 11412 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11413 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.7738, + "step": 11414 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 6.6245, + "step": 11415 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11416 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.8047, + "step": 11417 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.8868, + "step": 11418 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 7.6355, + "step": 11419 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 1.892, + "step": 11420 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11421 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 2.5608, + "step": 11422 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11423 + }, + { + "epoch": 0.95, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11424 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 4.0691, + "step": 11425 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 8.5468, + "step": 11426 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 6.3829, + "step": 11427 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11428 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9061, + "step": 11429 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11430 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 7.0258, + "step": 11431 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 5.9997, + "step": 11432 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11433 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11434 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11435 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 2.0744, + "step": 11436 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11437 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11438 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 6.2825, + "step": 11439 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11440 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.9226, + "step": 11441 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.1922, + "step": 11442 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11443 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11444 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11445 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11446 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 4.503, + "step": 11447 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9263, + "step": 11448 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11449 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 2.0953, + "step": 11450 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11451 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11452 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9459, + "step": 11453 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 2.2508, + "step": 11454 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 7.4621, + "step": 11455 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11456 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.0497, + "step": 11457 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.0942, + "step": 11458 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.7365, + "step": 11459 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11460 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11461 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11462 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.9396, + "step": 11463 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11464 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.7398, + "step": 11465 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.2459, + "step": 11466 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.2305, + "step": 11467 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.8564, + "step": 11468 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11469 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11470 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11471 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11472 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11473 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11474 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.2543, + "step": 11475 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 2.3272, + "step": 11476 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.0001, + "step": 11477 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.3352, + "step": 11478 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11479 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11480 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9023, + "step": 11481 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11482 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11483 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 4.0061, + "step": 11484 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11485 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11486 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11487 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11488 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11489 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.1871, + "step": 11490 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9635, + "step": 11491 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11492 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11493 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 4.2909, + "step": 11494 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 4.6606, + "step": 11495 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 4.1267, + "step": 11496 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11497 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11498 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 7.1881, + "step": 11499 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.169, + "step": 11500 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.0839, + "step": 11501 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.8618, + "step": 11502 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.6421, + "step": 11503 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11504 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 7.9622, + "step": 11505 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11506 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11507 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9682, + "step": 11508 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11509 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11510 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11511 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11512 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11513 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.4953, + "step": 11514 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 2.9273, + "step": 11515 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.8316, + "step": 11516 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 5.2784, + "step": 11517 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 4.0091, + "step": 11518 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 5.0115, + "step": 11519 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.1472, + "step": 11520 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11521 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.7174, + "step": 11522 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11523 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9124, + "step": 11524 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.1652, + "step": 11525 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 10.2018, + "step": 11526 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.8246, + "step": 11527 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11528 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.8052, + "step": 11529 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11530 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 2.075, + "step": 11531 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 1.7207, + "step": 11532 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11533 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11534 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.868, + "step": 11535 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9882, + "step": 11536 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.9059, + "step": 11537 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11538 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 8.4977, + "step": 11539 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11540 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11541 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11542 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.4964, + "step": 11543 + }, + { + "epoch": 0.96, + "learning_rate": 0.00019997213475631847, + "loss": 3.3838, + "step": 11544 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11545 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.7847, + "step": 11546 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 5.1284, + "step": 11547 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 4.2743, + "step": 11548 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11549 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11550 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.8937, + "step": 11551 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 7.5363, + "step": 11552 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11553 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11554 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.1046, + "step": 11555 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.732, + "step": 11556 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11557 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.9917, + "step": 11558 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11559 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11560 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11561 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11562 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11563 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.6422, + "step": 11564 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11565 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 8.7395, + "step": 11566 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 10.1284, + "step": 11567 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11568 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.1351, + "step": 11569 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11570 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.1676, + "step": 11571 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.0916, + "step": 11572 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11573 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.844, + "step": 11574 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11575 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.9614, + "step": 11576 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11577 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.4953, + "step": 11578 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11579 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11580 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11581 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.9668, + "step": 11582 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11583 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.1519, + "step": 11584 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.29, + "step": 11585 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.8989, + "step": 11586 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.1401, + "step": 11587 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11588 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11589 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11590 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 9.8889, + "step": 11591 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 5.4262, + "step": 11592 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11593 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.7311, + "step": 11594 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.0909, + "step": 11595 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.7583, + "step": 11596 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 8.0696, + "step": 11597 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11598 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.6783, + "step": 11599 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11600 + }, + { + "epoch": 0.97, + "eval_loss": NaN, + "eval_runtime": 40.046, + "eval_samples_per_second": 62.428, + "eval_steps_per_second": 7.816, + "step": 11600 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11601 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11602 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 4.6946, + "step": 11603 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11604 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.8871, + "step": 11605 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.8806, + "step": 11606 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.6929, + "step": 11607 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11608 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11609 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11610 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11611 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11612 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11613 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.0851, + "step": 11614 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11615 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.2427, + "step": 11616 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.6279, + "step": 11617 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.5627, + "step": 11618 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 5.6925, + "step": 11619 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 8.2644, + "step": 11620 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 7.1141, + "step": 11621 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.8926, + "step": 11622 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11623 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.862, + "step": 11624 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 4.1425, + "step": 11625 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11626 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11627 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.6816, + "step": 11628 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.7438, + "step": 11629 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11630 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11631 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 6.1403, + "step": 11632 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11633 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11634 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11635 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.9785, + "step": 11636 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.9555, + "step": 11637 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11638 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 6.5444, + "step": 11639 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 5.2592, + "step": 11640 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11641 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.2541, + "step": 11642 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.4067, + "step": 11643 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 4.9399, + "step": 11644 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.9705, + "step": 11645 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11646 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11647 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 6.346, + "step": 11648 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.9949, + "step": 11649 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 3.3321, + "step": 11650 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 5.6785, + "step": 11651 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11652 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11653 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 2.1627, + "step": 11654 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 8.2073, + "step": 11655 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 1.0927, + "step": 11656 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 8.2511, + "step": 11657 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11658 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 7.8496, + "step": 11659 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11660 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11661 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 7.3853, + "step": 11662 + }, + { + "epoch": 0.97, + "learning_rate": 0.00019997213475631847, + "loss": 4.7668, + "step": 11663 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11664 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 4.627, + "step": 11665 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 9.043, + "step": 11666 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.9261, + "step": 11667 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11668 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 4.3307, + "step": 11669 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 4.0727, + "step": 11670 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11671 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.9696, + "step": 11672 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11673 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 8.0544, + "step": 11674 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11675 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.6624, + "step": 11676 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 8.8434, + "step": 11677 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11678 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11679 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 7.4559, + "step": 11680 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11681 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11682 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11683 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 4.1794, + "step": 11684 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 8.2, + "step": 11685 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11686 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 5.6244, + "step": 11687 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.0314, + "step": 11688 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.4398, + "step": 11689 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 2.7535, + "step": 11690 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11691 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.8608, + "step": 11692 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11693 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11694 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.9788, + "step": 11695 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 5.1984, + "step": 11696 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11697 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 2.116, + "step": 11698 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11699 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 4.2388, + "step": 11700 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11701 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 2.415, + "step": 11702 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.6163, + "step": 11703 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.0916, + "step": 11704 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 8.2184, + "step": 11705 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11706 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.953, + "step": 11707 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.9571, + "step": 11708 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 2.0656, + "step": 11709 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11710 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 5.3666, + "step": 11711 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.9404, + "step": 11712 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.9302, + "step": 11713 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.9512, + "step": 11714 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 7.6986, + "step": 11715 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11716 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.4645, + "step": 11717 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 8.6893, + "step": 11718 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11719 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 6.9359, + "step": 11720 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 6.3952, + "step": 11721 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11722 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11723 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.9673, + "step": 11724 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11725 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.748, + "step": 11726 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 6.1952, + "step": 11727 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 7.7158, + "step": 11728 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.698, + "step": 11729 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11730 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.6955, + "step": 11731 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11732 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11733 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 4.6062, + "step": 11734 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11735 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11736 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.7031, + "step": 11737 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11738 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.7914, + "step": 11739 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11740 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 2.4037, + "step": 11741 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.9147, + "step": 11742 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.8506, + "step": 11743 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 1.9718, + "step": 11744 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.9377, + "step": 11745 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.708, + "step": 11746 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 3.2305, + "step": 11747 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11748 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11749 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 0.0, + "step": 11750 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019997213475631847, + "loss": 8.5009, + "step": 11751 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 4.5106, + "step": 11752 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 6.5993, + "step": 11753 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 9.0129, + "step": 11754 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 8.2034, + "step": 11755 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 2.9806, + "step": 11756 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11757 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11758 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11759 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11760 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 5.491, + "step": 11761 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 4.7624, + "step": 11762 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.9737, + "step": 11763 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 1.8549, + "step": 11764 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 2.6986, + "step": 11765 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 4.8005, + "step": 11766 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11767 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 3.3541, + "step": 11768 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11769 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 2.0906, + "step": 11770 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.9152, + "step": 11771 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 7.5658, + "step": 11772 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 3.1449, + "step": 11773 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11774 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11775 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11776 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11777 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11778 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11779 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.9733, + "step": 11780 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 1.0184, + "step": 11781 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 0.9032, + "step": 11782 + }, + { + "epoch": 0.98, + "learning_rate": 0.00019996656170758216, + "loss": 4.0172, + "step": 11783 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11784 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.5492, + "step": 11785 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11786 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.246, + "step": 11787 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 3.6134, + "step": 11788 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 14.2764, + "step": 11789 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.8601, + "step": 11790 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11791 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11792 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.4907, + "step": 11793 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11794 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11795 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11796 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.7205, + "step": 11797 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.0077, + "step": 11798 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11799 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 3.9469, + "step": 11800 + }, + { + "epoch": 0.99, + "eval_loss": NaN, + "eval_runtime": 40.0544, + "eval_samples_per_second": 62.415, + "eval_steps_per_second": 7.814, + "step": 11800 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.7857, + "step": 11801 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 6.9492, + "step": 11802 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11803 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.3026, + "step": 11804 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11805 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11806 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.9426, + "step": 11807 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 5.8071, + "step": 11808 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.1659, + "step": 11809 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11810 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.9349, + "step": 11811 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.2573, + "step": 11812 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.1832, + "step": 11813 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.5341, + "step": 11814 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 6.2385, + "step": 11815 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 5.9183, + "step": 11816 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 10.1184, + "step": 11817 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11818 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11819 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.8527, + "step": 11820 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 3.4748, + "step": 11821 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.2944, + "step": 11822 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11823 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 9.3602, + "step": 11824 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11825 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11826 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 4.0053, + "step": 11827 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.7695, + "step": 11828 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 3.4551, + "step": 11829 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11830 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11831 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 4.3255, + "step": 11832 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11833 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11834 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11835 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.0422, + "step": 11836 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.9261, + "step": 11837 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.5122, + "step": 11838 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.5792, + "step": 11839 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.1476, + "step": 11840 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.2627, + "step": 11841 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11842 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11843 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.031, + "step": 11844 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11845 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 3.0505, + "step": 11846 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.8813, + "step": 11847 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 5.8006, + "step": 11848 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 5.0141, + "step": 11849 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.058, + "step": 11850 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.8632, + "step": 11851 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11852 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11853 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 5.7102, + "step": 11854 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.2379, + "step": 11855 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.7868, + "step": 11856 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.0748, + "step": 11857 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11858 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 3.9994, + "step": 11859 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11860 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.1906, + "step": 11861 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11862 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 6.4821, + "step": 11863 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.3103, + "step": 11864 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.4196, + "step": 11865 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 6.3272, + "step": 11866 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11867 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.9326, + "step": 11868 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11869 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11870 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 9.9375, + "step": 11871 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11872 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 4.0554, + "step": 11873 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.156, + "step": 11874 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11875 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 9.5893, + "step": 11876 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 10.5871, + "step": 11877 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11878 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.9939, + "step": 11879 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.0253, + "step": 11880 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.087, + "step": 11881 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 7.2986, + "step": 11882 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.0657, + "step": 11883 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11884 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 5.3699, + "step": 11885 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.337, + "step": 11886 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 4.1997, + "step": 11887 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11888 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 8.1271, + "step": 11889 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.8778, + "step": 11890 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11891 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11892 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 1.8234, + "step": 11893 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 3.0187, + "step": 11894 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11895 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11896 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 4.1078, + "step": 11897 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11898 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11899 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.1567, + "step": 11900 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 2.0521, + "step": 11901 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11902 + }, + { + "epoch": 0.99, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11903 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.9406, + "step": 11904 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11905 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 6.6722, + "step": 11906 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11907 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11908 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11909 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 4.2873, + "step": 11910 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11911 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11912 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.9689, + "step": 11913 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11914 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.3308, + "step": 11915 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11916 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.8543, + "step": 11917 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11918 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11919 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11920 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.4506, + "step": 11921 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11922 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11923 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.8493, + "step": 11924 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.8954, + "step": 11925 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 2.0707, + "step": 11926 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 8.4445, + "step": 11927 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.0519, + "step": 11928 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.0428, + "step": 11929 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 9.4393, + "step": 11930 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.6502, + "step": 11931 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11932 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11933 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 6.4621, + "step": 11934 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.4322, + "step": 11935 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 4.9213, + "step": 11936 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11937 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 8.5968, + "step": 11938 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11939 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.9016, + "step": 11940 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11941 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11942 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 2.2193, + "step": 11943 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 9.459, + "step": 11944 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.1841, + "step": 11945 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11946 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 2.2336, + "step": 11947 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11948 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11949 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 4.5312, + "step": 11950 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.9097, + "step": 11951 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11952 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 2.8386, + "step": 11953 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11954 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.9049, + "step": 11955 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11956 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 4.3773, + "step": 11957 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11958 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.7772, + "step": 11959 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11960 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 2.9213, + "step": 11961 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11962 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.1119, + "step": 11963 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 7.838, + "step": 11964 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11965 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 2.6525, + "step": 11966 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11967 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11968 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11969 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11970 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.7154, + "step": 11971 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11972 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11973 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.5772, + "step": 11974 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11975 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 4.4593, + "step": 11976 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.0598, + "step": 11977 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11978 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 4.5271, + "step": 11979 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11980 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.1238, + "step": 11981 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.3662, + "step": 11982 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.8961, + "step": 11983 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 5.4438, + "step": 11984 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.7753, + "step": 11985 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 8.7066, + "step": 11986 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.0013, + "step": 11987 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11988 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 5.0049, + "step": 11989 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11990 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11991 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 6.7131, + "step": 11992 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11993 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.8446, + "step": 11994 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11995 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11996 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.1874, + "step": 11997 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11998 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 11999 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12000 + }, + { + "epoch": 1.0, + "eval_loss": NaN, + "eval_runtime": 40.0675, + "eval_samples_per_second": 62.395, + "eval_steps_per_second": 7.812, + "step": 12000 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.941, + "step": 12001 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 8.5386, + "step": 12002 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 3.7395, + "step": 12003 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12004 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12005 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12006 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12007 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 9.149, + "step": 12008 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 6.2473, + "step": 12009 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 4.1111, + "step": 12010 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 1.0483, + "step": 12011 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12012 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 2.1726, + "step": 12013 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12014 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12015 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12016 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 7.366, + "step": 12017 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12018 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12019 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 5.6374, + "step": 12020 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12021 + }, + { + "epoch": 1.0, + "learning_rate": 0.00019996656170758216, + "loss": 5.6857, + "step": 12022 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12023 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 3.4747, + "step": 12024 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12025 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12026 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 8.192, + "step": 12027 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.6419, + "step": 12028 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 4.0195, + "step": 12029 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12030 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12031 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12032 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.6629, + "step": 12033 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12034 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.9689, + "step": 12035 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.6705, + "step": 12036 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12037 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.9874, + "step": 12038 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12039 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.1197, + "step": 12040 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 4.1228, + "step": 12041 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12042 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12043 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12044 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12045 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12046 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12047 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.3184, + "step": 12048 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12049 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12050 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.0916, + "step": 12051 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12052 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.9622, + "step": 12053 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.7136, + "step": 12054 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 7.8236, + "step": 12055 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 3.0919, + "step": 12056 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12057 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12058 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.4339, + "step": 12059 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.8488, + "step": 12060 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 8.0725, + "step": 12061 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12062 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12063 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12064 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.8237, + "step": 12065 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12066 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.6802, + "step": 12067 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12068 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12069 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.1065, + "step": 12070 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12071 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12072 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.0685, + "step": 12073 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12074 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.8565, + "step": 12075 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 3.0855, + "step": 12076 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.067, + "step": 12077 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.858, + "step": 12078 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12079 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12080 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 8.1096, + "step": 12081 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12082 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 6.9308, + "step": 12083 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12084 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12085 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12086 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.9947, + "step": 12087 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12088 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12089 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 13.2085, + "step": 12090 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.2939, + "step": 12091 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 5.9405, + "step": 12092 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12093 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.687, + "step": 12094 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 6.4266, + "step": 12095 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.0826, + "step": 12096 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12097 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.7583, + "step": 12098 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.9732, + "step": 12099 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 7.9323, + "step": 12100 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 3.5036, + "step": 12101 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.0708, + "step": 12102 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12103 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12104 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12105 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12106 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12107 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.1508, + "step": 12108 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12109 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.0615, + "step": 12110 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 8.5667, + "step": 12111 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 3.5593, + "step": 12112 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12113 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12114 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.089, + "step": 12115 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.0282, + "step": 12116 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12117 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12118 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12119 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12120 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.9441, + "step": 12121 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 4.0542, + "step": 12122 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12123 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12124 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12125 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 3.8181, + "step": 12126 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.7133, + "step": 12127 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.9848, + "step": 12128 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12129 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.1309, + "step": 12130 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12131 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 4.2758, + "step": 12132 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.6619, + "step": 12133 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 2.2016, + "step": 12134 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12135 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 7.3787, + "step": 12136 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12137 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.8825, + "step": 12138 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12139 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 4.0619, + "step": 12140 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 1.9877, + "step": 12141 + }, + { + "epoch": 1.01, + "learning_rate": 0.00019996656170758216, + "loss": 4.9976, + "step": 12142 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.1384, + "step": 12143 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 6.3393, + "step": 12144 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.1392, + "step": 12145 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12146 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.031, + "step": 12147 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12148 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 4.8447, + "step": 12149 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12150 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.8993, + "step": 12151 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12152 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12153 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12154 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 8.0597, + "step": 12155 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12156 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12157 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12158 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 6.994, + "step": 12159 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.4572, + "step": 12160 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 7.8728, + "step": 12161 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12162 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 6.2249, + "step": 12163 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12164 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.7453, + "step": 12165 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.277, + "step": 12166 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12167 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 6.9775, + "step": 12168 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 8.5195, + "step": 12169 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.9191, + "step": 12170 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 4.4317, + "step": 12171 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12172 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.9725, + "step": 12173 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.1733, + "step": 12174 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.7959, + "step": 12175 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12176 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12177 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.8326, + "step": 12178 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12179 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12180 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12181 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.964, + "step": 12182 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.3968, + "step": 12183 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12184 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12185 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 6.9977, + "step": 12186 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12187 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12188 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12189 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 8.0204, + "step": 12190 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12191 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.8826, + "step": 12192 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 4.9734, + "step": 12193 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.7938, + "step": 12194 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.9446, + "step": 12195 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 4.3495, + "step": 12196 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12197 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.8645, + "step": 12198 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.8588, + "step": 12199 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.4925, + "step": 12200 + }, + { + "epoch": 1.02, + "eval_loss": NaN, + "eval_runtime": 40.0774, + "eval_samples_per_second": 62.379, + "eval_steps_per_second": 7.81, + "step": 12200 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12201 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 9.4512, + "step": 12202 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 7.4017, + "step": 12203 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.8463, + "step": 12204 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12205 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12206 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.9408, + "step": 12207 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12208 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.8499, + "step": 12209 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 7.6364, + "step": 12210 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12211 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.9792, + "step": 12212 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 7.5924, + "step": 12213 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12214 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.1066, + "step": 12215 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12216 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12217 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.4863, + "step": 12218 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 8.4748, + "step": 12219 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 8.7769, + "step": 12220 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.1144, + "step": 12221 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12222 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 4.6244, + "step": 12223 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12224 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12225 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 4.4317, + "step": 12226 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 6.9904, + "step": 12227 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 5.0857, + "step": 12228 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12229 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12230 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.2245, + "step": 12231 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 8.5466, + "step": 12232 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 7.1595, + "step": 12233 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.0495, + "step": 12234 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12235 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 7.2796, + "step": 12236 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12237 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12238 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12239 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.9003, + "step": 12240 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12241 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12242 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.8395, + "step": 12243 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.9692, + "step": 12244 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12245 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.8635, + "step": 12246 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12247 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12248 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 3.6848, + "step": 12249 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12250 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.8961, + "step": 12251 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.9062, + "step": 12252 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12253 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 4.3495, + "step": 12254 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.944, + "step": 12255 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12256 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12257 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.8292, + "step": 12258 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12259 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 2.0807, + "step": 12260 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 1.8689, + "step": 12261 + }, + { + "epoch": 1.02, + "learning_rate": 0.00019996656170758216, + "loss": 8.4684, + "step": 12262 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.6194, + "step": 12263 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12264 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.9734, + "step": 12265 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12266 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 2.0117, + "step": 12267 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 9.284, + "step": 12268 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.0123, + "step": 12269 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 4.1105, + "step": 12270 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12271 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.0763, + "step": 12272 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12273 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 8.0353, + "step": 12274 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.8074, + "step": 12275 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.1369, + "step": 12276 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12277 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12278 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12279 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12280 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.5156, + "step": 12281 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12282 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12283 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.7527, + "step": 12284 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.0822, + "step": 12285 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12286 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.8634, + "step": 12287 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12288 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12289 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 5.562, + "step": 12290 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.159, + "step": 12291 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12292 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12293 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12294 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12295 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 2.9055, + "step": 12296 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12297 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12298 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12299 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12300 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 4.1145, + "step": 12301 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.6265, + "step": 12302 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.6849, + "step": 12303 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12304 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12305 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12306 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.8864, + "step": 12307 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12308 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 8.1598, + "step": 12309 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12310 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 4.8456, + "step": 12311 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12312 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12313 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.9677, + "step": 12314 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12315 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12316 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12317 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12318 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.1783, + "step": 12319 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12320 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12321 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 5.8778, + "step": 12322 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12323 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.9718, + "step": 12324 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12325 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 2.3334, + "step": 12326 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12327 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12328 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.8245, + "step": 12329 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.8307, + "step": 12330 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.2557, + "step": 12331 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.7277, + "step": 12332 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 8.2571, + "step": 12333 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.3547, + "step": 12334 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 4.3116, + "step": 12335 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12336 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 6.6232, + "step": 12337 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12338 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12339 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 6.5578, + "step": 12340 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 7.6739, + "step": 12341 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12342 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12343 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12344 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 2.2897, + "step": 12345 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12346 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12347 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12348 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.9281, + "step": 12349 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12350 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 2.2395, + "step": 12351 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12352 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.0132, + "step": 12353 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 2.323, + "step": 12354 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.7634, + "step": 12355 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.4867, + "step": 12356 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12357 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12358 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.9701, + "step": 12359 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 8.7394, + "step": 12360 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 4.1964, + "step": 12361 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.6199, + "step": 12362 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12363 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12364 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12365 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.0655, + "step": 12366 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12367 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12368 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 5.4613, + "step": 12369 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12370 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12371 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 4.7621, + "step": 12372 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12373 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.96, + "step": 12374 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12375 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.9404, + "step": 12376 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.8343, + "step": 12377 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 1.2115, + "step": 12378 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12379 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 3.9944, + "step": 12380 + }, + { + "epoch": 1.03, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12381 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.2902, + "step": 12382 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.445, + "step": 12383 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12384 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.0601, + "step": 12385 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12386 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12387 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12388 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12389 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12390 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.0394, + "step": 12391 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12392 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.8524, + "step": 12393 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 7.6011, + "step": 12394 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12395 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12396 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12397 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 2.0373, + "step": 12398 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12399 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12400 + }, + { + "epoch": 1.04, + "eval_loss": NaN, + "eval_runtime": 40.163, + "eval_samples_per_second": 62.246, + "eval_steps_per_second": 7.793, + "step": 12400 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12401 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12402 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12403 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12404 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.3252, + "step": 12405 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12406 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12407 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.4911, + "step": 12408 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.0285, + "step": 12409 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.1389, + "step": 12410 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12411 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12412 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12413 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 8.0378, + "step": 12414 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.0959, + "step": 12415 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12416 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12417 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.8877, + "step": 12418 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.8538, + "step": 12419 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.1014, + "step": 12420 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.967, + "step": 12421 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 7.0966, + "step": 12422 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12423 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 5.3887, + "step": 12424 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.4902, + "step": 12425 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12426 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12427 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 2.0369, + "step": 12428 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.0977, + "step": 12429 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 7.8401, + "step": 12430 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12431 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 5.2799, + "step": 12432 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.3927, + "step": 12433 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12434 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12435 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.4087, + "step": 12436 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.8202, + "step": 12437 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 8.2247, + "step": 12438 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12439 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 7.5717, + "step": 12440 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.8261, + "step": 12441 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12442 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12443 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 5.7409, + "step": 12444 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.355, + "step": 12445 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12446 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12447 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 2.1215, + "step": 12448 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12449 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12450 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.0402, + "step": 12451 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12452 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12453 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12454 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.7713, + "step": 12455 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12456 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12457 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 4.5559, + "step": 12458 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 5.719, + "step": 12459 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12460 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12461 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.7385, + "step": 12462 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12463 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.0372, + "step": 12464 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 7.8614, + "step": 12465 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.0042, + "step": 12466 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 8.965, + "step": 12467 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12468 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.0533, + "step": 12469 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12470 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.8745, + "step": 12471 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12472 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12473 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.8012, + "step": 12474 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.9864, + "step": 12475 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 9.8144, + "step": 12476 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12477 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 2.6312, + "step": 12478 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 8.4932, + "step": 12479 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 7.1419, + "step": 12480 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.9505, + "step": 12481 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12482 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12483 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.9428, + "step": 12484 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 2.2791, + "step": 12485 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.7791, + "step": 12486 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12487 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12488 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.9531, + "step": 12489 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 7.7207, + "step": 12490 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12491 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12492 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.8264, + "step": 12493 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 6.8021, + "step": 12494 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12495 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12496 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 3.3961, + "step": 12497 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.842, + "step": 12498 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 1.4463, + "step": 12499 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12500 + }, + { + "epoch": 1.04, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12501 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 2.2728, + "step": 12502 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12503 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12504 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.0194, + "step": 12505 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12506 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12507 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 6.465, + "step": 12508 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.9572, + "step": 12509 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.6659, + "step": 12510 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.9287, + "step": 12511 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12512 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12513 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12514 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12515 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12516 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12517 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.719, + "step": 12518 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.9165, + "step": 12519 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 7.2115, + "step": 12520 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12521 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 4.0826, + "step": 12522 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12523 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.7951, + "step": 12524 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12525 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.8776, + "step": 12526 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.9953, + "step": 12527 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 6.1617, + "step": 12528 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 2.7928, + "step": 12529 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12530 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 6.2523, + "step": 12531 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12532 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12533 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12534 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.0705, + "step": 12535 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 7.9447, + "step": 12536 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 4.1279, + "step": 12537 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.3131, + "step": 12538 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 2.013, + "step": 12539 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12540 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 2.3138, + "step": 12541 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.8453, + "step": 12542 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12543 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.0935, + "step": 12544 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.8482, + "step": 12545 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12546 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12547 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12548 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.966, + "step": 12549 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12550 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12551 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12552 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.079, + "step": 12553 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12554 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12555 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12556 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12557 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.467, + "step": 12558 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.851, + "step": 12559 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12560 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12561 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 5.1056, + "step": 12562 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12563 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12564 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 3.7771, + "step": 12565 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.8597, + "step": 12566 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12567 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12568 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12569 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.8428, + "step": 12570 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12571 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.8139, + "step": 12572 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 4.065, + "step": 12573 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12574 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.9389, + "step": 12575 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.7279, + "step": 12576 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12577 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12578 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.1443, + "step": 12579 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12580 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12581 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.7739, + "step": 12582 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12583 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 7.7651, + "step": 12584 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12585 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.9231, + "step": 12586 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.05, + "step": 12587 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.1588, + "step": 12588 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 10.36, + "step": 12589 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12590 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 7.6981, + "step": 12591 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 6.2889, + "step": 12592 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12593 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12594 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12595 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 2.8539, + "step": 12596 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 7.9066, + "step": 12597 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.9843, + "step": 12598 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12599 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12600 + }, + { + "epoch": 1.05, + "eval_loss": NaN, + "eval_runtime": 40.1402, + "eval_samples_per_second": 62.282, + "eval_steps_per_second": 7.798, + "step": 12600 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.6577, + "step": 12601 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 8.4481, + "step": 12602 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 8.0645, + "step": 12603 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12604 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12605 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.9103, + "step": 12606 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12607 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.9009, + "step": 12608 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 6.9498, + "step": 12609 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12610 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.5471, + "step": 12611 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.8573, + "step": 12612 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.8155, + "step": 12613 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 6.867, + "step": 12614 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12615 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 1.2875, + "step": 12616 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 4.0816, + "step": 12617 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 5.718, + "step": 12618 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12619 + }, + { + "epoch": 1.05, + "learning_rate": 0.00019996656170758216, + "loss": 0.8314, + "step": 12620 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12621 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.2166, + "step": 12622 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 2.1397, + "step": 12623 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 3.8535, + "step": 12624 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 3.0044, + "step": 12625 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 3.928, + "step": 12626 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12627 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12628 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 7.4228, + "step": 12629 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12630 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12631 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12632 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.6448, + "step": 12633 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12634 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12635 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 2.3731, + "step": 12636 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 2.0511, + "step": 12637 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.8479, + "step": 12638 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12639 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.9168, + "step": 12640 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.9907, + "step": 12641 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.8813, + "step": 12642 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12643 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12644 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12645 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 9.0655, + "step": 12646 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12647 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.0428, + "step": 12648 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12649 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12650 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 7.6557, + "step": 12651 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12652 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12653 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12654 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12655 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 3.6453, + "step": 12656 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12657 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12658 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12659 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12660 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12661 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 8.8217, + "step": 12662 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 8.1571, + "step": 12663 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.9127, + "step": 12664 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12665 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 9.823, + "step": 12666 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12667 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.9303, + "step": 12668 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12669 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.1197, + "step": 12670 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12671 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12672 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.9835, + "step": 12673 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.0133, + "step": 12674 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.5011, + "step": 12675 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12676 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12677 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 9.9378, + "step": 12678 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.6946, + "step": 12679 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12680 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.8337, + "step": 12681 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.8665, + "step": 12682 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12683 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12684 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12685 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 7.4874, + "step": 12686 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12687 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12688 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 5.8407, + "step": 12689 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 8.3018, + "step": 12690 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 2.0758, + "step": 12691 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 6.8552, + "step": 12692 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12693 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12694 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.5912, + "step": 12695 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.7214, + "step": 12696 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 3.2178, + "step": 12697 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12698 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 2.7423, + "step": 12699 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12700 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 3.9155, + "step": 12701 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 2.4898, + "step": 12702 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12703 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12704 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 8.5569, + "step": 12705 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.1076, + "step": 12706 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12707 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12708 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.476, + "step": 12709 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12710 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 3.3108, + "step": 12711 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 5.1525, + "step": 12712 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 7.428, + "step": 12713 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 7.7919, + "step": 12714 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12715 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12716 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 9.7659, + "step": 12717 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.7896, + "step": 12718 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12719 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12720 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12721 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12722 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 6.1235, + "step": 12723 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.7997, + "step": 12724 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 8.3667, + "step": 12725 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.5159, + "step": 12726 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12727 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12728 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.585, + "step": 12729 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.0485, + "step": 12730 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12731 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.1908, + "step": 12732 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12733 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.4591, + "step": 12734 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 4.3404, + "step": 12735 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 5.0352, + "step": 12736 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.7286, + "step": 12737 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 1.8486, + "step": 12738 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 5.8943, + "step": 12739 + }, + { + "epoch": 1.06, + "learning_rate": 0.00019996656170758216, + "loss": 8.1447, + "step": 12740 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.8583, + "step": 12741 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12742 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12743 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12744 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 7.441, + "step": 12745 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 2.1238, + "step": 12746 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 7.6652, + "step": 12747 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 2.7685, + "step": 12748 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.1017, + "step": 12749 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12750 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.1738, + "step": 12751 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.8678, + "step": 12752 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12753 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12754 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.7573, + "step": 12755 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12756 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12757 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 5.3975, + "step": 12758 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12759 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.0241, + "step": 12760 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12761 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.7384, + "step": 12762 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.5249, + "step": 12763 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12764 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.4327, + "step": 12765 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12766 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12767 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 5.4723, + "step": 12768 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12769 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.5322, + "step": 12770 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12771 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.4977, + "step": 12772 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.5351, + "step": 12773 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12774 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 8.019, + "step": 12775 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.0116, + "step": 12776 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 2.4234, + "step": 12777 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 7.8928, + "step": 12778 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12779 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12780 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 2.1953, + "step": 12781 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 5.1676, + "step": 12782 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 13.9908, + "step": 12783 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.965, + "step": 12784 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12785 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12786 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.1466, + "step": 12787 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12788 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12789 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12790 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12791 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.0116, + "step": 12792 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12793 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12794 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.9309, + "step": 12795 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12796 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12797 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12798 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 5.3944, + "step": 12799 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12800 + }, + { + "epoch": 1.07, + "eval_loss": NaN, + "eval_runtime": 40.2173, + "eval_samples_per_second": 62.162, + "eval_steps_per_second": 7.783, + "step": 12800 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12801 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 5.2189, + "step": 12802 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 2.0225, + "step": 12803 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.9847, + "step": 12804 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.66, + "step": 12805 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12806 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 8.7076, + "step": 12807 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12808 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.0447, + "step": 12809 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.8021, + "step": 12810 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12811 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 7.5215, + "step": 12812 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.8176, + "step": 12813 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12814 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12815 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12816 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 7.0764, + "step": 12817 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12818 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 5.6006, + "step": 12819 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.5101, + "step": 12820 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12821 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 7.9789, + "step": 12822 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.2124, + "step": 12823 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.943, + "step": 12824 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.2733, + "step": 12825 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 2.8219, + "step": 12826 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.5157, + "step": 12827 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12828 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.5317, + "step": 12829 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 6.9799, + "step": 12830 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 6.4406, + "step": 12831 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.4155, + "step": 12832 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.3574, + "step": 12833 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.0435, + "step": 12834 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12835 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12836 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12837 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12838 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12839 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12840 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12841 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12842 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12843 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.7656, + "step": 12844 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.7739, + "step": 12845 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.7485, + "step": 12846 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.796, + "step": 12847 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 2.1567, + "step": 12848 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.2009, + "step": 12849 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12850 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12851 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 7.4818, + "step": 12852 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12853 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 3.83, + "step": 12854 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12855 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12856 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12857 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 4.6097, + "step": 12858 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 1.1216, + "step": 12859 + }, + { + "epoch": 1.07, + "learning_rate": 0.00019996656170758216, + "loss": 8.4397, + "step": 12860 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 4.0339, + "step": 12861 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12862 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12863 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12864 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 8.0325, + "step": 12865 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 8.7505, + "step": 12866 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.5997, + "step": 12867 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12868 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 8.5626, + "step": 12869 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.1469, + "step": 12870 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12871 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.5938, + "step": 12872 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 10.1838, + "step": 12873 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12874 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 4.8134, + "step": 12875 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12876 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.7955, + "step": 12877 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 8.0003, + "step": 12878 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.9854, + "step": 12879 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12880 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.1251, + "step": 12881 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 6.8344, + "step": 12882 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.8274, + "step": 12883 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 6.0974, + "step": 12884 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.9297, + "step": 12885 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.1874, + "step": 12886 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12887 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12888 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12889 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12890 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12891 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.0354, + "step": 12892 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 4.4288, + "step": 12893 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 6.528, + "step": 12894 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.6164, + "step": 12895 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.1044, + "step": 12896 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 5.9087, + "step": 12897 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.893, + "step": 12898 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.1964, + "step": 12899 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.9117, + "step": 12900 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 6.5983, + "step": 12901 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12902 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12903 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12904 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12905 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12906 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.8983, + "step": 12907 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.1899, + "step": 12908 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.2775, + "step": 12909 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12910 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.1461, + "step": 12911 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.9393, + "step": 12912 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.0611, + "step": 12913 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12914 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12915 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.8611, + "step": 12916 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12917 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12918 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.7751, + "step": 12919 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12920 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12921 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12922 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12923 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.0463, + "step": 12924 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12925 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12926 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12927 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12928 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12929 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.2125, + "step": 12930 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.2783, + "step": 12931 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12932 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.9621, + "step": 12933 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.8736, + "step": 12934 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12935 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.6308, + "step": 12936 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.3674, + "step": 12937 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12938 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12939 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12940 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.5769, + "step": 12941 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12942 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.8363, + "step": 12943 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.9896, + "step": 12944 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12945 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.5023, + "step": 12946 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.2922, + "step": 12947 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12948 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.6169, + "step": 12949 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.6347, + "step": 12950 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 4.3574, + "step": 12951 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12952 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.7289, + "step": 12953 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 8.4613, + "step": 12954 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12955 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 9.962, + "step": 12956 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12957 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.4838, + "step": 12958 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 8.8302, + "step": 12959 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.6151, + "step": 12960 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12961 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.0909, + "step": 12962 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.8858, + "step": 12963 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12964 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.9576, + "step": 12965 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12966 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12967 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 6.8449, + "step": 12968 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12969 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.6275, + "step": 12970 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 1.8549, + "step": 12971 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 4.0728, + "step": 12972 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 4.2551, + "step": 12973 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12974 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 3.6497, + "step": 12975 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.2705, + "step": 12976 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 0.8821, + "step": 12977 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 7.0507, + "step": 12978 + }, + { + "epoch": 1.08, + "learning_rate": 0.00019996656170758216, + "loss": 2.0045, + "step": 12979 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.1733, + "step": 12980 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 6.1605, + "step": 12981 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.7876, + "step": 12982 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 2.9857, + "step": 12983 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12984 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12985 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.7829, + "step": 12986 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.2716, + "step": 12987 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12988 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.076, + "step": 12989 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 14.0789, + "step": 12990 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 8.2861, + "step": 12991 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12992 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12993 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12994 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12995 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12996 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.8462, + "step": 12997 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12998 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 12999 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13000 + }, + { + "epoch": 1.09, + "eval_loss": NaN, + "eval_runtime": 40.2234, + "eval_samples_per_second": 62.153, + "eval_steps_per_second": 7.782, + "step": 13000 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.1784, + "step": 13001 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.8757, + "step": 13002 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13003 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13004 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.0875, + "step": 13005 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13006 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.8092, + "step": 13007 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13008 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.0486, + "step": 13009 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13010 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.2242, + "step": 13011 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 2.2599, + "step": 13012 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.7718, + "step": 13013 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.8652, + "step": 13014 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13015 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13016 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13017 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13018 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.993, + "step": 13019 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.6868, + "step": 13020 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13021 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13022 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13023 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13024 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 5.1113, + "step": 13025 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13026 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.0973, + "step": 13027 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.0987, + "step": 13028 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13029 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 2.7976, + "step": 13030 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.9853, + "step": 13031 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13032 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13033 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.9723, + "step": 13034 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 7.5925, + "step": 13035 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 2.4798, + "step": 13036 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 5.7221, + "step": 13037 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 6.7187, + "step": 13038 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 8.3828, + "step": 13039 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.5219, + "step": 13040 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13041 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13042 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 13.38, + "step": 13043 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13044 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 7.674, + "step": 13045 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13046 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13047 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 7.7291, + "step": 13048 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.0111, + "step": 13049 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.0546, + "step": 13050 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13051 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 6.1081, + "step": 13052 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13053 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 5.7509, + "step": 13054 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 8.3068, + "step": 13055 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.6585, + "step": 13056 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13057 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 2.0609, + "step": 13058 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.0862, + "step": 13059 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 7.5444, + "step": 13060 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 8.2535, + "step": 13061 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13062 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.2946, + "step": 13063 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.5624, + "step": 13064 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.9546, + "step": 13065 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13066 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.9996, + "step": 13067 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13068 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 3.3758, + "step": 13069 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13070 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13071 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13072 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13073 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13074 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.0914, + "step": 13075 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.9264, + "step": 13076 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.1405, + "step": 13077 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13078 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 2.1591, + "step": 13079 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13080 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.1056, + "step": 13081 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13082 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.9321, + "step": 13083 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.1962, + "step": 13084 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13085 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 7.712, + "step": 13086 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13087 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.2048, + "step": 13088 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13089 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13090 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13091 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 4.0766, + "step": 13092 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 1.032, + "step": 13093 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13094 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13095 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 8.098, + "step": 13096 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13097 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13098 + }, + { + "epoch": 1.09, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13099 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.1249, + "step": 13100 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13101 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13102 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13103 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13104 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.9217, + "step": 13105 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13106 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.7204, + "step": 13107 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13108 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13109 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.758, + "step": 13110 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 5.1577, + "step": 13111 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 4.3093, + "step": 13112 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.9196, + "step": 13113 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.0148, + "step": 13114 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.1124, + "step": 13115 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 7.7259, + "step": 13116 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 5.4313, + "step": 13117 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.1093, + "step": 13118 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13119 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.2003, + "step": 13120 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.3271, + "step": 13121 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13122 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 6.3489, + "step": 13123 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.9711, + "step": 13124 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13125 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13126 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13127 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13128 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13129 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 8.1281, + "step": 13130 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.9846, + "step": 13131 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 4.3737, + "step": 13132 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 6.6493, + "step": 13133 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.0772, + "step": 13134 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13135 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.1658, + "step": 13136 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.6655, + "step": 13137 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13138 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13139 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13140 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.0519, + "step": 13141 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.4716, + "step": 13142 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 8.9034, + "step": 13143 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13144 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.0628, + "step": 13145 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.024, + "step": 13146 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.2796, + "step": 13147 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13148 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.8333, + "step": 13149 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 8.1713, + "step": 13150 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13151 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13152 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.9767, + "step": 13153 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13154 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.9304, + "step": 13155 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13156 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.1128, + "step": 13157 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13158 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 3.545, + "step": 13159 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 8.1386, + "step": 13160 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13161 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 3.8113, + "step": 13162 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.6078, + "step": 13163 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13164 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13165 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13166 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13167 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13168 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 6.8068, + "step": 13169 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13170 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.9614, + "step": 13171 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13172 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 9.0967, + "step": 13173 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 7.3714, + "step": 13174 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 5.2588, + "step": 13175 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.1197, + "step": 13176 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 4.0475, + "step": 13177 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13178 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.7099, + "step": 13179 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13180 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13181 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13182 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13183 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 5.265, + "step": 13184 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.7607, + "step": 13185 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 1.8874, + "step": 13186 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.7715, + "step": 13187 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 4.3021, + "step": 13188 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 3.806, + "step": 13189 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 7.1397, + "step": 13190 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13191 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13192 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 5.0319, + "step": 13193 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.6693, + "step": 13194 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 4.2832, + "step": 13195 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.4126, + "step": 13196 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 10.7788, + "step": 13197 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 3.8583, + "step": 13198 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13199 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13200 + }, + { + "epoch": 1.1, + "eval_loss": NaN, + "eval_runtime": 40.2696, + "eval_samples_per_second": 62.082, + "eval_steps_per_second": 7.773, + "step": 13200 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13201 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13202 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.8518, + "step": 13203 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13204 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 4.1672, + "step": 13205 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13206 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 6.6102, + "step": 13207 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 3.3545, + "step": 13208 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.0724, + "step": 13209 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.9673, + "step": 13210 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.1132, + "step": 13211 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 8.1352, + "step": 13212 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.9428, + "step": 13213 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13214 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 2.5354, + "step": 13215 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13216 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13217 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13218 + }, + { + "epoch": 1.1, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13219 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.902, + "step": 13220 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13221 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13222 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.9093, + "step": 13223 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13224 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 4.3675, + "step": 13225 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 2.0177, + "step": 13226 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 4.3259, + "step": 13227 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13228 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13229 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.9769, + "step": 13230 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13231 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13232 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.046, + "step": 13233 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 5.9803, + "step": 13234 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13235 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13236 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 8.8233, + "step": 13237 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13238 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 1.2349, + "step": 13239 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13240 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.7902, + "step": 13241 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.8767, + "step": 13242 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13243 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 5.1714, + "step": 13244 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 8.6326, + "step": 13245 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13246 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13247 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 4.0391, + "step": 13248 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13249 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13250 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.7627, + "step": 13251 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 2.8118, + "step": 13252 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13253 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 7.1623, + "step": 13254 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.5877, + "step": 13255 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13256 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13257 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13258 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13259 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 1.0128, + "step": 13260 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13261 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13262 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13263 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13264 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.8765, + "step": 13265 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 6.1832, + "step": 13266 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 9.5374, + "step": 13267 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 8.2063, + "step": 13268 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13269 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13270 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13271 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.7361, + "step": 13272 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 4.1297, + "step": 13273 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 1.9445, + "step": 13274 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13275 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13276 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13277 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.1973, + "step": 13278 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.5003, + "step": 13279 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13280 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13281 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 7.4144, + "step": 13282 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13283 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13284 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13285 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 7.2227, + "step": 13286 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13287 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.3875, + "step": 13288 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 2.1321, + "step": 13289 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.9678, + "step": 13290 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13291 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13292 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 7.9972, + "step": 13293 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 2.4977, + "step": 13294 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13295 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 7.0408, + "step": 13296 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13297 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13298 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13299 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13300 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 7.7001, + "step": 13301 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 10.6887, + "step": 13302 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13303 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.9608, + "step": 13304 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 4.118, + "step": 13305 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13306 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 5.0837, + "step": 13307 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.6634, + "step": 13308 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 1.9463, + "step": 13309 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.8541, + "step": 13310 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.9203, + "step": 13311 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13312 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13313 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 4.1837, + "step": 13314 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 1.8399, + "step": 13315 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 5.1835, + "step": 13316 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 5.2897, + "step": 13317 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 1.8938, + "step": 13318 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13319 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13320 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13321 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13322 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.6528, + "step": 13323 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13324 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 5.0244, + "step": 13325 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 10.8553, + "step": 13326 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13327 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 2.113, + "step": 13328 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.9557, + "step": 13329 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.8697, + "step": 13330 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13331 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 4.8599, + "step": 13332 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 2.4056, + "step": 13333 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 3.3534, + "step": 13334 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13335 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13336 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 2.4249, + "step": 13337 + }, + { + "epoch": 1.11, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13338 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13339 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 4.969, + "step": 13340 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13341 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13342 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 6.2104, + "step": 13343 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.3597, + "step": 13344 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13345 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13346 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13347 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 1.2298, + "step": 13348 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.348, + "step": 13349 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 1.7069, + "step": 13350 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13351 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.175, + "step": 13352 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13353 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13354 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 3.7678, + "step": 13355 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.0104, + "step": 13356 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 5.742, + "step": 13357 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13358 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 1.7149, + "step": 13359 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13360 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13361 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 3.6873, + "step": 13362 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13363 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 1.0071, + "step": 13364 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.1476, + "step": 13365 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 1.276, + "step": 13366 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 7.4603, + "step": 13367 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 3.4634, + "step": 13368 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13369 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 6.8322, + "step": 13370 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 1.4819, + "step": 13371 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.443, + "step": 13372 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 8.3778, + "step": 13373 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13374 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13375 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13376 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13377 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13378 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 1.8265, + "step": 13379 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13380 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13381 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 5.7285, + "step": 13382 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 8.3808, + "step": 13383 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.048, + "step": 13384 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 4.1553, + "step": 13385 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13386 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13387 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.6077, + "step": 13388 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13389 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13390 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 7.291, + "step": 13391 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 7.8399, + "step": 13392 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 3.6281, + "step": 13393 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 3.3652, + "step": 13394 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.1708, + "step": 13395 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.9777, + "step": 13396 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13397 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.0493, + "step": 13398 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 4.9818, + "step": 13399 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 2.2816, + "step": 13400 + }, + { + "epoch": 1.12, + "eval_loss": NaN, + "eval_runtime": 40.2586, + "eval_samples_per_second": 62.099, + "eval_steps_per_second": 7.775, + "step": 13400 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13401 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13402 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13403 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13404 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13405 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 0.0, + "step": 13406 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996656170758216, + "loss": 4.3003, + "step": 13407 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 4.0551, + "step": 13408 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 8.5256, + "step": 13409 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13410 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.9355, + "step": 13411 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13412 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13413 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13414 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 1.0815, + "step": 13415 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 8.4675, + "step": 13416 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13417 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 3.4265, + "step": 13418 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 5.3988, + "step": 13419 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13420 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13421 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 2.7109, + "step": 13422 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13423 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 7.2955, + "step": 13424 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 1.7369, + "step": 13425 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13426 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 6.9077, + "step": 13427 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13428 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13429 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13430 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 5.5902, + "step": 13431 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 1.5406, + "step": 13432 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 3.1353, + "step": 13433 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13434 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.8648, + "step": 13435 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13436 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 4.5619, + "step": 13437 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 7.2302, + "step": 13438 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13439 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13440 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 3.1687, + "step": 13441 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 4.68, + "step": 13442 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 3.8115, + "step": 13443 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13444 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 1.0745, + "step": 13445 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.9049, + "step": 13446 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13447 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13448 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 3.8986, + "step": 13449 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13450 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13451 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 3.8614, + "step": 13452 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13453 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13454 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 1.8652, + "step": 13455 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 9.1713, + "step": 13456 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 3.2913, + "step": 13457 + }, + { + "epoch": 1.12, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13458 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13459 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13460 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13461 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13462 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 2.0938, + "step": 13463 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 7.7256, + "step": 13464 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13465 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13466 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 2.2007, + "step": 13467 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13468 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.8192, + "step": 13469 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 6.5947, + "step": 13470 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13471 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 4.4753, + "step": 13472 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13473 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13474 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13475 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 1.164, + "step": 13476 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13477 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 1.9509, + "step": 13478 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 3.1251, + "step": 13479 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13480 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 3.3339, + "step": 13481 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13482 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13483 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13484 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13485 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 1.0319, + "step": 13486 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 2.9115, + "step": 13487 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 5.3026, + "step": 13488 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13489 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13490 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 3.0391, + "step": 13491 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13492 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13493 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13494 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.8363, + "step": 13495 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 6.3835, + "step": 13496 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 6.3566, + "step": 13497 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.9782, + "step": 13498 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 2.4832, + "step": 13499 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 8.3442, + "step": 13500 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 4.1694, + "step": 13501 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 3.0821, + "step": 13502 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 3.7933, + "step": 13503 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 1.8089, + "step": 13504 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 4.7878, + "step": 13505 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 3.5944, + "step": 13506 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 4.5674, + "step": 13507 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 6.7099, + "step": 13508 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 10.6454, + "step": 13509 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13510 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 2.4278, + "step": 13511 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019996098865884582, + "loss": 0.0, + "step": 13512 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 4.2654, + "step": 13513 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 7.6868, + "step": 13514 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 5.919, + "step": 13515 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.9236, + "step": 13516 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.6241, + "step": 13517 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 2.4019, + "step": 13518 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 8.2211, + "step": 13519 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 1.1382, + "step": 13520 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.8834, + "step": 13521 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 8.3094, + "step": 13522 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13523 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.9235, + "step": 13524 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13525 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 2.9077, + "step": 13526 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.9747, + "step": 13527 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 6.4241, + "step": 13528 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13529 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13530 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 7.9832, + "step": 13531 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13532 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 4.6154, + "step": 13533 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 5.8803, + "step": 13534 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13535 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 7.2922, + "step": 13536 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 1.9675, + "step": 13537 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13538 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13539 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13540 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13541 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 1.8777, + "step": 13542 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13543 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.6655, + "step": 13544 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.9824, + "step": 13545 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13546 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13547 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 7.9749, + "step": 13548 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13549 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 9.9696, + "step": 13550 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.9427, + "step": 13551 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 9.3002, + "step": 13552 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13553 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.4664, + "step": 13554 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.6368, + "step": 13555 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 3.0297, + "step": 13556 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 8.1379, + "step": 13557 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 2.1211, + "step": 13558 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13559 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 5.5352, + "step": 13560 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13561 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13562 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13563 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 5.347, + "step": 13564 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13565 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13566 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13567 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 1.6324, + "step": 13568 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13569 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13570 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.8592, + "step": 13571 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13572 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13573 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 8.2678, + "step": 13574 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 7.016, + "step": 13575 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 1.6663, + "step": 13576 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 1.0395, + "step": 13577 + }, + { + "epoch": 1.13, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13578 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.7767, + "step": 13579 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 4.4869, + "step": 13580 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13581 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 8.071, + "step": 13582 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13583 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 4.324, + "step": 13584 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.6485, + "step": 13585 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 9.0331, + "step": 13586 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.5054, + "step": 13587 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13588 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 4.3251, + "step": 13589 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13590 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.834, + "step": 13591 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.9117, + "step": 13592 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13593 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13594 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13595 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 8.8359, + "step": 13596 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13597 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.2857, + "step": 13598 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.8236, + "step": 13599 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13600 + }, + { + "epoch": 1.14, + "eval_loss": NaN, + "eval_runtime": 40.2596, + "eval_samples_per_second": 62.097, + "eval_steps_per_second": 7.775, + "step": 13600 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 8.669, + "step": 13601 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13602 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13603 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13604 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.024, + "step": 13605 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.9625, + "step": 13606 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13607 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.4789, + "step": 13608 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 5.266, + "step": 13609 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.1162, + "step": 13610 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13611 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.6738, + "step": 13612 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.0836, + "step": 13613 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13614 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.4057, + "step": 13615 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.4254, + "step": 13616 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13617 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13618 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.1258, + "step": 13619 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.5546, + "step": 13620 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 8.2865, + "step": 13621 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.6485, + "step": 13622 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.8204, + "step": 13623 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13624 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.2185, + "step": 13625 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13626 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.9812, + "step": 13627 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.7861, + "step": 13628 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 5.418, + "step": 13629 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13630 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13631 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 4.1816, + "step": 13632 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13633 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.7922, + "step": 13634 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 4.3678, + "step": 13635 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.9808, + "step": 13636 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13637 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.9203, + "step": 13638 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.9767, + "step": 13639 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13640 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13641 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13642 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13643 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13644 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13645 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13646 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13647 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13648 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.0976, + "step": 13649 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.0196, + "step": 13650 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13651 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13652 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.2137, + "step": 13653 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 4.4345, + "step": 13654 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.3933, + "step": 13655 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13656 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 9.6531, + "step": 13657 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13658 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.7695, + "step": 13659 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 4.2976, + "step": 13660 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.0062, + "step": 13661 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13662 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13663 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.092, + "step": 13664 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13665 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 8.1631, + "step": 13666 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.0433, + "step": 13667 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13668 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.984, + "step": 13669 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.0818, + "step": 13670 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13671 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13672 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.8824, + "step": 13673 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.7635, + "step": 13674 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.1346, + "step": 13675 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13676 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 5.4511, + "step": 13677 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13678 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13679 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 1.0467, + "step": 13680 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.551, + "step": 13681 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13682 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13683 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13684 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13685 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.3565, + "step": 13686 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 2.1076, + "step": 13687 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13688 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.5182, + "step": 13689 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 9.4475, + "step": 13690 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 3.2546, + "step": 13691 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13692 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 6.6064, + "step": 13693 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13694 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13695 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 7.7558, + "step": 13696 + }, + { + "epoch": 1.14, + "learning_rate": 0.00019995541561010954, + "loss": 0.8996, + "step": 13697 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.2575, + "step": 13698 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13699 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13700 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13701 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.6275, + "step": 13702 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 5.5154, + "step": 13703 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 6.2432, + "step": 13704 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 5.4215, + "step": 13705 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13706 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.2692, + "step": 13707 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.891, + "step": 13708 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.2623, + "step": 13709 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.8277, + "step": 13710 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.8402, + "step": 13711 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13712 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13713 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13714 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.3225, + "step": 13715 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13716 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.4972, + "step": 13717 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 8.4834, + "step": 13718 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 9.0139, + "step": 13719 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13720 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13721 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13722 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.9388, + "step": 13723 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.4711, + "step": 13724 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.9792, + "step": 13725 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.1212, + "step": 13726 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.4214, + "step": 13727 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.8024, + "step": 13728 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13729 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.8518, + "step": 13730 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13731 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13732 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13733 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.096, + "step": 13734 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13735 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.6857, + "step": 13736 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.4716, + "step": 13737 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13738 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.0427, + "step": 13739 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.9187, + "step": 13740 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13741 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.8659, + "step": 13742 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.2785, + "step": 13743 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.3765, + "step": 13744 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13745 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13746 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.4631, + "step": 13747 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.3856, + "step": 13748 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.0826, + "step": 13749 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13750 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13751 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13752 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13753 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13754 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13755 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.6479, + "step": 13756 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13757 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13758 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.6642, + "step": 13759 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13760 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13761 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13762 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13763 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.3064, + "step": 13764 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13765 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13766 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13767 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13768 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 5.2491, + "step": 13769 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.8004, + "step": 13770 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13771 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.6296, + "step": 13772 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.001, + "step": 13773 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13774 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.167, + "step": 13775 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.5072, + "step": 13776 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 16.4385, + "step": 13777 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13778 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13779 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.4874, + "step": 13780 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.3835, + "step": 13781 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13782 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 5.3053, + "step": 13783 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.9289, + "step": 13784 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 4.3105, + "step": 13785 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.8471, + "step": 13786 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13787 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.0716, + "step": 13788 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13789 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13790 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13791 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.9775, + "step": 13792 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13793 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13794 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.0529, + "step": 13795 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 1.1189, + "step": 13796 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 8.4846, + "step": 13797 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13798 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13799 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.7694, + "step": 13800 + }, + { + "epoch": 1.15, + "eval_loss": NaN, + "eval_runtime": 40.2461, + "eval_samples_per_second": 62.118, + "eval_steps_per_second": 7.777, + "step": 13800 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 7.7605, + "step": 13801 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13802 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13803 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13804 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 8.4435, + "step": 13805 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 3.1954, + "step": 13806 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13807 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.7558, + "step": 13808 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 9.414, + "step": 13809 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13810 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13811 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13812 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.9654, + "step": 13813 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13814 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.4614, + "step": 13815 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13816 + }, + { + "epoch": 1.15, + "learning_rate": 0.00019995541561010954, + "loss": 2.9063, + "step": 13817 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 6.2763, + "step": 13818 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 7.0908, + "step": 13819 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.3642, + "step": 13820 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.4268, + "step": 13821 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 3.7466, + "step": 13822 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.8437, + "step": 13823 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13824 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.0916, + "step": 13825 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13826 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 6.7873, + "step": 13827 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.8595, + "step": 13828 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13829 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13830 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13831 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.0233, + "step": 13832 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 3.9633, + "step": 13833 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.9378, + "step": 13834 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 14.7658, + "step": 13835 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.6771, + "step": 13836 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 7.9793, + "step": 13837 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13838 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.0704, + "step": 13839 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13840 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13841 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13842 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13843 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13844 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13845 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13846 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13847 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13848 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13849 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.4342, + "step": 13850 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13851 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.335, + "step": 13852 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 3.1553, + "step": 13853 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13854 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.0413, + "step": 13855 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13856 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 8.6431, + "step": 13857 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13858 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13859 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.1932, + "step": 13860 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.1743, + "step": 13861 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.2292, + "step": 13862 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13863 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.3266, + "step": 13864 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13865 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13866 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.167, + "step": 13867 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13868 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13869 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13870 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 3.9485, + "step": 13871 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13872 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.1734, + "step": 13873 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 7.5853, + "step": 13874 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13875 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13876 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.7184, + "step": 13877 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.9283, + "step": 13878 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13879 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 5.1576, + "step": 13880 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.4771, + "step": 13881 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.5276, + "step": 13882 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13883 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13884 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13885 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.4302, + "step": 13886 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13887 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13888 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 5.4666, + "step": 13889 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 5.1789, + "step": 13890 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13891 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13892 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.2952, + "step": 13893 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13894 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.1005, + "step": 13895 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13896 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13897 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13898 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.3393, + "step": 13899 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 7.0465, + "step": 13900 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13901 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13902 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13903 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.648, + "step": 13904 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13905 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13906 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13907 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.5947, + "step": 13908 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 3.2579, + "step": 13909 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 3.6257, + "step": 13910 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.4547, + "step": 13911 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 8.7494, + "step": 13912 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 6.4024, + "step": 13913 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13914 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.9316, + "step": 13915 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 3.7949, + "step": 13916 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13917 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.0214, + "step": 13918 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13919 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13920 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13921 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13922 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13923 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.9231, + "step": 13924 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.8277, + "step": 13925 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.0754, + "step": 13926 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.9042, + "step": 13927 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13928 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 2.1078, + "step": 13929 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.3711, + "step": 13930 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 4.2954, + "step": 13931 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 1.0371, + "step": 13932 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13933 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13934 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13935 + }, + { + "epoch": 1.16, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13936 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13937 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13938 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13939 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13940 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.1661, + "step": 13941 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 5.9003, + "step": 13942 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 3.9423, + "step": 13943 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13944 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13945 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13946 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13947 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13948 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13949 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 2.3571, + "step": 13950 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13951 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.8425, + "step": 13952 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.7127, + "step": 13953 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13954 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13955 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 5.4636, + "step": 13956 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13957 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 3.9404, + "step": 13958 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13959 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.1942, + "step": 13960 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.636, + "step": 13961 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.9228, + "step": 13962 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.9936, + "step": 13963 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 2.596, + "step": 13964 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13965 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 2.9922, + "step": 13966 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.2153, + "step": 13967 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 2.1909, + "step": 13968 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13969 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13970 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13971 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13972 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.3674, + "step": 13973 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.8852, + "step": 13974 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.0115, + "step": 13975 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13976 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13977 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.4104, + "step": 13978 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.615, + "step": 13979 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13980 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.1646, + "step": 13981 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 5.1869, + "step": 13982 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13983 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13984 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.9249, + "step": 13985 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.2206, + "step": 13986 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13987 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13988 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 3.9987, + "step": 13989 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.6524, + "step": 13990 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 8.5167, + "step": 13991 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 8.1982, + "step": 13992 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 13993 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.0052, + "step": 13994 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.9516, + "step": 13995 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.656, + "step": 13996 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.2557, + "step": 13997 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 10.2894, + "step": 13998 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.4842, + "step": 13999 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14000 + }, + { + "epoch": 1.17, + "eval_loss": NaN, + "eval_runtime": 40.2905, + "eval_samples_per_second": 62.049, + "eval_steps_per_second": 7.769, + "step": 14000 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.9643, + "step": 14001 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14002 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14003 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.7375, + "step": 14004 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14005 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 6.1287, + "step": 14006 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14007 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14008 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14009 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.9824, + "step": 14010 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14011 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14012 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.694, + "step": 14013 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 3.8641, + "step": 14014 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.655, + "step": 14015 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 5.4816, + "step": 14016 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14017 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 2.6717, + "step": 14018 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.0911, + "step": 14019 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 8.5091, + "step": 14020 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.0246, + "step": 14021 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.9794, + "step": 14022 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14023 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14024 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 3.8328, + "step": 14025 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14026 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14027 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 3.6869, + "step": 14028 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14029 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14030 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.9591, + "step": 14031 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14032 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.9789, + "step": 14033 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 3.5945, + "step": 14034 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.0337, + "step": 14035 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14036 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 6.9876, + "step": 14037 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.6445, + "step": 14038 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14039 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14040 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14041 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.6858, + "step": 14042 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 2.1281, + "step": 14043 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 7.4731, + "step": 14044 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14045 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 2.6987, + "step": 14046 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14047 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14048 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 1.1246, + "step": 14049 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14050 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14051 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14052 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.9622, + "step": 14053 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 4.1458, + "step": 14054 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14055 + }, + { + "epoch": 1.17, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14056 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14057 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 9.2064, + "step": 14058 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 1.2189, + "step": 14059 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14060 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 1.6544, + "step": 14061 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 1.9058, + "step": 14062 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14063 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14064 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 5.1288, + "step": 14065 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 7.3777, + "step": 14066 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14067 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14068 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14069 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 3.2294, + "step": 14070 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 3.5852, + "step": 14071 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 7.3194, + "step": 14072 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14073 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14074 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14075 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14076 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14077 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 2.0357, + "step": 14078 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14079 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14080 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 2.2657, + "step": 14081 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 1.2621, + "step": 14082 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14083 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14084 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14085 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 3.9587, + "step": 14086 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14087 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14088 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14089 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14090 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14091 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 5.9605, + "step": 14092 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14093 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 2.5691, + "step": 14094 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 4.0732, + "step": 14095 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14096 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14097 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14098 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14099 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14100 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14101 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14102 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14103 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14104 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14105 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 5.6966, + "step": 14106 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 9.1386, + "step": 14107 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 8.6396, + "step": 14108 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14109 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14110 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 7.1364, + "step": 14111 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 3.0901, + "step": 14112 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 4.0274, + "step": 14113 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14114 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14115 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 6.854, + "step": 14116 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14117 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 2.1028, + "step": 14118 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 4.2425, + "step": 14119 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14120 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 0.0, + "step": 14121 + }, + { + "epoch": 1.18, + "learning_rate": 0.00019995541561010954, + "loss": 5.4394, + "step": 14122 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 4.0022, + "step": 14123 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.7552, + "step": 14124 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 6.4422, + "step": 14125 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14126 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 3.866, + "step": 14127 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 8.1817, + "step": 14128 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14129 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 2.2392, + "step": 14130 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 9.2389, + "step": 14131 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14132 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14133 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.0497, + "step": 14134 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14135 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 2.5131, + "step": 14136 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.593, + "step": 14137 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14138 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 2.3314, + "step": 14139 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14140 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14141 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14142 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 7.1644, + "step": 14143 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14144 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14145 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 2.7257, + "step": 14146 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 8.4786, + "step": 14147 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14148 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14149 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14150 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.3267, + "step": 14151 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14152 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 5.2591, + "step": 14153 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.7656, + "step": 14154 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14155 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 2.1039, + "step": 14156 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14157 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 3.9002, + "step": 14158 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 3.9921, + "step": 14159 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14160 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14161 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14162 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14163 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14164 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 5.1855, + "step": 14165 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 4.0797, + "step": 14166 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.9716, + "step": 14167 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14168 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.9908, + "step": 14169 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.5779, + "step": 14170 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14171 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14172 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14173 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 8.4918, + "step": 14174 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 1.8042, + "step": 14175 + }, + { + "epoch": 1.18, + "learning_rate": 0.0001999498425613732, + "loss": 3.874, + "step": 14176 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 3.9602, + "step": 14177 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 1.1306, + "step": 14178 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14179 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14180 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 8.4832, + "step": 14181 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 5.3742, + "step": 14182 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.8689, + "step": 14183 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 8.0898, + "step": 14184 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14185 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 4.0604, + "step": 14186 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 4.0294, + "step": 14187 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14188 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 3.6708, + "step": 14189 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14190 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 3.8092, + "step": 14191 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.9227, + "step": 14192 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 3.2928, + "step": 14193 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 2.255, + "step": 14194 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 7.2928, + "step": 14195 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 2.7713, + "step": 14196 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14197 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 7.3248, + "step": 14198 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 3.9837, + "step": 14199 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14200 + }, + { + "epoch": 1.19, + "eval_loss": NaN, + "eval_runtime": 40.3067, + "eval_samples_per_second": 62.024, + "eval_steps_per_second": 7.765, + "step": 14200 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14201 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14202 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14203 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 4.517, + "step": 14204 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 2.0928, + "step": 14205 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 2.9589, + "step": 14206 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 1.1382, + "step": 14207 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 3.967, + "step": 14208 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14209 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14210 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14211 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14212 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14213 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 2.1734, + "step": 14214 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 1.7555, + "step": 14215 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14216 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 0.0, + "step": 14217 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 8.9915, + "step": 14218 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 4.0232, + "step": 14219 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999498425613732, + "loss": 8.4126, + "step": 14220 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 3.9166, + "step": 14221 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.1946, + "step": 14222 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.8253, + "step": 14223 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 2.2837, + "step": 14224 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 2.1883, + "step": 14225 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 3.7659, + "step": 14226 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14227 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.9965, + "step": 14228 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14229 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14230 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14231 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14232 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.9917, + "step": 14233 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14234 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 8.5099, + "step": 14235 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14236 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 7.477, + "step": 14237 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14238 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 3.8096, + "step": 14239 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.351, + "step": 14240 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14241 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14242 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14243 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14244 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14245 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 6.1736, + "step": 14246 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 8.5167, + "step": 14247 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 7.9498, + "step": 14248 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 4.4294, + "step": 14249 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 7.9807, + "step": 14250 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 8.7647, + "step": 14251 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.2258, + "step": 14252 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14253 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14254 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14255 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.701, + "step": 14256 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.015, + "step": 14257 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 7.0232, + "step": 14258 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14259 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14260 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 6.6458, + "step": 14261 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14262 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14263 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14264 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.8385, + "step": 14265 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 2.2695, + "step": 14266 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.8652, + "step": 14267 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 2.264, + "step": 14268 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 2.1526, + "step": 14269 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14270 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 3.8723, + "step": 14271 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 7.9658, + "step": 14272 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 3.9507, + "step": 14273 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 8.0694, + "step": 14274 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 2.0243, + "step": 14275 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14276 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 4.0563, + "step": 14277 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 3.2709, + "step": 14278 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14279 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14280 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.8894, + "step": 14281 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14282 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 3.3127, + "step": 14283 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14284 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14285 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 7.5094, + "step": 14286 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 4.2827, + "step": 14287 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 2.6629, + "step": 14288 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14289 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14290 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14291 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.3911, + "step": 14292 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14293 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 1.9172, + "step": 14294 + }, + { + "epoch": 1.19, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14295 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14296 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14297 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 6.2393, + "step": 14298 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.7888, + "step": 14299 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.8763, + "step": 14300 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 4.7459, + "step": 14301 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14302 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.8205, + "step": 14303 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 6.3812, + "step": 14304 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.7098, + "step": 14305 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.2266, + "step": 14306 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.2226, + "step": 14307 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14308 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.9581, + "step": 14309 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14310 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.4555, + "step": 14311 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14312 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 6.1389, + "step": 14313 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 7.619, + "step": 14314 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14315 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14316 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.0252, + "step": 14317 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14318 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.995, + "step": 14319 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 4.1864, + "step": 14320 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 7.8052, + "step": 14321 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14322 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14323 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14324 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14325 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.9328, + "step": 14326 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14327 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.8952, + "step": 14328 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14329 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.3521, + "step": 14330 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14331 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14332 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14333 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.1197, + "step": 14334 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14335 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14336 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14337 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14338 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14339 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.1746, + "step": 14340 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.0892, + "step": 14341 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.5354, + "step": 14342 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.4125, + "step": 14343 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 6.4556, + "step": 14344 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.9442, + "step": 14345 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.4152, + "step": 14346 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.3109, + "step": 14347 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14348 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 8.5655, + "step": 14349 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.9328, + "step": 14350 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14351 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.2109, + "step": 14352 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14353 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14354 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.2745, + "step": 14355 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 4.2871, + "step": 14356 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14357 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14358 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14359 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.8557, + "step": 14360 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 8.5659, + "step": 14361 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14362 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14363 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.9471, + "step": 14364 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14365 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.4735, + "step": 14366 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.2598, + "step": 14367 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.9028, + "step": 14368 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.968, + "step": 14369 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14370 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.2554, + "step": 14371 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14372 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 7.139, + "step": 14373 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.7335, + "step": 14374 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 5.5881, + "step": 14375 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 6.6491, + "step": 14376 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14377 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14378 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.914, + "step": 14379 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.0995, + "step": 14380 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14381 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14382 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14383 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14384 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14385 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 4.2307, + "step": 14386 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.4233, + "step": 14387 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14388 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.3882, + "step": 14389 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 4.1555, + "step": 14390 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14391 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14392 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14393 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14394 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.9197, + "step": 14395 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.4942, + "step": 14396 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.8389, + "step": 14397 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.7421, + "step": 14398 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 11.3361, + "step": 14399 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14400 + }, + { + "epoch": 1.2, + "eval_loss": NaN, + "eval_runtime": 40.2815, + "eval_samples_per_second": 62.063, + "eval_steps_per_second": 7.77, + "step": 14400 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.9041, + "step": 14401 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14402 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.8547, + "step": 14403 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.2173, + "step": 14404 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14405 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.0206, + "step": 14406 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 2.3783, + "step": 14407 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 1.2793, + "step": 14408 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 3.121, + "step": 14409 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 9.1103, + "step": 14410 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14411 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14412 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14413 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 0.9411, + "step": 14414 + }, + { + "epoch": 1.2, + "learning_rate": 0.0001999442695126369, + "loss": 6.4641, + "step": 14415 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14416 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14417 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14418 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.3378, + "step": 14419 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14420 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 7.0946, + "step": 14421 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14422 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 7.4783, + "step": 14423 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.3942, + "step": 14424 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14425 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14426 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.7004, + "step": 14427 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14428 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.9423, + "step": 14429 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.2602, + "step": 14430 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14431 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 7.7933, + "step": 14432 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14433 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.2206, + "step": 14434 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.86, + "step": 14435 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.374, + "step": 14436 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14437 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14438 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14439 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14440 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.5555, + "step": 14441 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14442 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14443 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14444 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.7164, + "step": 14445 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.4104, + "step": 14446 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14447 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 5.3348, + "step": 14448 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.1091, + "step": 14449 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.4344, + "step": 14450 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.9073, + "step": 14451 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.9886, + "step": 14452 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 6.1376, + "step": 14453 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14454 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.9605, + "step": 14455 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14456 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.45, + "step": 14457 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 8.151, + "step": 14458 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.8437, + "step": 14459 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 13.0096, + "step": 14460 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 7.7068, + "step": 14461 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.7336, + "step": 14462 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14463 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.0329, + "step": 14464 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14465 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.9855, + "step": 14466 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.1708, + "step": 14467 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.0552, + "step": 14468 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14469 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.9439, + "step": 14470 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 8.2004, + "step": 14471 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14472 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14473 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.9927, + "step": 14474 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.7842, + "step": 14475 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14476 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14477 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 5.4629, + "step": 14478 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 6.8611, + "step": 14479 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14480 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.0602, + "step": 14481 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.5803, + "step": 14482 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14483 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.1299, + "step": 14484 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14485 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.8504, + "step": 14486 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.5971, + "step": 14487 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14488 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14489 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.9212, + "step": 14490 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.2959, + "step": 14491 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 7.732, + "step": 14492 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.755, + "step": 14493 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 5.3603, + "step": 14494 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.3158, + "step": 14495 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14496 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14497 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14498 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.8579, + "step": 14499 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14500 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14501 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.6517, + "step": 14502 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.9691, + "step": 14503 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.4999, + "step": 14504 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.7809, + "step": 14505 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.9192, + "step": 14506 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14507 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.8627, + "step": 14508 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.1623, + "step": 14509 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14510 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 5.9548, + "step": 14511 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 7.1117, + "step": 14512 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14513 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.9397, + "step": 14514 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14515 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14516 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14517 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.4034, + "step": 14518 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14519 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14520 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.7126, + "step": 14521 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14522 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14523 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 8.2984, + "step": 14524 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14525 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14526 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.8782, + "step": 14527 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 8.3837, + "step": 14528 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 1.9115, + "step": 14529 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14530 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.2178, + "step": 14531 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 3.9339, + "step": 14532 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 4.0329, + "step": 14533 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14534 + }, + { + "epoch": 1.21, + "learning_rate": 0.0001999442695126369, + "loss": 2.0053, + "step": 14535 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14536 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.8133, + "step": 14537 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.8521, + "step": 14538 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14539 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 7.8945, + "step": 14540 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14541 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.7988, + "step": 14542 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14543 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.5353, + "step": 14544 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.0699, + "step": 14545 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14546 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.49, + "step": 14547 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14548 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14549 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14550 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.9598, + "step": 14551 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.8916, + "step": 14552 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.4374, + "step": 14553 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14554 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14555 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.8668, + "step": 14556 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14557 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14558 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.3004, + "step": 14559 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.8892, + "step": 14560 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 9.5133, + "step": 14561 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.2433, + "step": 14562 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14563 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14564 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14565 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14566 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14567 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.0122, + "step": 14568 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.9443, + "step": 14569 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.0289, + "step": 14570 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14571 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14572 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14573 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14574 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14575 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14576 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14577 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14578 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 4.2004, + "step": 14579 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 4.0932, + "step": 14580 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14581 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.9794, + "step": 14582 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 7.3898, + "step": 14583 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.7337, + "step": 14584 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14585 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 4.4355, + "step": 14586 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14587 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14588 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14589 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14590 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 10.2609, + "step": 14591 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14592 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14593 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.5914, + "step": 14594 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.3111, + "step": 14595 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14596 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 4.1414, + "step": 14597 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 4.5706, + "step": 14598 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14599 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.0768, + "step": 14600 + }, + { + "epoch": 1.22, + "eval_loss": NaN, + "eval_runtime": 40.2926, + "eval_samples_per_second": 62.046, + "eval_steps_per_second": 7.768, + "step": 14600 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.9419, + "step": 14601 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 6.3893, + "step": 14602 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14603 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14604 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14605 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.5339, + "step": 14606 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 5.7174, + "step": 14607 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14608 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.8792, + "step": 14609 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14610 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14611 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.0013, + "step": 14612 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14613 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14614 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.2945, + "step": 14615 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 7.8675, + "step": 14616 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14617 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.2432, + "step": 14618 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14619 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.1886, + "step": 14620 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.9219, + "step": 14621 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14622 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14623 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14624 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.7608, + "step": 14625 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14626 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 9.5818, + "step": 14627 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14628 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14629 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.0311, + "step": 14630 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.251, + "step": 14631 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14632 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.2135, + "step": 14633 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14634 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.2901, + "step": 14635 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14636 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 2.2453, + "step": 14637 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.0881, + "step": 14638 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 1.995, + "step": 14639 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14640 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14641 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14642 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14643 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 6.987, + "step": 14644 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.9427, + "step": 14645 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14646 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14647 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14648 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 3.5651, + "step": 14649 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14650 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14651 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14652 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 7.5864, + "step": 14653 + }, + { + "epoch": 1.22, + "learning_rate": 0.0001999442695126369, + "loss": 8.7888, + "step": 14654 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.6784, + "step": 14655 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.6282, + "step": 14656 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 8.458, + "step": 14657 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.6682, + "step": 14658 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14659 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.6083, + "step": 14660 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.9593, + "step": 14661 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.2108, + "step": 14662 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14663 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14664 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14665 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 2.8736, + "step": 14666 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14667 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.2256, + "step": 14668 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14669 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 8.2105, + "step": 14670 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 6.0863, + "step": 14671 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.7966, + "step": 14672 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14673 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14674 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.9752, + "step": 14675 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.4137, + "step": 14676 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 9.0939, + "step": 14677 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14678 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.6683, + "step": 14679 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 9.9196, + "step": 14680 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14681 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.0709, + "step": 14682 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14683 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 6.5869, + "step": 14684 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14685 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14686 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.7131, + "step": 14687 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14688 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.2034, + "step": 14689 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.9916, + "step": 14690 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14691 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.2177, + "step": 14692 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14693 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14694 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.2187, + "step": 14695 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 8.1703, + "step": 14696 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14697 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14698 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 2.0742, + "step": 14699 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 9.3407, + "step": 14700 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14701 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14702 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 5.0937, + "step": 14703 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.6867, + "step": 14704 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.7076, + "step": 14705 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 2.0988, + "step": 14706 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.1384, + "step": 14707 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.8554, + "step": 14708 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.8203, + "step": 14709 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14710 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14711 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14712 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.6996, + "step": 14713 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.8016, + "step": 14714 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14715 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14716 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14717 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.8934, + "step": 14718 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14719 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 9.608, + "step": 14720 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14721 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14722 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.7271, + "step": 14723 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 5.7193, + "step": 14724 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.9093, + "step": 14725 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14726 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 7.9793, + "step": 14727 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 2.045, + "step": 14728 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 9.2426, + "step": 14729 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14730 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.9019, + "step": 14731 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14732 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14733 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 5.406, + "step": 14734 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.1471, + "step": 14735 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 8.2145, + "step": 14736 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.9622, + "step": 14737 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 6.8569, + "step": 14738 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14739 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.0294, + "step": 14740 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.9753, + "step": 14741 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14742 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14743 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.0122, + "step": 14744 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14745 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 6.6101, + "step": 14746 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14747 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14748 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.759, + "step": 14749 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14750 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 8.1441, + "step": 14751 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14752 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14753 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14754 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 5.1708, + "step": 14755 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.4717, + "step": 14756 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 3.9674, + "step": 14757 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14758 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 1.0335, + "step": 14759 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14760 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14761 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14762 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 5.2125, + "step": 14763 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 2.3497, + "step": 14764 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14765 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.6947, + "step": 14766 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14767 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14768 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14769 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14770 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 7.7929, + "step": 14771 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 5.101, + "step": 14772 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 4.5963, + "step": 14773 + }, + { + "epoch": 1.23, + "learning_rate": 0.0001999442695126369, + "loss": 2.0961, + "step": 14774 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14775 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14776 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14777 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 6.3369, + "step": 14778 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14779 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14780 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 2.0885, + "step": 14781 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.9461, + "step": 14782 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 1.0085, + "step": 14783 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14784 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 8.0818, + "step": 14785 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14786 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14787 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 1.1082, + "step": 14788 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14789 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.0, + "step": 14790 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 1.0629, + "step": 14791 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 0.961, + "step": 14792 + }, + { + "epoch": 1.24, + "learning_rate": 0.0001999442695126369, + "loss": 3.5155, + "step": 14793 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 4.5476, + "step": 14794 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14795 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 7.5082, + "step": 14796 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.9956, + "step": 14797 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14798 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 9.7467, + "step": 14799 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14800 + }, + { + "epoch": 1.24, + "eval_loss": NaN, + "eval_runtime": 40.2444, + "eval_samples_per_second": 62.12, + "eval_steps_per_second": 7.777, + "step": 14800 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14801 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14802 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 7.8189, + "step": 14803 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14804 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.2588, + "step": 14805 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 2.2531, + "step": 14806 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14807 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14808 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14809 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14810 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14811 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14812 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14813 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 2.1893, + "step": 14814 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.8867, + "step": 14815 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14816 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 8.7623, + "step": 14817 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.4509, + "step": 14818 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.8882, + "step": 14819 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 6.6287, + "step": 14820 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.1117, + "step": 14821 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14822 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14823 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14824 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14825 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 2.646, + "step": 14826 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.0527, + "step": 14827 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.9844, + "step": 14828 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 2.9465, + "step": 14829 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.7862, + "step": 14830 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.8977, + "step": 14831 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14832 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14833 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 6.1689, + "step": 14834 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14835 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.3677, + "step": 14836 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14837 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14838 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14839 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14840 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 6.1138, + "step": 14841 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 4.0164, + "step": 14842 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.8977, + "step": 14843 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14844 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14845 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14846 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14847 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 7.7489, + "step": 14848 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 2.5383, + "step": 14849 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.4773, + "step": 14850 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 4.1139, + "step": 14851 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14852 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 4.5326, + "step": 14853 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.6827, + "step": 14854 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14855 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14856 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14857 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.3844, + "step": 14858 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14859 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 9.7507, + "step": 14860 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.9876, + "step": 14861 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 4.0797, + "step": 14862 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 2.0439, + "step": 14863 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 6.7344, + "step": 14864 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 9.5809, + "step": 14865 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14866 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.5341, + "step": 14867 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.9148, + "step": 14868 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14869 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.8094, + "step": 14870 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14871 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14872 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 8.326, + "step": 14873 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.9146, + "step": 14874 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14875 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14876 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14877 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.8716, + "step": 14878 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14879 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.2525, + "step": 14880 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 3.4617, + "step": 14881 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14882 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14883 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 1.0753, + "step": 14884 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 7.8899, + "step": 14885 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 2.0196, + "step": 14886 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14887 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14888 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 6.3569, + "step": 14889 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.8192, + "step": 14890 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.9679, + "step": 14891 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14892 + }, + { + "epoch": 1.24, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14893 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14894 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14895 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.478, + "step": 14896 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 7.6001, + "step": 14897 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.9219, + "step": 14898 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 8.1513, + "step": 14899 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 5.7694, + "step": 14900 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.8041, + "step": 14901 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 2.0337, + "step": 14902 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14903 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14904 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 8.33, + "step": 14905 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14906 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 6.1478, + "step": 14907 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14908 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.35, + "step": 14909 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.3599, + "step": 14910 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.0615, + "step": 14911 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14912 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 7.9339, + "step": 14913 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14914 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14915 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14916 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14917 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14918 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14919 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14920 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.1491, + "step": 14921 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 7.8044, + "step": 14922 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.9838, + "step": 14923 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14924 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14925 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 2.5789, + "step": 14926 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14927 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.1481, + "step": 14928 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 2.0407, + "step": 14929 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14930 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14931 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14932 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14933 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 7.4401, + "step": 14934 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14935 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 6.2696, + "step": 14936 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14937 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 7.789, + "step": 14938 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.8917, + "step": 14939 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14940 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 5.5608, + "step": 14941 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.942, + "step": 14942 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.8538, + "step": 14943 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14944 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.4326, + "step": 14945 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.832, + "step": 14946 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 5.328, + "step": 14947 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14948 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 2.1571, + "step": 14949 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14950 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14951 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 12.6299, + "step": 14952 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14953 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.9128, + "step": 14954 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.232, + "step": 14955 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 8.6497, + "step": 14956 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14957 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14958 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14959 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14960 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.9152, + "step": 14961 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14962 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14963 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 7.4858, + "step": 14964 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.9133, + "step": 14965 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 6.584, + "step": 14966 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14967 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 8.2114, + "step": 14968 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14969 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14970 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.0114, + "step": 14971 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.5956, + "step": 14972 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.7883, + "step": 14973 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 9.4753, + "step": 14974 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14975 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14976 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14977 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14978 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14979 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14980 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.7947, + "step": 14981 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14982 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14983 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 5.538, + "step": 14984 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.1891, + "step": 14985 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14986 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14987 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14988 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 6.1843, + "step": 14989 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14990 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14991 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.2159, + "step": 14992 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 6.6868, + "step": 14993 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14994 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.1044, + "step": 14995 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.6803, + "step": 14996 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14997 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14998 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 14999 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15000 + }, + { + "epoch": 1.25, + "eval_loss": NaN, + "eval_runtime": 40.2218, + "eval_samples_per_second": 62.155, + "eval_steps_per_second": 7.782, + "step": 15000 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15001 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.1283, + "step": 15002 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 8.5923, + "step": 15003 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 3.8644, + "step": 15004 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 6.3838, + "step": 15005 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 1.7793, + "step": 15006 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 12.6914, + "step": 15007 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15008 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 4.2575, + "step": 15009 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 6.5865, + "step": 15010 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15011 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15012 + }, + { + "epoch": 1.25, + "learning_rate": 0.00019993869646390059, + "loss": 10.0744, + "step": 15013 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15014 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15015 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15016 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.0815, + "step": 15017 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15018 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.0809, + "step": 15019 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15020 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 7.8004, + "step": 15021 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.1048, + "step": 15022 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 7.4954, + "step": 15023 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.0569, + "step": 15024 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.9612, + "step": 15025 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15026 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15027 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 6.9917, + "step": 15028 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 3.5069, + "step": 15029 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 5.7722, + "step": 15030 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 3.6817, + "step": 15031 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.8422, + "step": 15032 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15033 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15034 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 3.1911, + "step": 15035 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15036 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15037 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15038 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15039 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 8.8489, + "step": 15040 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15041 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15042 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 7.8477, + "step": 15043 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15044 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.8869, + "step": 15045 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.3587, + "step": 15046 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.2511, + "step": 15047 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.8435, + "step": 15048 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15049 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 4.222, + "step": 15050 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15051 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 8.4567, + "step": 15052 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 4.0494, + "step": 15053 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 5.3106, + "step": 15054 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15055 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 4.5209, + "step": 15056 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15057 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15058 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.51, + "step": 15059 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 4.5424, + "step": 15060 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.7455, + "step": 15061 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.3385, + "step": 15062 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15063 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15064 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 7.9175, + "step": 15065 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15066 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.0517, + "step": 15067 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15068 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.8815, + "step": 15069 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15070 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.8178, + "step": 15071 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15072 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 7.8243, + "step": 15073 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15074 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.5799, + "step": 15075 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15076 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15077 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.537, + "step": 15078 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15079 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 3.7746, + "step": 15080 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15081 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15082 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 5.924, + "step": 15083 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.7366, + "step": 15084 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15085 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 5.7486, + "step": 15086 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15087 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.9177, + "step": 15088 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 5.8749, + "step": 15089 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 8.1231, + "step": 15090 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 6.0056, + "step": 15091 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 8.9057, + "step": 15092 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15093 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15094 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.0781, + "step": 15095 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15096 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 8.186, + "step": 15097 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15098 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15099 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 3.7711, + "step": 15100 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.7275, + "step": 15101 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15102 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.6789, + "step": 15103 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.9381, + "step": 15104 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.7324, + "step": 15105 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 6.232, + "step": 15106 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.8289, + "step": 15107 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15108 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15109 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15110 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15111 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15112 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15113 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15114 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15115 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 8.0788, + "step": 15116 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 5.1153, + "step": 15117 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.218, + "step": 15118 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 4.0083, + "step": 15119 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 4.0149, + "step": 15120 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 6.3874, + "step": 15121 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.878, + "step": 15122 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.7727, + "step": 15123 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15124 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15125 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 1.0103, + "step": 15126 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 7.8308, + "step": 15127 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 3.0478, + "step": 15128 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 2.0079, + "step": 15129 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15130 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15131 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15132 + }, + { + "epoch": 1.26, + "learning_rate": 0.00019993869646390059, + "loss": 3.9977, + "step": 15133 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15134 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 7.2233, + "step": 15135 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15136 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15137 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15138 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15139 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15140 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.293, + "step": 15141 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.9636, + "step": 15142 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15143 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 2.5364, + "step": 15144 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15145 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.1039, + "step": 15146 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.955, + "step": 15147 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 5.5664, + "step": 15148 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 7.8319, + "step": 15149 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 5.2531, + "step": 15150 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 7.2155, + "step": 15151 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.9402, + "step": 15152 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 7.4325, + "step": 15153 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15154 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15155 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15156 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15157 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15158 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15159 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.9802, + "step": 15160 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.8143, + "step": 15161 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15162 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15163 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15164 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.8766, + "step": 15165 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.0102, + "step": 15166 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.1174, + "step": 15167 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.7992, + "step": 15168 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 3.2428, + "step": 15169 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 6.3053, + "step": 15170 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.3212, + "step": 15171 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15172 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15173 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15174 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15175 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.7674, + "step": 15176 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 6.4564, + "step": 15177 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.8046, + "step": 15178 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 3.2256, + "step": 15179 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15180 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15181 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15182 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15183 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.9674, + "step": 15184 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 2.0171, + "step": 15185 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15186 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.0891, + "step": 15187 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15188 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.794, + "step": 15189 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.441, + "step": 15190 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15191 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15192 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 2.8814, + "step": 15193 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 8.2003, + "step": 15194 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.0656, + "step": 15195 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.4419, + "step": 15196 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15197 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15198 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15199 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.0235, + "step": 15200 + }, + { + "epoch": 1.27, + "eval_loss": NaN, + "eval_runtime": 40.0749, + "eval_samples_per_second": 62.383, + "eval_steps_per_second": 7.81, + "step": 15200 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.0848, + "step": 15201 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15202 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15203 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 2.1204, + "step": 15204 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.964, + "step": 15205 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.8884, + "step": 15206 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 2.0772, + "step": 15207 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15208 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15209 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15210 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 3.9941, + "step": 15211 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.0488, + "step": 15212 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15213 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.941, + "step": 15214 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.9965, + "step": 15215 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15216 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15217 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15218 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 7.3138, + "step": 15219 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 6.0109, + "step": 15220 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15221 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 7.77, + "step": 15222 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15223 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.6734, + "step": 15224 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15225 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 2.3773, + "step": 15226 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 8.5341, + "step": 15227 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15228 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 9.8485, + "step": 15229 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15230 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15231 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15232 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.945, + "step": 15233 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15234 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 9.1083, + "step": 15235 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 3.3985, + "step": 15236 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 8.4996, + "step": 15237 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.5002, + "step": 15238 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15239 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.2889, + "step": 15240 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 3.9135, + "step": 15241 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.8156, + "step": 15242 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15243 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15244 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.9679, + "step": 15245 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 3.4233, + "step": 15246 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.7418, + "step": 15247 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15248 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 1.7951, + "step": 15249 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 4.1381, + "step": 15250 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15251 + }, + { + "epoch": 1.27, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15252 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15253 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.2088, + "step": 15254 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.2862, + "step": 15255 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15256 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.6641, + "step": 15257 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15258 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15259 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.0733, + "step": 15260 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15261 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.8553, + "step": 15262 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15263 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.9435, + "step": 15264 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15265 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 4.3253, + "step": 15266 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15267 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 7.4459, + "step": 15268 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15269 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15270 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15271 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.2716, + "step": 15272 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 6.7482, + "step": 15273 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.5322, + "step": 15274 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15275 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.8035, + "step": 15276 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.9003, + "step": 15277 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15278 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.9301, + "step": 15279 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.9534, + "step": 15280 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.8549, + "step": 15281 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 2.0197, + "step": 15282 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15283 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 7.4838, + "step": 15284 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 7.7848, + "step": 15285 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15286 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15287 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 5.4358, + "step": 15288 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15289 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15290 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.9114, + "step": 15291 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.2195, + "step": 15292 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15293 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15294 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.7793, + "step": 15295 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 4.1085, + "step": 15296 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 5.9282, + "step": 15297 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15298 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.2342, + "step": 15299 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.6364, + "step": 15300 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15301 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15302 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.9817, + "step": 15303 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15304 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15305 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15306 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.4883, + "step": 15307 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15308 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15309 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 10.0784, + "step": 15310 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15311 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 6.8687, + "step": 15312 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15313 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15314 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.5098, + "step": 15315 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.1291, + "step": 15316 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15317 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15318 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15319 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15320 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.1941, + "step": 15321 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15322 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15323 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 2.0779, + "step": 15324 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 7.2083, + "step": 15325 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15326 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 4.6485, + "step": 15327 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.14, + "step": 15328 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15329 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.9495, + "step": 15330 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.7873, + "step": 15331 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.83, + "step": 15332 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 6.2904, + "step": 15333 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15334 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15335 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15336 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.2499, + "step": 15337 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 5.2108, + "step": 15338 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 4.0701, + "step": 15339 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 7.267, + "step": 15340 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15341 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.6391, + "step": 15342 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 4.4106, + "step": 15343 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15344 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15345 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.5701, + "step": 15346 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 5.3099, + "step": 15347 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.0982, + "step": 15348 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.8318, + "step": 15349 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15350 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 7.4049, + "step": 15351 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15352 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15353 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15354 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15355 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.2829, + "step": 15356 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15357 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.9934, + "step": 15358 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15359 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 1.2827, + "step": 15360 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 8.0794, + "step": 15361 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15362 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 7.3243, + "step": 15363 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15364 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15365 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 3.4298, + "step": 15366 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 11.1586, + "step": 15367 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15368 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15369 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15370 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15371 + }, + { + "epoch": 1.28, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15372 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15373 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.3701, + "step": 15374 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.0244, + "step": 15375 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 8.1185, + "step": 15376 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.2467, + "step": 15377 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.2341, + "step": 15378 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15379 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15380 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15381 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15382 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.2276, + "step": 15383 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15384 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15385 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.3487, + "step": 15386 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15387 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15388 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15389 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 5.8249, + "step": 15390 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15391 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.835, + "step": 15392 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15393 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.9851, + "step": 15394 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.1608, + "step": 15395 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.6307, + "step": 15396 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15397 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 8.5095, + "step": 15398 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.9907, + "step": 15399 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.9727, + "step": 15400 + }, + { + "epoch": 1.29, + "eval_loss": NaN, + "eval_runtime": 40.024, + "eval_samples_per_second": 62.463, + "eval_steps_per_second": 7.82, + "step": 15400 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15401 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15402 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.8503, + "step": 15403 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15404 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.6442, + "step": 15405 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.969, + "step": 15406 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15407 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15408 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.6257, + "step": 15409 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.996, + "step": 15410 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15411 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.8767, + "step": 15412 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15413 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15414 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.5974, + "step": 15415 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 7.4345, + "step": 15416 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.9163, + "step": 15417 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15418 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15419 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15420 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15421 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.7051, + "step": 15422 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 8.1499, + "step": 15423 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.7288, + "step": 15424 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15425 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15426 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15427 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 8.767, + "step": 15428 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.2217, + "step": 15429 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15430 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15431 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.964, + "step": 15432 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.0021, + "step": 15433 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.044, + "step": 15434 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15435 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.2915, + "step": 15436 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15437 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15438 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.4019, + "step": 15439 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15440 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15441 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15442 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.1435, + "step": 15443 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15444 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15445 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.9596, + "step": 15446 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.4722, + "step": 15447 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.5532, + "step": 15448 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.1655, + "step": 15449 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15450 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.1224, + "step": 15451 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15452 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.7181, + "step": 15453 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.2459, + "step": 15454 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 8.4179, + "step": 15455 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 3.3368, + "step": 15456 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15457 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 8.4399, + "step": 15458 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15459 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.7884, + "step": 15460 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15461 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.9169, + "step": 15462 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15463 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15464 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15465 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15466 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.2195, + "step": 15467 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 1.882, + "step": 15468 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.7529, + "step": 15469 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.5917, + "step": 15470 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.0199, + "step": 15471 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.9618, + "step": 15472 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.5194, + "step": 15473 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15474 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 8.3861, + "step": 15475 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15476 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15477 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.692, + "step": 15478 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15479 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15480 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15481 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 7.4679, + "step": 15482 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.4083, + "step": 15483 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15484 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 6.1343, + "step": 15485 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.8623, + "step": 15486 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15487 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.0193, + "step": 15488 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 2.0699, + "step": 15489 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15490 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 4.0108, + "step": 15491 + }, + { + "epoch": 1.29, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15492 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.9664, + "step": 15493 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 4.7121, + "step": 15494 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15495 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.4447, + "step": 15496 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.4345, + "step": 15497 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15498 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 4.5363, + "step": 15499 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15500 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.9874, + "step": 15501 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 2.0143, + "step": 15502 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 5.7859, + "step": 15503 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.2539, + "step": 15504 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15505 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.2239, + "step": 15506 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15507 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15508 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 5.4545, + "step": 15509 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15510 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15511 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15512 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15513 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15514 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 10.3562, + "step": 15515 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15516 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15517 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 2.9205, + "step": 15518 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15519 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.0684, + "step": 15520 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.8762, + "step": 15521 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.9271, + "step": 15522 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15523 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.5043, + "step": 15524 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15525 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15526 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 2.2164, + "step": 15527 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15528 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15529 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15530 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.1642, + "step": 15531 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.9619, + "step": 15532 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15533 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 2.1114, + "step": 15534 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15535 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.9058, + "step": 15536 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.8988, + "step": 15537 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15538 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.7599, + "step": 15539 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 8.2427, + "step": 15540 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15541 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.0569, + "step": 15542 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.6152, + "step": 15543 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 4.5933, + "step": 15544 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15545 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.7548, + "step": 15546 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 6.9387, + "step": 15547 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15548 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.9022, + "step": 15549 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.8132, + "step": 15550 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 4.5687, + "step": 15551 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15552 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 2.1356, + "step": 15553 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.0988, + "step": 15554 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15555 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15556 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.433, + "step": 15557 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 8.369, + "step": 15558 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15559 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15560 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 11.5319, + "step": 15561 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15562 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15563 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15564 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 4.3114, + "step": 15565 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.9836, + "step": 15566 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15567 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 8.1206, + "step": 15568 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15569 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15570 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15571 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15572 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15573 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15574 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15575 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15576 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 2.6602, + "step": 15577 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.6401, + "step": 15578 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15579 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15580 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 5.851, + "step": 15581 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.511, + "step": 15582 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.1834, + "step": 15583 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15584 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 7.2056, + "step": 15585 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15586 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.8232, + "step": 15587 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15588 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 6.5386, + "step": 15589 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15590 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15591 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.9412, + "step": 15592 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 8.9349, + "step": 15593 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.086, + "step": 15594 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.9131, + "step": 15595 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 1.0277, + "step": 15596 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15597 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15598 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15599 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 4.0243, + "step": 15600 + }, + { + "epoch": 1.3, + "eval_loss": NaN, + "eval_runtime": 39.9446, + "eval_samples_per_second": 62.587, + "eval_steps_per_second": 7.836, + "step": 15600 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15601 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 5.9657, + "step": 15602 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15603 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15604 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 5.9427, + "step": 15605 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15606 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15607 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15608 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15609 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 3.5988, + "step": 15610 + }, + { + "epoch": 1.3, + "learning_rate": 0.00019993869646390059, + "loss": 4.0117, + "step": 15611 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15612 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15613 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15614 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 6.89, + "step": 15615 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15616 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15617 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15618 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15619 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 9.5521, + "step": 15620 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.1489, + "step": 15621 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 3.7516, + "step": 15622 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15623 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15624 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15625 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15626 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15627 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 7.6068, + "step": 15628 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15629 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15630 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15631 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.8573, + "step": 15632 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15633 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15634 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15635 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15636 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.9783, + "step": 15637 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 6.474, + "step": 15638 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 7.0526, + "step": 15639 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 5.0692, + "step": 15640 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15641 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15642 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15643 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 9.2684, + "step": 15644 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15645 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 3.2394, + "step": 15646 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.523, + "step": 15647 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15648 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 3.5771, + "step": 15649 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 3.7207, + "step": 15650 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15651 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15652 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15653 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15654 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15655 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.8517, + "step": 15656 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.9145, + "step": 15657 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.8297, + "step": 15658 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15659 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.1828, + "step": 15660 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.2074, + "step": 15661 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15662 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.0485, + "step": 15663 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15664 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15665 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15666 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15667 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15668 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15669 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15670 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15671 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 10.1294, + "step": 15672 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.5016, + "step": 15673 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.7856, + "step": 15674 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 8.938, + "step": 15675 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15676 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 6.0787, + "step": 15677 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15678 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15679 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.8633, + "step": 15680 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15681 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.3998, + "step": 15682 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15683 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 5.026, + "step": 15684 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15685 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15686 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.0511, + "step": 15687 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.3916, + "step": 15688 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.8014, + "step": 15689 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15690 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.5355, + "step": 15691 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 6.6784, + "step": 15692 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.9487, + "step": 15693 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.1271, + "step": 15694 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 6.025, + "step": 15695 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.6625, + "step": 15696 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.2916, + "step": 15697 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15698 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15699 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 5.7211, + "step": 15700 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.418, + "step": 15701 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 8.6389, + "step": 15702 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15703 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.0884, + "step": 15704 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15705 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15706 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15707 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 6.7966, + "step": 15708 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.4163, + "step": 15709 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15710 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.7201, + "step": 15711 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15712 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.2362, + "step": 15713 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15714 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.7299, + "step": 15715 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 3.8153, + "step": 15716 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15717 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 8.6521, + "step": 15718 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 3.1753, + "step": 15719 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15720 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 1.0565, + "step": 15721 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15722 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.7574, + "step": 15723 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.501, + "step": 15724 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15725 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.5473, + "step": 15726 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15727 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 4.036, + "step": 15728 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 2.0612, + "step": 15729 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15730 + }, + { + "epoch": 1.31, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15731 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15732 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15733 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 8.2154, + "step": 15734 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15735 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15736 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 6.3462, + "step": 15737 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15738 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15739 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.3488, + "step": 15740 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.9181, + "step": 15741 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 9.0375, + "step": 15742 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.9177, + "step": 15743 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15744 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15745 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.0841, + "step": 15746 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15747 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.3838, + "step": 15748 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.6044, + "step": 15749 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.7886, + "step": 15750 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15751 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15752 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 10.5221, + "step": 15753 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.8936, + "step": 15754 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 6.4141, + "step": 15755 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15756 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 5.0727, + "step": 15757 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 7.2742, + "step": 15758 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 5.2435, + "step": 15759 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15760 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.2842, + "step": 15761 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15762 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15763 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 7.826, + "step": 15764 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15765 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.5537, + "step": 15766 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 7.7798, + "step": 15767 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.859, + "step": 15768 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15769 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15770 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.1017, + "step": 15771 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.9858, + "step": 15772 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.6203, + "step": 15773 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.7918, + "step": 15774 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15775 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.91, + "step": 15776 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15777 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15778 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.0591, + "step": 15779 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.791, + "step": 15780 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.3564, + "step": 15781 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15782 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15783 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15784 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15785 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.1096, + "step": 15786 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15787 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15788 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15789 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.4456, + "step": 15790 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15791 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15792 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15793 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.756, + "step": 15794 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 5.0245, + "step": 15795 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.9607, + "step": 15796 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.2523, + "step": 15797 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15798 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15799 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.8035, + "step": 15800 + }, + { + "epoch": 1.32, + "eval_loss": NaN, + "eval_runtime": 39.92, + "eval_samples_per_second": 62.625, + "eval_steps_per_second": 7.841, + "step": 15800 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15801 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15802 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 6.1481, + "step": 15803 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15804 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15805 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15806 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.9951, + "step": 15807 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.1744, + "step": 15808 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 5.3582, + "step": 15809 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15810 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15811 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.738, + "step": 15812 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15813 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.9985, + "step": 15814 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 5.6033, + "step": 15815 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15816 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.807, + "step": 15817 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.908, + "step": 15818 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15819 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.5829, + "step": 15820 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.2736, + "step": 15821 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15822 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 7.151, + "step": 15823 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 7.0724, + "step": 15824 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.0171, + "step": 15825 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.9319, + "step": 15826 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.9373, + "step": 15827 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.9384, + "step": 15828 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.4907, + "step": 15829 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.6538, + "step": 15830 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 8.2479, + "step": 15831 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15832 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.0702, + "step": 15833 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.0566, + "step": 15834 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.8426, + "step": 15835 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 2.209, + "step": 15836 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.7256, + "step": 15837 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 5.115, + "step": 15838 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15839 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15840 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 6.4728, + "step": 15841 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15842 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.942, + "step": 15843 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 4.2687, + "step": 15844 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15845 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 1.9992, + "step": 15846 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 3.6529, + "step": 15847 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15848 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15849 + }, + { + "epoch": 1.32, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15850 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 1.0789, + "step": 15851 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15852 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15853 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 3.457, + "step": 15854 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.936, + "step": 15855 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 8.4037, + "step": 15856 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 7.1586, + "step": 15857 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15858 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 3.5668, + "step": 15859 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15860 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15861 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 3.8081, + "step": 15862 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 1.6728, + "step": 15863 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15864 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 7.1724, + "step": 15865 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 4.2369, + "step": 15866 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 2.6545, + "step": 15867 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 1.0154, + "step": 15868 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 1.6925, + "step": 15869 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 6.3713, + "step": 15870 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 9.4506, + "step": 15871 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15872 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 6.9087, + "step": 15873 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15874 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15875 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 5.0666, + "step": 15876 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 3.8514, + "step": 15877 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15878 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15879 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 2.8884, + "step": 15880 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15881 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15882 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 2.6809, + "step": 15883 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15884 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 5.9013, + "step": 15885 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15886 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 2.0532, + "step": 15887 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 3.6975, + "step": 15888 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 1.5149, + "step": 15889 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 2.5565, + "step": 15890 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15891 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 7.8188, + "step": 15892 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.8603, + "step": 15893 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 1.746, + "step": 15894 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15895 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15896 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 1.766, + "step": 15897 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993869646390059, + "loss": 0.0, + "step": 15898 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 4.6504, + "step": 15899 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 2.4272, + "step": 15900 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 1.3733, + "step": 15901 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 2.5019, + "step": 15902 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15903 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.8135, + "step": 15904 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.9252, + "step": 15905 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15906 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15907 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 3.9862, + "step": 15908 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15909 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 11.2047, + "step": 15910 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15911 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15912 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15913 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.9791, + "step": 15914 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 7.9323, + "step": 15915 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 2.2391, + "step": 15916 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15917 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15918 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 6.956, + "step": 15919 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 5.4023, + "step": 15920 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15921 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15922 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 2.1574, + "step": 15923 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15924 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 5.8367, + "step": 15925 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 7.5112, + "step": 15926 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15927 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 2.0085, + "step": 15928 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 4.0179, + "step": 15929 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15930 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 1.0355, + "step": 15931 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 4.984, + "step": 15932 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 7.4308, + "step": 15933 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 3.3266, + "step": 15934 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15935 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15936 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15937 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15938 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15939 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15940 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15941 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.9273, + "step": 15942 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 7.7353, + "step": 15943 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15944 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15945 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 1.7685, + "step": 15946 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15947 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15948 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15949 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15950 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 2.85, + "step": 15951 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15952 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 4.6697, + "step": 15953 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.8434, + "step": 15954 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15955 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15956 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15957 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15958 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 3.5712, + "step": 15959 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 7.5573, + "step": 15960 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15961 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15962 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 1.7479, + "step": 15963 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 3.2012, + "step": 15964 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15965 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 6.1731, + "step": 15966 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15967 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15968 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 1.3059, + "step": 15969 + }, + { + "epoch": 1.33, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15970 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15971 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.7746, + "step": 15972 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15973 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15974 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15975 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.0583, + "step": 15976 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 5.5938, + "step": 15977 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 7.0967, + "step": 15978 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15979 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15980 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15981 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.0185, + "step": 15982 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.9042, + "step": 15983 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15984 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 6.9027, + "step": 15985 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15986 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15987 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.5272, + "step": 15988 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15989 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15990 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15991 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.9973, + "step": 15992 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.0394, + "step": 15993 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.9484, + "step": 15994 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 15995 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.6742, + "step": 15996 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.1136, + "step": 15997 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 8.274, + "step": 15998 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.5073, + "step": 15999 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16000 + }, + { + "epoch": 1.34, + "eval_loss": NaN, + "eval_runtime": 39.898, + "eval_samples_per_second": 62.66, + "eval_steps_per_second": 7.845, + "step": 16000 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 3.9444, + "step": 16001 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16002 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 3.2303, + "step": 16003 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 8.1357, + "step": 16004 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.1232, + "step": 16005 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 6.7859, + "step": 16006 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16007 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.0224, + "step": 16008 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16009 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16010 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16011 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 5.6499, + "step": 16012 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.1943, + "step": 16013 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16014 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.3594, + "step": 16015 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16016 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 10.7847, + "step": 16017 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.8914, + "step": 16018 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 6.0319, + "step": 16019 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 7.4556, + "step": 16020 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 3.505, + "step": 16021 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.0344, + "step": 16022 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16023 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16024 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16025 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16026 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16027 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.0743, + "step": 16028 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.4606, + "step": 16029 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16030 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 5.7509, + "step": 16031 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.2893, + "step": 16032 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 9.2583, + "step": 16033 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16034 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16035 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16036 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 7.0673, + "step": 16037 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.4036, + "step": 16038 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16039 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16040 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16041 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 8.6553, + "step": 16042 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.2218, + "step": 16043 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16044 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16045 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 11.6815, + "step": 16046 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16047 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16048 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16049 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16050 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 3.6527, + "step": 16051 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.9347, + "step": 16052 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.344, + "step": 16053 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 6.9302, + "step": 16054 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16055 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.0827, + "step": 16056 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.5208, + "step": 16057 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.6913, + "step": 16058 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 6.8746, + "step": 16059 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 9.0873, + "step": 16060 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16061 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 8.5362, + "step": 16062 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16063 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16064 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 3.9224, + "step": 16065 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.6019, + "step": 16066 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16067 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.7921, + "step": 16068 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 8.3479, + "step": 16069 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16070 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 8.0278, + "step": 16071 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.8894, + "step": 16072 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16073 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.0385, + "step": 16074 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16075 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16076 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16077 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 2.0788, + "step": 16078 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 3.0098, + "step": 16079 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.0017, + "step": 16080 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16081 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16082 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 3.7214, + "step": 16083 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16084 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 5.8962, + "step": 16085 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.0153, + "step": 16086 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 4.8638, + "step": 16087 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.036, + "step": 16088 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 1.0732, + "step": 16089 + }, + { + "epoch": 1.34, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16090 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 2.0925, + "step": 16091 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.9762, + "step": 16092 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.9889, + "step": 16093 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16094 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.0355, + "step": 16095 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 2.0148, + "step": 16096 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.0144, + "step": 16097 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.9358, + "step": 16098 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16099 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 6.0668, + "step": 16100 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16101 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16102 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16103 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.9996, + "step": 16104 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 8.1367, + "step": 16105 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 6.1297, + "step": 16106 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.2129, + "step": 16107 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 6.1952, + "step": 16108 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16109 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 8.3485, + "step": 16110 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.5451, + "step": 16111 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 2.4459, + "step": 16112 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.8686, + "step": 16113 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 8.8279, + "step": 16114 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16115 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16116 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.5528, + "step": 16117 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16118 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16119 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.5219, + "step": 16120 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.1415, + "step": 16121 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.5045, + "step": 16122 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.9206, + "step": 16123 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 5.5209, + "step": 16124 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16125 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 8.5949, + "step": 16126 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16127 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 7.474, + "step": 16128 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16129 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16130 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.7769, + "step": 16131 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16132 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16133 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.9423, + "step": 16134 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16135 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16136 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.4865, + "step": 16137 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 7.6175, + "step": 16138 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16139 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16140 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16141 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16142 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16143 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.2289, + "step": 16144 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16145 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16146 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16147 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 5.8605, + "step": 16148 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.91, + "step": 16149 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.0065, + "step": 16150 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.6501, + "step": 16151 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16152 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 6.3101, + "step": 16153 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16154 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.5318, + "step": 16155 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16156 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 5.3878, + "step": 16157 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 5.1015, + "step": 16158 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16159 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16160 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16161 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.1272, + "step": 16162 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.9731, + "step": 16163 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16164 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16165 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.0983, + "step": 16166 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 7.6534, + "step": 16167 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 7.1945, + "step": 16168 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16169 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.107, + "step": 16170 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16171 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.9621, + "step": 16172 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.5801, + "step": 16173 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16174 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16175 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 6.5555, + "step": 16176 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16177 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16178 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.0356, + "step": 16179 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.8634, + "step": 16180 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.2134, + "step": 16181 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16182 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.1164, + "step": 16183 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16184 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16185 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 8.0524, + "step": 16186 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.7508, + "step": 16187 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.0373, + "step": 16188 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.4302, + "step": 16189 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16190 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 7.9359, + "step": 16191 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16192 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16193 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 7.5285, + "step": 16194 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16195 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 7.4912, + "step": 16196 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.4253, + "step": 16197 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.1452, + "step": 16198 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16199 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.6404, + "step": 16200 + }, + { + "epoch": 1.35, + "eval_loss": NaN, + "eval_runtime": 39.9121, + "eval_samples_per_second": 62.638, + "eval_steps_per_second": 7.842, + "step": 16200 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.5656, + "step": 16201 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16202 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16203 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 4.1764, + "step": 16204 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16205 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16206 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 3.3951, + "step": 16207 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16208 + }, + { + "epoch": 1.35, + "learning_rate": 0.00019993312341516428, + "loss": 1.9268, + "step": 16209 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 7.1642, + "step": 16210 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16211 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 8.0086, + "step": 16212 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16213 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 5.3456, + "step": 16214 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.2756, + "step": 16215 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.8139, + "step": 16216 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16217 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16218 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.9361, + "step": 16219 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16220 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16221 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16222 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 3.8364, + "step": 16223 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16224 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.1125, + "step": 16225 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.9594, + "step": 16226 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16227 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.215, + "step": 16228 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16229 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16230 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.9058, + "step": 16231 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16232 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16233 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 8.7851, + "step": 16234 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.0074, + "step": 16235 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.8904, + "step": 16236 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.1473, + "step": 16237 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16238 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 8.4153, + "step": 16239 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16240 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.2141, + "step": 16241 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.0001, + "step": 16242 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.7829, + "step": 16243 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.7559, + "step": 16244 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16245 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16246 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16247 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.4008, + "step": 16248 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16249 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16250 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16251 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16252 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.2022, + "step": 16253 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16254 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.0088, + "step": 16255 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16256 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.4019, + "step": 16257 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16258 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16259 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 8.3795, + "step": 16260 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16261 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 9.5665, + "step": 16262 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.9262, + "step": 16263 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16264 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16265 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16266 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.7444, + "step": 16267 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 3.8973, + "step": 16268 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.0582, + "step": 16269 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16270 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.1142, + "step": 16271 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.5706, + "step": 16272 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16273 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16274 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16275 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 8.861, + "step": 16276 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.0508, + "step": 16277 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.9369, + "step": 16278 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16279 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 6.9491, + "step": 16280 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.0337, + "step": 16281 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 7.9791, + "step": 16282 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 1.4768, + "step": 16283 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16284 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16285 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16286 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16287 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.6678, + "step": 16288 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 3.2744, + "step": 16289 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16290 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.4089, + "step": 16291 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16292 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16293 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.866, + "step": 16294 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.2525, + "step": 16295 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16296 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16297 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 7.3303, + "step": 16298 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 3.0001, + "step": 16299 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16300 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16301 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.0865, + "step": 16302 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16303 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.2722, + "step": 16304 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16305 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.7987, + "step": 16306 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.7662, + "step": 16307 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16308 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.2483, + "step": 16309 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 3.8845, + "step": 16310 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16311 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16312 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16313 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16314 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.0575, + "step": 16315 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 7.067, + "step": 16316 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.617, + "step": 16317 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16318 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 3.4776, + "step": 16319 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16320 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16321 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16322 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16323 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 2.195, + "step": 16324 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 6.3871, + "step": 16325 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16326 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 4.0969, + "step": 16327 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 0.0, + "step": 16328 + }, + { + "epoch": 1.36, + "learning_rate": 0.00019993312341516428, + "loss": 8.4266, + "step": 16329 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.1197, + "step": 16330 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16331 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.4224, + "step": 16332 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.5939, + "step": 16333 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.9138, + "step": 16334 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16335 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 6.7042, + "step": 16336 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.2844, + "step": 16337 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16338 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16339 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16340 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16341 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 9.2476, + "step": 16342 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.9422, + "step": 16343 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.0028, + "step": 16344 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.8702, + "step": 16345 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.5776, + "step": 16346 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.9847, + "step": 16347 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16348 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.6156, + "step": 16349 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.6527, + "step": 16350 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16351 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.1726, + "step": 16352 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16353 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.9576, + "step": 16354 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.0009, + "step": 16355 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16356 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16357 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.498, + "step": 16358 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 8.4018, + "step": 16359 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.4317, + "step": 16360 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.8722, + "step": 16361 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16362 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16363 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.648, + "step": 16364 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16365 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16366 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.2396, + "step": 16367 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.8241, + "step": 16368 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.9198, + "step": 16369 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16370 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.865, + "step": 16371 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 14.359, + "step": 16372 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 7.4437, + "step": 16373 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.7752, + "step": 16374 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16375 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.0803, + "step": 16376 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16377 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16378 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16379 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.3593, + "step": 16380 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.1799, + "step": 16381 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.698, + "step": 16382 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 6.7915, + "step": 16383 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.6609, + "step": 16384 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 6.763, + "step": 16385 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16386 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.7439, + "step": 16387 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.5688, + "step": 16388 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16389 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.8681, + "step": 16390 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16391 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.0392, + "step": 16392 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16393 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.936, + "step": 16394 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16395 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.1126, + "step": 16396 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.7767, + "step": 16397 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.748, + "step": 16398 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16399 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.9407, + "step": 16400 + }, + { + "epoch": 1.37, + "eval_loss": NaN, + "eval_runtime": 39.9069, + "eval_samples_per_second": 62.646, + "eval_steps_per_second": 7.843, + "step": 16400 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.4438, + "step": 16401 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16402 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16403 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16404 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.5822, + "step": 16405 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.0135, + "step": 16406 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.878, + "step": 16407 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16408 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16409 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.9364, + "step": 16410 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.7413, + "step": 16411 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.6165, + "step": 16412 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16413 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16414 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16415 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16416 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 7.1987, + "step": 16417 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.7511, + "step": 16418 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16419 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.541, + "step": 16420 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16421 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16422 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16423 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.7384, + "step": 16424 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16425 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.695, + "step": 16426 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.1161, + "step": 16427 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 5.8608, + "step": 16428 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.4743, + "step": 16429 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16430 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16431 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 7.4408, + "step": 16432 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 2.4708, + "step": 16433 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.72, + "step": 16434 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 1.14, + "step": 16435 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16436 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 4.4796, + "step": 16437 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 7.5537, + "step": 16438 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16439 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16440 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 9.5012, + "step": 16441 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16442 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16443 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 7.9457, + "step": 16444 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16445 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 8.0449, + "step": 16446 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 3.7943, + "step": 16447 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16448 + }, + { + "epoch": 1.37, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16449 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16450 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16451 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 1.0322, + "step": 16452 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16453 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 4.1789, + "step": 16454 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16455 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16456 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.8928, + "step": 16457 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16458 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 6.8416, + "step": 16459 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 4.1927, + "step": 16460 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.7787, + "step": 16461 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16462 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16463 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16464 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16465 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 2.52, + "step": 16466 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 3.2309, + "step": 16467 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16468 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 1.9547, + "step": 16469 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16470 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 5.396, + "step": 16471 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 6.6477, + "step": 16472 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16473 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 1.9298, + "step": 16474 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16475 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 6.7503, + "step": 16476 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16477 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 3.3307, + "step": 16478 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 4.7963, + "step": 16479 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16480 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16481 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 3.186, + "step": 16482 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16483 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 8.7115, + "step": 16484 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 2.1934, + "step": 16485 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16486 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 2.4554, + "step": 16487 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 2.258, + "step": 16488 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.9672, + "step": 16489 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16490 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 1.0997, + "step": 16491 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 5.7722, + "step": 16492 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16493 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 1.2351, + "step": 16494 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16495 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16496 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 14.7071, + "step": 16497 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.0, + "step": 16498 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 3.0712, + "step": 16499 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992755036642797, + "loss": 0.8814, + "step": 16500 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 3.6277, + "step": 16501 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16502 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.9124, + "step": 16503 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16504 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16505 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16506 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.8013, + "step": 16507 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 6.9142, + "step": 16508 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 7.9425, + "step": 16509 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16510 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.1858, + "step": 16511 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 8.7853, + "step": 16512 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16513 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16514 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16515 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16516 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16517 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16518 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.0619, + "step": 16519 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 4.0577, + "step": 16520 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16521 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16522 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 5.8692, + "step": 16523 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16524 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16525 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16526 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 4.2073, + "step": 16527 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 8.9108, + "step": 16528 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.6101, + "step": 16529 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16530 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16531 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 4.6312, + "step": 16532 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16533 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16534 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.7896, + "step": 16535 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16536 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16537 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16538 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.7684, + "step": 16539 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16540 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 2.3398, + "step": 16541 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 4.921, + "step": 16542 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16543 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.7779, + "step": 16544 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16545 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16546 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.742, + "step": 16547 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.8683, + "step": 16548 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16549 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16550 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 3.9935, + "step": 16551 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16552 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.7477, + "step": 16553 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16554 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16555 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16556 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 6.6103, + "step": 16557 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 4.294, + "step": 16558 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 5.8218, + "step": 16559 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16560 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16561 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16562 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16563 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16564 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16565 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16566 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 1.0727, + "step": 16567 + }, + { + "epoch": 1.38, + "learning_rate": 0.00019992197731769163, + "loss": 3.6788, + "step": 16568 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16569 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16570 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.994, + "step": 16571 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16572 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16573 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 8.3629, + "step": 16574 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.7907, + "step": 16575 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16576 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 4.3171, + "step": 16577 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.1517, + "step": 16578 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 4.1618, + "step": 16579 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16580 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16581 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.7444, + "step": 16582 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 8.3247, + "step": 16583 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.2184, + "step": 16584 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16585 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 6.7481, + "step": 16586 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16587 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.7923, + "step": 16588 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.9085, + "step": 16589 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 5.6502, + "step": 16590 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 5.623, + "step": 16591 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16592 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16593 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.021, + "step": 16594 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.3691, + "step": 16595 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16596 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16597 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.0187, + "step": 16598 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.9884, + "step": 16599 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16600 + }, + { + "epoch": 1.39, + "eval_loss": NaN, + "eval_runtime": 40.9904, + "eval_samples_per_second": 60.99, + "eval_steps_per_second": 7.636, + "step": 16600 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16601 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.7473, + "step": 16602 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16603 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.9688, + "step": 16604 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.0077, + "step": 16605 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.9513, + "step": 16606 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16607 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.3521, + "step": 16608 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.7415, + "step": 16609 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16610 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16611 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.8434, + "step": 16612 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16613 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16614 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.9848, + "step": 16615 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.9653, + "step": 16616 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.5243, + "step": 16617 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16618 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16619 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 10.7406, + "step": 16620 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 6.8178, + "step": 16621 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 4.3497, + "step": 16622 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.9436, + "step": 16623 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16624 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16625 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16626 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16627 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.6246, + "step": 16628 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.402, + "step": 16629 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16630 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 6.4027, + "step": 16631 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.954, + "step": 16632 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 8.0201, + "step": 16633 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16634 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.032, + "step": 16635 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16636 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16637 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16638 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16639 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 4.7771, + "step": 16640 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16641 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.1239, + "step": 16642 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.7859, + "step": 16643 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16644 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16645 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16646 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16647 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 4.4308, + "step": 16648 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16649 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16650 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16651 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.8516, + "step": 16652 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16653 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.0971, + "step": 16654 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16655 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16656 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16657 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 8.3075, + "step": 16658 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16659 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16660 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.2136, + "step": 16661 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16662 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 7.9517, + "step": 16663 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 6.7996, + "step": 16664 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.0293, + "step": 16665 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16666 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.7758, + "step": 16667 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16668 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16669 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16670 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 5.8433, + "step": 16671 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.0619, + "step": 16672 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 9.102, + "step": 16673 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.6878, + "step": 16674 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 8.037, + "step": 16675 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 1.6716, + "step": 16676 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 9.3463, + "step": 16677 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16678 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 2.7072, + "step": 16679 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16680 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16681 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16682 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16683 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16684 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16685 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16686 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 0.9826, + "step": 16687 + }, + { + "epoch": 1.39, + "learning_rate": 0.00019992197731769163, + "loss": 3.3099, + "step": 16688 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 8.1451, + "step": 16689 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16690 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16691 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16692 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16693 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 7.3248, + "step": 16694 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 6.6525, + "step": 16695 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.6551, + "step": 16696 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16697 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16698 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16699 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.3536, + "step": 16700 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16701 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16702 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16703 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.9375, + "step": 16704 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16705 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16706 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16707 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.5139, + "step": 16708 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 10.3756, + "step": 16709 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16710 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.8654, + "step": 16711 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.1417, + "step": 16712 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16713 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.0092, + "step": 16714 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.5661, + "step": 16715 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.3069, + "step": 16716 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.9684, + "step": 16717 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.7768, + "step": 16718 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 7.7801, + "step": 16719 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 8.6664, + "step": 16720 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16721 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 4.0351, + "step": 16722 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16723 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 6.9916, + "step": 16724 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16725 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16726 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 7.3721, + "step": 16727 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16728 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16729 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16730 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16731 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 4.1895, + "step": 16732 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16733 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16734 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16735 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.3221, + "step": 16736 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16737 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16738 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.9688, + "step": 16739 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16740 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 6.5865, + "step": 16741 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 7.7095, + "step": 16742 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16743 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.1455, + "step": 16744 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.0156, + "step": 16745 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16746 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16747 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16748 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16749 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.4281, + "step": 16750 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16751 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.9044, + "step": 16752 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16753 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.8156, + "step": 16754 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 10.6471, + "step": 16755 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.2093, + "step": 16756 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.5232, + "step": 16757 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.7545, + "step": 16758 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.006, + "step": 16759 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.8423, + "step": 16760 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16761 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16762 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16763 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16764 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16765 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16766 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.8033, + "step": 16767 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16768 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16769 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16770 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16771 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.1379, + "step": 16772 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16773 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16774 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16775 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.0417, + "step": 16776 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 9.3252, + "step": 16777 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 4.0049, + "step": 16778 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16779 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.0703, + "step": 16780 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.1462, + "step": 16781 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16782 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16783 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16784 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 10.1926, + "step": 16785 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16786 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 8.7987, + "step": 16787 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.7718, + "step": 16788 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.5508, + "step": 16789 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.9636, + "step": 16790 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 1.986, + "step": 16791 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16792 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.0626, + "step": 16793 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16794 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.2976, + "step": 16795 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16796 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 10.6566, + "step": 16797 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.5538, + "step": 16798 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 3.5457, + "step": 16799 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 7.6869, + "step": 16800 + }, + { + "epoch": 1.4, + "eval_loss": NaN, + "eval_runtime": 40.6617, + "eval_samples_per_second": 61.483, + "eval_steps_per_second": 7.698, + "step": 16800 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 8.1242, + "step": 16801 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16802 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16803 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16804 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 8.1386, + "step": 16805 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 2.0249, + "step": 16806 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16807 + }, + { + "epoch": 1.4, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16808 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.1078, + "step": 16809 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.9709, + "step": 16810 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16811 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.462, + "step": 16812 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.8125, + "step": 16813 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16814 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16815 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16816 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.0929, + "step": 16817 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.7547, + "step": 16818 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16819 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16820 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.2617, + "step": 16821 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 9.1808, + "step": 16822 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16823 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16824 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 8.1149, + "step": 16825 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16826 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 7.4779, + "step": 16827 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16828 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 6.0097, + "step": 16829 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16830 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.1207, + "step": 16831 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.6003, + "step": 16832 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.0326, + "step": 16833 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16834 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16835 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.0841, + "step": 16836 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 5.4324, + "step": 16837 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 6.3817, + "step": 16838 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.6963, + "step": 16839 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.1498, + "step": 16840 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16841 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.0039, + "step": 16842 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.1141, + "step": 16843 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.8974, + "step": 16844 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.0503, + "step": 16845 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16846 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.2682, + "step": 16847 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 9.5302, + "step": 16848 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 5.0503, + "step": 16849 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16850 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16851 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.9458, + "step": 16852 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16853 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.9782, + "step": 16854 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 7.0424, + "step": 16855 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.7571, + "step": 16856 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.1862, + "step": 16857 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16858 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16859 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 5.7555, + "step": 16860 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16861 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16862 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 5.3327, + "step": 16863 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16864 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.0409, + "step": 16865 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16866 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 6.0459, + "step": 16867 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 8.6462, + "step": 16868 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.6068, + "step": 16869 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.01, + "step": 16870 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 9.5561, + "step": 16871 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16872 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.4504, + "step": 16873 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16874 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16875 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16876 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.4636, + "step": 16877 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.6501, + "step": 16878 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16879 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 6.6206, + "step": 16880 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 7.5497, + "step": 16881 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16882 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16883 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.1592, + "step": 16884 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.4752, + "step": 16885 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 5.4962, + "step": 16886 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16887 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16888 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.0672, + "step": 16889 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16890 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16891 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.1111, + "step": 16892 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16893 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.8446, + "step": 16894 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.6099, + "step": 16895 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.9841, + "step": 16896 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 8.0755, + "step": 16897 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.1938, + "step": 16898 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16899 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16900 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16901 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16902 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16903 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 2.1688, + "step": 16904 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16905 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16906 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 7.6257, + "step": 16907 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.7257, + "step": 16908 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16909 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16910 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16911 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16912 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.243, + "step": 16913 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 1.9262, + "step": 16914 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16915 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16916 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 4.1554, + "step": 16917 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16918 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16919 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16920 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16921 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.2548, + "step": 16922 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 3.1017, + "step": 16923 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 8.0692, + "step": 16924 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16925 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 5.252, + "step": 16926 + }, + { + "epoch": 1.41, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16927 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16928 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.908, + "step": 16929 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16930 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16931 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 4.0177, + "step": 16932 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.8123, + "step": 16933 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16934 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16935 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16936 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.134, + "step": 16937 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 2.0357, + "step": 16938 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.8671, + "step": 16939 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16940 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16941 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16942 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16943 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.6196, + "step": 16944 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16945 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16946 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 5.2431, + "step": 16947 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 3.2271, + "step": 16948 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16949 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 8.186, + "step": 16950 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16951 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16952 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16953 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16954 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 5.9774, + "step": 16955 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 5.0973, + "step": 16956 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16957 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16958 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16959 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.9624, + "step": 16960 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.8842, + "step": 16961 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.927, + "step": 16962 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.9645, + "step": 16963 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.0716, + "step": 16964 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 2.0465, + "step": 16965 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 7.6945, + "step": 16966 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16967 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16968 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 8.459, + "step": 16969 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16970 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.8856, + "step": 16971 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.8374, + "step": 16972 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16973 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.223, + "step": 16974 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 5.3494, + "step": 16975 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 5.3943, + "step": 16976 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 4.2779, + "step": 16977 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 4.5928, + "step": 16978 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16979 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 5.6362, + "step": 16980 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16981 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16982 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 2.0173, + "step": 16983 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16984 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16985 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16986 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 3.1602, + "step": 16987 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.8998, + "step": 16988 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16989 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 7.8324, + "step": 16990 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 4.6311, + "step": 16991 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16992 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16993 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16994 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.7118, + "step": 16995 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16996 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 6.4678, + "step": 16997 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16998 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 16999 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17000 + }, + { + "epoch": 1.42, + "eval_loss": NaN, + "eval_runtime": 41.1705, + "eval_samples_per_second": 60.723, + "eval_steps_per_second": 7.603, + "step": 17000 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 8.9443, + "step": 17001 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17002 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 9.4875, + "step": 17003 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17004 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 9.0717, + "step": 17005 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 4.2662, + "step": 17006 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17007 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 4.9741, + "step": 17008 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.8693, + "step": 17009 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 9.4221, + "step": 17010 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17011 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 2.3056, + "step": 17012 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17013 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 7.0443, + "step": 17014 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.2379, + "step": 17015 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.8737, + "step": 17016 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 3.684, + "step": 17017 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.1928, + "step": 17018 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17019 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17020 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17021 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17022 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.9663, + "step": 17023 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.9996, + "step": 17024 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17025 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17026 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17027 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17028 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 2.0977, + "step": 17029 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.1241, + "step": 17030 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17031 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.8425, + "step": 17032 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17033 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 9.4727, + "step": 17034 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17035 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17036 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 7.1308, + "step": 17037 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17038 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 4.4139, + "step": 17039 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 3.466, + "step": 17040 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 1.0255, + "step": 17041 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17042 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17043 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 5.7551, + "step": 17044 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17045 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 3.4793, + "step": 17046 + }, + { + "epoch": 1.42, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17047 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17048 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 3.0831, + "step": 17049 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17050 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 8.6365, + "step": 17051 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.6023, + "step": 17052 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17053 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17054 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17055 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17056 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.9202, + "step": 17057 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17058 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.0909, + "step": 17059 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17060 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.1348, + "step": 17061 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 2.1545, + "step": 17062 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.6401, + "step": 17063 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 2.7261, + "step": 17064 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.9176, + "step": 17065 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 2.5284, + "step": 17066 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.4887, + "step": 17067 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17068 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17069 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.6218, + "step": 17070 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.7761, + "step": 17071 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.366, + "step": 17072 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17073 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.1603, + "step": 17074 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17075 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.6905, + "step": 17076 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 3.9659, + "step": 17077 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17078 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.8506, + "step": 17079 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17080 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.9244, + "step": 17081 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17082 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.0401, + "step": 17083 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 5.6946, + "step": 17084 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17085 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 3.0819, + "step": 17086 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 6.9048, + "step": 17087 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 6.8839, + "step": 17088 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17089 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17090 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 3.2701, + "step": 17091 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.9402, + "step": 17092 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.8837, + "step": 17093 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17094 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17095 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.1879, + "step": 17096 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 2.0512, + "step": 17097 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.9108, + "step": 17098 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17099 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.7621, + "step": 17100 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.9333, + "step": 17101 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.3722, + "step": 17102 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17103 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17104 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17105 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17106 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17107 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17108 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17109 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 6.687, + "step": 17110 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 8.555, + "step": 17111 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 3.9526, + "step": 17112 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17113 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.7005, + "step": 17114 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17115 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.7034, + "step": 17116 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.8119, + "step": 17117 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.6156, + "step": 17118 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17119 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.1677, + "step": 17120 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.1271, + "step": 17121 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 3.8077, + "step": 17122 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 5.5099, + "step": 17123 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17124 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 5.1437, + "step": 17125 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.3436, + "step": 17126 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17127 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17128 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 6.2585, + "step": 17129 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 8.2708, + "step": 17130 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.603, + "step": 17131 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17132 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.0238, + "step": 17133 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17134 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.9956, + "step": 17135 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.8776, + "step": 17136 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.9603, + "step": 17137 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 6.3635, + "step": 17138 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17139 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.8705, + "step": 17140 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17141 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 5.4705, + "step": 17142 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17143 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 9.8327, + "step": 17144 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17145 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17146 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.9705, + "step": 17147 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17148 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17149 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17150 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.0927, + "step": 17151 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 6.5109, + "step": 17152 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17153 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.1116, + "step": 17154 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17155 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.8238, + "step": 17156 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17157 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.909, + "step": 17158 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 1.0913, + "step": 17159 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.6752, + "step": 17160 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17161 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 7.4225, + "step": 17162 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 8.2336, + "step": 17163 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 2.1517, + "step": 17164 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17165 + }, + { + "epoch": 1.43, + "learning_rate": 0.00019992197731769163, + "loss": 4.7492, + "step": 17166 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 5.6026, + "step": 17167 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.9827, + "step": 17168 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 9.0695, + "step": 17169 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17170 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.2085, + "step": 17171 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.251, + "step": 17172 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.7085, + "step": 17173 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17174 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17175 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17176 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 9.8665, + "step": 17177 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17178 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17179 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.362, + "step": 17180 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.7238, + "step": 17181 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17182 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.9013, + "step": 17183 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 7.8017, + "step": 17184 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 7.655, + "step": 17185 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.6235, + "step": 17186 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.4195, + "step": 17187 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17188 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.8972, + "step": 17189 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.9899, + "step": 17190 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17191 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 8.4897, + "step": 17192 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.8835, + "step": 17193 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17194 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.0034, + "step": 17195 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17196 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.8909, + "step": 17197 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.0058, + "step": 17198 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17199 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.747, + "step": 17200 + }, + { + "epoch": 1.44, + "eval_loss": NaN, + "eval_runtime": 40.276, + "eval_samples_per_second": 62.072, + "eval_steps_per_second": 7.771, + "step": 17200 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 9.0176, + "step": 17201 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17202 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.0779, + "step": 17203 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17204 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.1036, + "step": 17205 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.8682, + "step": 17206 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.6654, + "step": 17207 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17208 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17209 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17210 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 7.72, + "step": 17211 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 8.044, + "step": 17212 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17213 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17214 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.1143, + "step": 17215 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.0243, + "step": 17216 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17217 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 8.2464, + "step": 17218 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.0361, + "step": 17219 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17220 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.1453, + "step": 17221 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.8011, + "step": 17222 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.6123, + "step": 17223 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17224 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17225 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17226 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17227 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.5857, + "step": 17228 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.6478, + "step": 17229 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17230 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17231 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 7.7036, + "step": 17232 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17233 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17234 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17235 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 5.7208, + "step": 17236 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 6.3127, + "step": 17237 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17238 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.8102, + "step": 17239 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.4376, + "step": 17240 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.0199, + "step": 17241 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.0141, + "step": 17242 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.5217, + "step": 17243 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17244 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 6.3887, + "step": 17245 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 6.8701, + "step": 17246 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.1, + "step": 17247 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17248 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17249 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17250 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.9308, + "step": 17251 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17252 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 7.0177, + "step": 17253 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.0386, + "step": 17254 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17255 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 5.8183, + "step": 17256 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.4143, + "step": 17257 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.6087, + "step": 17258 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17259 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 6.5079, + "step": 17260 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17261 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17262 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17263 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.044, + "step": 17264 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.8855, + "step": 17265 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 8.5376, + "step": 17266 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17267 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17268 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.3765, + "step": 17269 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.7796, + "step": 17270 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.7747, + "step": 17271 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17272 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.7461, + "step": 17273 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17274 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 4.3045, + "step": 17275 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 10.6105, + "step": 17276 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17277 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17278 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17279 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 8.9975, + "step": 17280 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.5534, + "step": 17281 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 5.0636, + "step": 17282 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17283 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 2.6823, + "step": 17284 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 1.5349, + "step": 17285 + }, + { + "epoch": 1.44, + "learning_rate": 0.00019992197731769163, + "loss": 3.3797, + "step": 17286 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17287 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.124, + "step": 17288 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.9914, + "step": 17289 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.8972, + "step": 17290 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 5.9941, + "step": 17291 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17292 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17293 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.1495, + "step": 17294 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 5.4172, + "step": 17295 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.6677, + "step": 17296 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17297 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.1984, + "step": 17298 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17299 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.1033, + "step": 17300 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.1126, + "step": 17301 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 4.0295, + "step": 17302 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 6.5137, + "step": 17303 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 6.0003, + "step": 17304 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.8046, + "step": 17305 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17306 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17307 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.8725, + "step": 17308 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.7098, + "step": 17309 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17310 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.25, + "step": 17311 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 8.0435, + "step": 17312 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17313 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.1036, + "step": 17314 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17315 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.0391, + "step": 17316 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17317 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.6171, + "step": 17318 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 5.6268, + "step": 17319 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.6147, + "step": 17320 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17321 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.3878, + "step": 17322 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 4.0886, + "step": 17323 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17324 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17325 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17326 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17327 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17328 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 6.3842, + "step": 17329 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17330 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.2004, + "step": 17331 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.8039, + "step": 17332 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 4.1209, + "step": 17333 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.773, + "step": 17334 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17335 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.6928, + "step": 17336 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 9.1037, + "step": 17337 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.8929, + "step": 17338 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17339 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17340 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.8753, + "step": 17341 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.0009, + "step": 17342 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.9658, + "step": 17343 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17344 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17345 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17346 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17347 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17348 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.9394, + "step": 17349 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17350 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17351 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17352 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17353 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.5057, + "step": 17354 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.1021, + "step": 17355 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17356 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.8156, + "step": 17357 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17358 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.9396, + "step": 17359 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 5.1673, + "step": 17360 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.8201, + "step": 17361 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17362 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.0832, + "step": 17363 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17364 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17365 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 6.6481, + "step": 17366 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17367 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.0704, + "step": 17368 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17369 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17370 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 9.4461, + "step": 17371 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 6.1816, + "step": 17372 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 6.1297, + "step": 17373 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.1793, + "step": 17374 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17375 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17376 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17377 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17378 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17379 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17380 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17381 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17382 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 9.5088, + "step": 17383 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17384 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.9475, + "step": 17385 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.913, + "step": 17386 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.9278, + "step": 17387 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.0049, + "step": 17388 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.8162, + "step": 17389 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 1.1627, + "step": 17390 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17391 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17392 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 7.5739, + "step": 17393 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.6705, + "step": 17394 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17395 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 9.3812, + "step": 17396 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17397 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 6.4775, + "step": 17398 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.2645, + "step": 17399 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 4.25, + "step": 17400 + }, + { + "epoch": 1.45, + "eval_loss": NaN, + "eval_runtime": 40.3229, + "eval_samples_per_second": 62.0, + "eval_steps_per_second": 7.762, + "step": 17400 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 8.0205, + "step": 17401 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 4.5333, + "step": 17402 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.6859, + "step": 17403 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 3.503, + "step": 17404 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17405 + }, + { + "epoch": 1.45, + "learning_rate": 0.00019992197731769163, + "loss": 2.1962, + "step": 17406 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.9716, + "step": 17407 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.8001, + "step": 17408 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.9155, + "step": 17409 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.8124, + "step": 17410 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17411 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 2.0119, + "step": 17412 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17413 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.0108, + "step": 17414 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 4.3639, + "step": 17415 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.9549, + "step": 17416 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.1092, + "step": 17417 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17418 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.0903, + "step": 17419 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.4132, + "step": 17420 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17421 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 4.9077, + "step": 17422 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.9725, + "step": 17423 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 10.1125, + "step": 17424 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17425 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17426 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.3627, + "step": 17427 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17428 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 7.8043, + "step": 17429 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17430 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 9.5714, + "step": 17431 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.9548, + "step": 17432 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 5.1697, + "step": 17433 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.706, + "step": 17434 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17435 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17436 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.0436, + "step": 17437 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17438 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 5.0422, + "step": 17439 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17440 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.8318, + "step": 17441 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17442 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17443 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 7.0509, + "step": 17444 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 6.7523, + "step": 17445 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 4.614, + "step": 17446 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.4043, + "step": 17447 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17448 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.6013, + "step": 17449 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17450 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.8238, + "step": 17451 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17452 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 8.3351, + "step": 17453 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 2.1895, + "step": 17454 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17455 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17456 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 5.8459, + "step": 17457 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17458 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 1.4572, + "step": 17459 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.3256, + "step": 17460 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17461 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 4.3654, + "step": 17462 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17463 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17464 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.8122, + "step": 17465 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17466 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 8.0536, + "step": 17467 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 6.7615, + "step": 17468 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 3.5799, + "step": 17469 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019992197731769163, + "loss": 0.0, + "step": 17470 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 3.9116, + "step": 17471 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 3.541, + "step": 17472 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 8.4991, + "step": 17473 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17474 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17475 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17476 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 7.5552, + "step": 17477 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17478 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 3.7879, + "step": 17479 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17480 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.1475, + "step": 17481 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 5.3566, + "step": 17482 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 7.2631, + "step": 17483 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 4.6024, + "step": 17484 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 4.5712, + "step": 17485 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17486 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.8386, + "step": 17487 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17488 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17489 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17490 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17491 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.0475, + "step": 17492 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17493 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17494 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.8075, + "step": 17495 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17496 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 4.4894, + "step": 17497 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17498 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17499 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 4.1588, + "step": 17500 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17501 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17502 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17503 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 2.2377, + "step": 17504 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17505 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.8997, + "step": 17506 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17507 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.7063, + "step": 17508 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.9994, + "step": 17509 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.0465, + "step": 17510 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17511 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17512 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17513 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 5.1039, + "step": 17514 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.3256, + "step": 17515 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17516 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17517 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 1.8564, + "step": 17518 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17519 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 2.0736, + "step": 17520 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17521 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17522 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 6.0406, + "step": 17523 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 8.376, + "step": 17524 + }, + { + "epoch": 1.46, + "learning_rate": 0.00019991640426895535, + "loss": 6.2802, + "step": 17525 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17526 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17527 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17528 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 2.1595, + "step": 17529 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17530 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17531 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 7.8976, + "step": 17532 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17533 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17534 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17535 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17536 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17537 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17538 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.6545, + "step": 17539 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.7622, + "step": 17540 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 3.7715, + "step": 17541 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17542 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17543 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 5.8638, + "step": 17544 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17545 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17546 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17547 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17548 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 5.9869, + "step": 17549 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17550 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17551 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17552 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17553 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.2847, + "step": 17554 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.6251, + "step": 17555 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 9.4046, + "step": 17556 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.581, + "step": 17557 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17558 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.0318, + "step": 17559 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17560 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 10.1867, + "step": 17561 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17562 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 5.7771, + "step": 17563 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17564 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17565 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17566 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17567 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17568 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.4654, + "step": 17569 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17570 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.2601, + "step": 17571 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17572 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 6.6699, + "step": 17573 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17574 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 2.0259, + "step": 17575 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 6.6158, + "step": 17576 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 6.3597, + "step": 17577 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 3.9332, + "step": 17578 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17579 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 5.5184, + "step": 17580 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 5.1023, + "step": 17581 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.9532, + "step": 17582 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17583 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 10.7562, + "step": 17584 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17585 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17586 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 2.0572, + "step": 17587 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17588 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17589 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.8642, + "step": 17590 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17591 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 3.7735, + "step": 17592 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.0461, + "step": 17593 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17594 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.1144, + "step": 17595 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 2.8293, + "step": 17596 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17597 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 3.9087, + "step": 17598 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17599 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.8696, + "step": 17600 + }, + { + "epoch": 1.47, + "eval_loss": NaN, + "eval_runtime": 40.6112, + "eval_samples_per_second": 61.559, + "eval_steps_per_second": 7.707, + "step": 17600 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 6.9856, + "step": 17601 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 5.5602, + "step": 17602 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17603 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17604 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17605 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.1314, + "step": 17606 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.6131, + "step": 17607 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 4.2292, + "step": 17608 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17609 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17610 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17611 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.8633, + "step": 17612 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.7158, + "step": 17613 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17614 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17615 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 9.7097, + "step": 17616 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.1752, + "step": 17617 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.0763, + "step": 17618 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17619 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17620 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17621 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17622 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17623 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17624 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17625 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.1275, + "step": 17626 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17627 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 1.0596, + "step": 17628 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 3.9896, + "step": 17629 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 3.2854, + "step": 17630 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17631 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17632 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17633 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17634 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17635 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 6.8036, + "step": 17636 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.933, + "step": 17637 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17638 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17639 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17640 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17641 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 6.4153, + "step": 17642 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17643 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17644 + }, + { + "epoch": 1.47, + "learning_rate": 0.00019991640426895535, + "loss": 7.6465, + "step": 17645 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17646 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.4845, + "step": 17647 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 7.3508, + "step": 17648 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.9715, + "step": 17649 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.9784, + "step": 17650 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.0103, + "step": 17651 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17652 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17653 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.1358, + "step": 17654 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 6.2524, + "step": 17655 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.262, + "step": 17656 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 9.2513, + "step": 17657 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17658 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17659 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17660 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 9.0003, + "step": 17661 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17662 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.034, + "step": 17663 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17664 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 5.4845, + "step": 17665 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17666 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.0804, + "step": 17667 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17668 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.6138, + "step": 17669 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 7.0182, + "step": 17670 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17671 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.174, + "step": 17672 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.5774, + "step": 17673 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.2077, + "step": 17674 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.1219, + "step": 17675 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17676 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17677 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17678 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17679 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17680 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17681 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.7657, + "step": 17682 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 4.2449, + "step": 17683 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.2929, + "step": 17684 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 6.3416, + "step": 17685 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.6361, + "step": 17686 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17687 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 6.9664, + "step": 17688 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17689 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17690 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17691 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 6.1118, + "step": 17692 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.555, + "step": 17693 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 7.8042, + "step": 17694 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17695 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17696 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17697 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 7.1606, + "step": 17698 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.7709, + "step": 17699 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 8.0022, + "step": 17700 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.3025, + "step": 17701 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.0043, + "step": 17702 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 9.3241, + "step": 17703 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17704 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.3436, + "step": 17705 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17706 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17707 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17708 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 8.2777, + "step": 17709 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17710 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 8.675, + "step": 17711 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 8.0961, + "step": 17712 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17713 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17714 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17715 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 6.9603, + "step": 17716 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.9241, + "step": 17717 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.8569, + "step": 17718 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 4.5026, + "step": 17719 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 7.1646, + "step": 17720 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17721 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17722 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 5.6619, + "step": 17723 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.8737, + "step": 17724 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 9.0062, + "step": 17725 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.9453, + "step": 17726 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.9926, + "step": 17727 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17728 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.8553, + "step": 17729 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 6.6158, + "step": 17730 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17731 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.3281, + "step": 17732 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.9192, + "step": 17733 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.764, + "step": 17734 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17735 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17736 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.7627, + "step": 17737 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.9154, + "step": 17738 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 6.7504, + "step": 17739 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17740 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17741 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17742 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.2547, + "step": 17743 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.1575, + "step": 17744 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 7.8795, + "step": 17745 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 5.1338, + "step": 17746 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17747 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.6725, + "step": 17748 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 8.6731, + "step": 17749 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.3889, + "step": 17750 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 5.2378, + "step": 17751 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.3193, + "step": 17752 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17753 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 9.0375, + "step": 17754 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 2.3304, + "step": 17755 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17756 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17757 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 3.3756, + "step": 17758 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.759, + "step": 17759 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 4.299, + "step": 17760 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 1.8009, + "step": 17761 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 8.4292, + "step": 17762 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17763 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 9.0046, + "step": 17764 + }, + { + "epoch": 1.48, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17765 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.0272, + "step": 17766 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17767 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17768 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17769 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17770 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.4143, + "step": 17771 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.7647, + "step": 17772 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17773 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17774 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17775 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.8326, + "step": 17776 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 10.2493, + "step": 17777 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.1665, + "step": 17778 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.2229, + "step": 17779 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 7.2113, + "step": 17780 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17781 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17782 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.9421, + "step": 17783 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17784 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17785 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 5.995, + "step": 17786 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 7.4681, + "step": 17787 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.6026, + "step": 17788 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 5.3389, + "step": 17789 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.2008, + "step": 17790 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.6065, + "step": 17791 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.778, + "step": 17792 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17793 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 7.5612, + "step": 17794 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 7.4864, + "step": 17795 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17796 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17797 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.8403, + "step": 17798 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 7.2722, + "step": 17799 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.9468, + "step": 17800 + }, + { + "epoch": 1.49, + "eval_loss": NaN, + "eval_runtime": 40.1554, + "eval_samples_per_second": 62.258, + "eval_steps_per_second": 7.795, + "step": 17800 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.9368, + "step": 17801 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.6165, + "step": 17802 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.8797, + "step": 17803 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.0294, + "step": 17804 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.1251, + "step": 17805 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.9628, + "step": 17806 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 8.7565, + "step": 17807 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17808 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17809 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.1008, + "step": 17810 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.6844, + "step": 17811 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 6.3715, + "step": 17812 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17813 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.0793, + "step": 17814 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17815 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17816 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.5718, + "step": 17817 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17818 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.8041, + "step": 17819 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.0798, + "step": 17820 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17821 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.4345, + "step": 17822 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 6.0365, + "step": 17823 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17824 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17825 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 5.875, + "step": 17826 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.7422, + "step": 17827 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.9138, + "step": 17828 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.5128, + "step": 17829 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17830 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17831 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17832 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17833 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.0633, + "step": 17834 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.1521, + "step": 17835 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.6904, + "step": 17836 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.943, + "step": 17837 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.3755, + "step": 17838 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.9957, + "step": 17839 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17840 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17841 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.5975, + "step": 17842 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.9027, + "step": 17843 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.033, + "step": 17844 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.7541, + "step": 17845 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 6.3036, + "step": 17846 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17847 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 8.2794, + "step": 17848 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.7866, + "step": 17849 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17850 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.9146, + "step": 17851 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17852 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.5401, + "step": 17853 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 5.9945, + "step": 17854 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 7.6014, + "step": 17855 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.9926, + "step": 17856 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17857 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17858 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17859 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 5.7517, + "step": 17860 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.9365, + "step": 17861 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.9642, + "step": 17862 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.0182, + "step": 17863 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.0718, + "step": 17864 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.3935, + "step": 17865 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17866 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.3358, + "step": 17867 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17868 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 2.4483, + "step": 17869 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17870 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 6.0775, + "step": 17871 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.7244, + "step": 17872 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 3.9864, + "step": 17873 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.3115, + "step": 17874 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.9148, + "step": 17875 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 1.0201, + "step": 17876 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17877 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 4.5785, + "step": 17878 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17879 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17880 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 5.3122, + "step": 17881 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17882 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17883 + }, + { + "epoch": 1.49, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17884 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17885 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17886 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 7.9584, + "step": 17887 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17888 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 9.743, + "step": 17889 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17890 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.912, + "step": 17891 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 8.4867, + "step": 17892 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17893 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.9923, + "step": 17894 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17895 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17896 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17897 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.9926, + "step": 17898 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.941, + "step": 17899 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17900 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.6622, + "step": 17901 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 3.8432, + "step": 17902 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.427, + "step": 17903 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17904 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 1.0362, + "step": 17905 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17906 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17907 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17908 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17909 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 1.9668, + "step": 17910 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17911 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.6782, + "step": 17912 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17913 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17914 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17915 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.2283, + "step": 17916 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.3333, + "step": 17917 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17918 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17919 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17920 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17921 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17922 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.0044, + "step": 17923 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.7189, + "step": 17924 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 3.579, + "step": 17925 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 7.106, + "step": 17926 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17927 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 3.1267, + "step": 17928 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 1.9327, + "step": 17929 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17930 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17931 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17932 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 7.6166, + "step": 17933 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17934 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17935 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.7277, + "step": 17936 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17937 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17938 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 3.7709, + "step": 17939 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17940 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17941 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17942 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17943 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 3.3376, + "step": 17944 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17945 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17946 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 5.1289, + "step": 17947 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 8.9765, + "step": 17948 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17949 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17950 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.0112, + "step": 17951 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 6.5707, + "step": 17952 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.9245, + "step": 17953 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.7892, + "step": 17954 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 14.002, + "step": 17955 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.3688, + "step": 17956 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.0475, + "step": 17957 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 1.9199, + "step": 17958 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17959 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17960 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17961 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17962 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17963 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.9288, + "step": 17964 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.6462, + "step": 17965 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 7.514, + "step": 17966 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.0584, + "step": 17967 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 1.9136, + "step": 17968 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.3637, + "step": 17969 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.7543, + "step": 17970 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 6.7293, + "step": 17971 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17972 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 5.7782, + "step": 17973 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17974 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 7.6035, + "step": 17975 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.1506, + "step": 17976 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17977 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 3.8186, + "step": 17978 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17979 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17980 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 8.2887, + "step": 17981 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17982 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17983 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17984 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 7.3246, + "step": 17985 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17986 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.0413, + "step": 17987 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.8492, + "step": 17988 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17989 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 1.0743, + "step": 17990 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17991 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17992 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 6.724, + "step": 17993 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 5.1987, + "step": 17994 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17995 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17996 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 8.8312, + "step": 17997 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 2.0438, + "step": 17998 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 17999 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18000 + }, + { + "epoch": 1.5, + "eval_loss": NaN, + "eval_runtime": 40.6283, + "eval_samples_per_second": 61.534, + "eval_steps_per_second": 7.704, + "step": 18000 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 5.8455, + "step": 18001 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 3.0104, + "step": 18002 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 4.0989, + "step": 18003 + }, + { + "epoch": 1.5, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18004 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18005 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.4287, + "step": 18006 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18007 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.9775, + "step": 18008 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18009 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18010 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.4158, + "step": 18011 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18012 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18013 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18014 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18015 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.1041, + "step": 18016 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 8.1411, + "step": 18017 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18018 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18019 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 4.7641, + "step": 18020 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 8.8257, + "step": 18021 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 2.096, + "step": 18022 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18023 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18024 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18025 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18026 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18027 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18028 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18029 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18030 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 6.5289, + "step": 18031 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18032 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 8.25, + "step": 18033 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 4.9197, + "step": 18034 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.5875, + "step": 18035 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18036 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18037 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18038 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 5.032, + "step": 18039 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.2179, + "step": 18040 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 5.3214, + "step": 18041 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18042 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18043 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.6616, + "step": 18044 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.0235, + "step": 18045 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18046 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.0153, + "step": 18047 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.3998, + "step": 18048 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18049 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.7874, + "step": 18050 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18051 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.8481, + "step": 18052 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18053 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 2.0236, + "step": 18054 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 2.9399, + "step": 18055 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.2155, + "step": 18056 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18057 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18058 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.9652, + "step": 18059 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.0092, + "step": 18060 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 6.8715, + "step": 18061 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18062 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 2.1771, + "step": 18063 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18064 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.0241, + "step": 18065 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18066 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18067 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.0376, + "step": 18068 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.6342, + "step": 18069 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.1276, + "step": 18070 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 8.1132, + "step": 18071 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.6774, + "step": 18072 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18073 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18074 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 4.3596, + "step": 18075 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18076 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18077 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.0296, + "step": 18078 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18079 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.6541, + "step": 18080 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18081 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18082 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.6616, + "step": 18083 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18084 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18085 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18086 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18087 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18088 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.9294, + "step": 18089 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 4.0818, + "step": 18090 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 2.0246, + "step": 18091 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.2932, + "step": 18092 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18093 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18094 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18095 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18096 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 2.6089, + "step": 18097 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18098 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18099 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.4827, + "step": 18100 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.6254, + "step": 18101 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18102 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18103 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.8603, + "step": 18104 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18105 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 2.6374, + "step": 18106 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18107 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18108 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 4.6589, + "step": 18109 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18110 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18111 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.8471, + "step": 18112 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18113 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.7827, + "step": 18114 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18115 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18116 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 1.8212, + "step": 18117 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 3.0708, + "step": 18118 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18119 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18120 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 7.8377, + "step": 18121 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18122 + }, + { + "epoch": 1.51, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18123 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18124 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.5793, + "step": 18125 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 4.4177, + "step": 18126 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18127 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.3716, + "step": 18128 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18129 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18130 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.8645, + "step": 18131 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18132 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18133 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 5.0913, + "step": 18134 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18135 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.9792, + "step": 18136 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18137 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 6.5559, + "step": 18138 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18139 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 10.291, + "step": 18140 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18141 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18142 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18143 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 7.3332, + "step": 18144 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18145 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 8.5087, + "step": 18146 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18147 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 7.988, + "step": 18148 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.8978, + "step": 18149 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18150 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18151 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18152 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18153 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18154 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18155 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18156 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18157 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 6.3708, + "step": 18158 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18159 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.6886, + "step": 18160 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18161 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18162 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18163 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18164 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.9769, + "step": 18165 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18166 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18167 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.4705, + "step": 18168 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 8.2998, + "step": 18169 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 6.6285, + "step": 18170 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18171 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18172 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.8743, + "step": 18173 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18174 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18175 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 5.5613, + "step": 18176 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.2876, + "step": 18177 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18178 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.358, + "step": 18179 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 5.9545, + "step": 18180 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18181 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.6653, + "step": 18182 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18183 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.9018, + "step": 18184 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 4.9261, + "step": 18185 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18186 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18187 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 9.825, + "step": 18188 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.9291, + "step": 18189 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.9235, + "step": 18190 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 6.6265, + "step": 18191 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 8.4962, + "step": 18192 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18193 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18194 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 6.1763, + "step": 18195 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.8703, + "step": 18196 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.8163, + "step": 18197 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18198 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18199 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.0548, + "step": 18200 + }, + { + "epoch": 1.52, + "eval_loss": NaN, + "eval_runtime": 40.6165, + "eval_samples_per_second": 61.551, + "eval_steps_per_second": 7.706, + "step": 18200 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.0663, + "step": 18201 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18202 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.0533, + "step": 18203 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18204 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18205 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18206 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18207 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.5427, + "step": 18208 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.9127, + "step": 18209 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18210 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18211 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.9519, + "step": 18212 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18213 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18214 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18215 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18216 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.7673, + "step": 18217 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18218 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.4919, + "step": 18219 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18220 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 4.1014, + "step": 18221 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18222 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18223 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18224 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 7.689, + "step": 18225 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.7716, + "step": 18226 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 5.026, + "step": 18227 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18228 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18229 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18230 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18231 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.7031, + "step": 18232 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18233 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.4293, + "step": 18234 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18235 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 3.7506, + "step": 18236 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 7.0448, + "step": 18237 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 7.1019, + "step": 18238 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18239 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18240 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.0568, + "step": 18241 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 1.9352, + "step": 18242 + }, + { + "epoch": 1.52, + "learning_rate": 0.00019991640426895535, + "loss": 2.0395, + "step": 18243 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18244 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18245 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.757, + "step": 18246 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18247 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.2716, + "step": 18248 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.7958, + "step": 18249 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18250 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.7376, + "step": 18251 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18252 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18253 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18254 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.8017, + "step": 18255 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 9.9856, + "step": 18256 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 6.6994, + "step": 18257 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18258 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.5288, + "step": 18259 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18260 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18261 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.1283, + "step": 18262 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18263 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.3378, + "step": 18264 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.5841, + "step": 18265 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.2033, + "step": 18266 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.934, + "step": 18267 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18268 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 9.1393, + "step": 18269 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.3927, + "step": 18270 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18271 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.3287, + "step": 18272 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.5477, + "step": 18273 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.9899, + "step": 18274 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18275 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18276 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18277 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.7, + "step": 18278 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18279 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18280 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.0783, + "step": 18281 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18282 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 10.6498, + "step": 18283 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18284 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18285 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18286 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.6712, + "step": 18287 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.1598, + "step": 18288 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.9421, + "step": 18289 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.9016, + "step": 18290 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 10.0241, + "step": 18291 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.8913, + "step": 18292 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18293 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18294 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18295 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 6.9615, + "step": 18296 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.3469, + "step": 18297 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.0404, + "step": 18298 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18299 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 7.8113, + "step": 18300 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.0867, + "step": 18301 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.7656, + "step": 18302 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 6.37, + "step": 18303 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.2975, + "step": 18304 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.2661, + "step": 18305 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 7.4212, + "step": 18306 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18307 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18308 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18309 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.3613, + "step": 18310 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18311 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18312 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.8249, + "step": 18313 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.0124, + "step": 18314 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18315 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.8621, + "step": 18316 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.9512, + "step": 18317 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.989, + "step": 18318 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18319 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18320 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18321 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.7701, + "step": 18322 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.6846, + "step": 18323 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18324 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.3582, + "step": 18325 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18326 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18327 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.384, + "step": 18328 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 6.5629, + "step": 18329 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18330 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.9755, + "step": 18331 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 8.8313, + "step": 18332 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18333 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18334 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.8367, + "step": 18335 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18336 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18337 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.3123, + "step": 18338 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18339 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18340 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.5004, + "step": 18341 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 8.6947, + "step": 18342 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.0284, + "step": 18343 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18344 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18345 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18346 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18347 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18348 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.944, + "step": 18349 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 8.4136, + "step": 18350 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18351 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.5871, + "step": 18352 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18353 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 1.7973, + "step": 18354 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 4.9765, + "step": 18355 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 6.5769, + "step": 18356 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18357 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 5.3889, + "step": 18358 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18359 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 7.6384, + "step": 18360 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 2.1041, + "step": 18361 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18362 + }, + { + "epoch": 1.53, + "learning_rate": 0.00019991640426895535, + "loss": 3.4405, + "step": 18363 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 5.4495, + "step": 18364 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 8.3529, + "step": 18365 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18366 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 4.5737, + "step": 18367 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18368 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 5.4604, + "step": 18369 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 3.9734, + "step": 18370 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.9493, + "step": 18371 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.3925, + "step": 18372 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 3.8321, + "step": 18373 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18374 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18375 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 6.3313, + "step": 18376 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 5.3238, + "step": 18377 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18378 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18379 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18380 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 5.3174, + "step": 18381 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18382 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18383 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.2057, + "step": 18384 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18385 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 12.0219, + "step": 18386 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18387 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 8.0454, + "step": 18388 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.2427, + "step": 18389 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 3.0504, + "step": 18390 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.6178, + "step": 18391 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 6.7886, + "step": 18392 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18393 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18394 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18395 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18396 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.7099, + "step": 18397 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18398 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.2116, + "step": 18399 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 5.2101, + "step": 18400 + }, + { + "epoch": 1.54, + "eval_loss": NaN, + "eval_runtime": 40.2395, + "eval_samples_per_second": 62.128, + "eval_steps_per_second": 7.778, + "step": 18400 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18401 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18402 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.9066, + "step": 18403 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18404 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.101, + "step": 18405 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18406 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18407 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18408 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 3.0227, + "step": 18409 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18410 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.8113, + "step": 18411 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.8537, + "step": 18412 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.7947, + "step": 18413 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.9316, + "step": 18414 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.9622, + "step": 18415 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18416 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18417 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18418 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18419 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18420 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18421 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18422 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18423 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18424 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.3824, + "step": 18425 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 6.022, + "step": 18426 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18427 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18428 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18429 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18430 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.1707, + "step": 18431 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18432 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.1541, + "step": 18433 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.8568, + "step": 18434 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.8322, + "step": 18435 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.9977, + "step": 18436 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18437 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18438 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 5.0464, + "step": 18439 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 8.3894, + "step": 18440 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18441 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18442 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 4.7068, + "step": 18443 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 3.3969, + "step": 18444 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.7147, + "step": 18445 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18446 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.7239, + "step": 18447 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18448 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18449 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18450 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18451 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 3.5451, + "step": 18452 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.9245, + "step": 18453 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 7.6934, + "step": 18454 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18455 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 4.6068, + "step": 18456 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18457 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18458 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18459 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18460 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 7.4287, + "step": 18461 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18462 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18463 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18464 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18465 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18466 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.5465, + "step": 18467 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 4.7937, + "step": 18468 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.0641, + "step": 18469 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18470 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18471 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 4.6577, + "step": 18472 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 3.8146, + "step": 18473 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.8347, + "step": 18474 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 1.0077, + "step": 18475 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 5.6214, + "step": 18476 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 2.0863, + "step": 18477 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 4.0853, + "step": 18478 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 8.9005, + "step": 18479 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18480 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18481 + }, + { + "epoch": 1.54, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18482 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.8518, + "step": 18483 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18484 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.2076, + "step": 18485 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 7.2244, + "step": 18486 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18487 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18488 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18489 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18490 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18491 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18492 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18493 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 6.9131, + "step": 18494 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.9288, + "step": 18495 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.1039, + "step": 18496 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18497 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 6.7151, + "step": 18498 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18499 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18500 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 6.5012, + "step": 18501 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18502 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18503 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18504 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18505 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 4.11, + "step": 18506 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18507 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 9.4072, + "step": 18508 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.9791, + "step": 18509 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.7365, + "step": 18510 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18511 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18512 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18513 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18514 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18515 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 8.655, + "step": 18516 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18517 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 5.248, + "step": 18518 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 2.2115, + "step": 18519 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.0055, + "step": 18520 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18521 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18522 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18523 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 3.8217, + "step": 18524 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 8.9059, + "step": 18525 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18526 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18527 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 2.5271, + "step": 18528 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 5.3104, + "step": 18529 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.3995, + "step": 18530 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 3.671, + "step": 18531 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18532 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18533 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18534 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.8603, + "step": 18535 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.9918, + "step": 18536 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 4.2074, + "step": 18537 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.8952, + "step": 18538 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 4.1095, + "step": 18539 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18540 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18541 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 2.641, + "step": 18542 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 6.0443, + "step": 18543 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.8185, + "step": 18544 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18545 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 8.4236, + "step": 18546 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 9.4367, + "step": 18547 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 2.0031, + "step": 18548 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18549 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 3.9742, + "step": 18550 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 10.0787, + "step": 18551 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 4.0998, + "step": 18552 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.3138, + "step": 18553 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.6646, + "step": 18554 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 3.6215, + "step": 18555 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18556 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 3.9025, + "step": 18557 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 7.4177, + "step": 18558 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18559 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 11.9434, + "step": 18560 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 8.1911, + "step": 18561 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18562 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.6687, + "step": 18563 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18564 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18565 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18566 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 8.7825, + "step": 18567 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 9.1555, + "step": 18568 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.9627, + "step": 18569 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18570 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 5.1725, + "step": 18571 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18572 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18573 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18574 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18575 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 7.2834, + "step": 18576 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.0877, + "step": 18577 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18578 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 7.1367, + "step": 18579 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 7.9174, + "step": 18580 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 5.6435, + "step": 18581 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.0128, + "step": 18582 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 5.2158, + "step": 18583 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 2.025, + "step": 18584 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.9843, + "step": 18585 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18586 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.3383, + "step": 18587 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18588 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.0303, + "step": 18589 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18590 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 3.5887, + "step": 18591 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 1.8282, + "step": 18592 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18593 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 4.0361, + "step": 18594 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 7.5951, + "step": 18595 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18596 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 2.6496, + "step": 18597 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 3.7415, + "step": 18598 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.6203, + "step": 18599 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 4.6063, + "step": 18600 + }, + { + "epoch": 1.55, + "eval_loss": NaN, + "eval_runtime": 41.279, + "eval_samples_per_second": 60.563, + "eval_steps_per_second": 7.583, + "step": 18600 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18601 + }, + { + "epoch": 1.55, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18602 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18603 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18604 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18605 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 3.4774, + "step": 18606 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18607 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 3.7795, + "step": 18608 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 5.8318, + "step": 18609 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 3.0821, + "step": 18610 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18611 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18612 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18613 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 4.557, + "step": 18614 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18615 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18616 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 1.1447, + "step": 18617 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 2.0133, + "step": 18618 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 1.3275, + "step": 18619 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18620 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18621 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 7.5511, + "step": 18622 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 1.7715, + "step": 18623 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18624 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 1.8121, + "step": 18625 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 4.5266, + "step": 18626 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18627 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18628 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 7.6272, + "step": 18629 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 1.916, + "step": 18630 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18631 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18632 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18633 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18634 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18635 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 7.0756, + "step": 18636 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18637 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 3.8325, + "step": 18638 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 1.0023, + "step": 18639 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18640 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 3.612, + "step": 18641 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18642 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 5.3212, + "step": 18643 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18644 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 6.5825, + "step": 18645 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18646 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 1.9094, + "step": 18647 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.9911, + "step": 18648 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.9184, + "step": 18649 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 4.7945, + "step": 18650 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 8.692, + "step": 18651 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18652 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18653 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.8923, + "step": 18654 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 2.1343, + "step": 18655 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18656 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18657 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 6.1484, + "step": 18658 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18659 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18660 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18661 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 8.635, + "step": 18662 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 6.0459, + "step": 18663 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 2.0913, + "step": 18664 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 2.7949, + "step": 18665 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18666 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 4.3343, + "step": 18667 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.7857, + "step": 18668 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18669 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 7.1614, + "step": 18670 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 6.8077, + "step": 18671 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 4.0303, + "step": 18672 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.5143, + "step": 18673 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18674 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.9457, + "step": 18675 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18676 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 3.5041, + "step": 18677 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 8.3807, + "step": 18678 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.9388, + "step": 18679 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 8.3566, + "step": 18680 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 5.2204, + "step": 18681 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18682 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 4.4962, + "step": 18683 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18684 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18685 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 2.2192, + "step": 18686 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 2.7093, + "step": 18687 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18688 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18689 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 3.3152, + "step": 18690 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18691 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 8.5567, + "step": 18692 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18693 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991640426895535, + "loss": 0.0, + "step": 18694 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 3.8819, + "step": 18695 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 2.1125, + "step": 18696 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 4.1682, + "step": 18697 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18698 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18699 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 7.3725, + "step": 18700 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 8.2115, + "step": 18701 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18702 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.9195, + "step": 18703 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 2.4086, + "step": 18704 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18705 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18706 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18707 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 1.9583, + "step": 18708 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 2.929, + "step": 18709 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 1.722, + "step": 18710 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18711 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18712 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 1.9277, + "step": 18713 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 8.1095, + "step": 18714 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18715 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18716 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18717 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 4.0175, + "step": 18718 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 2.8012, + "step": 18719 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18720 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18721 + }, + { + "epoch": 1.56, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18722 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 2.0753, + "step": 18723 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.6576, + "step": 18724 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 8.5338, + "step": 18725 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.8867, + "step": 18726 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18727 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.339, + "step": 18728 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18729 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18730 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18731 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18732 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 10.9714, + "step": 18733 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.1168, + "step": 18734 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18735 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18736 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.7577, + "step": 18737 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 9.0012, + "step": 18738 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.4614, + "step": 18739 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18740 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18741 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18742 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.945, + "step": 18743 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18744 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 6.557, + "step": 18745 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18746 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18747 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 7.1106, + "step": 18748 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 7.4931, + "step": 18749 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 7.8353, + "step": 18750 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18751 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.3761, + "step": 18752 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18753 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.2128, + "step": 18754 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 2.1585, + "step": 18755 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18756 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.8351, + "step": 18757 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.0193, + "step": 18758 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.5444, + "step": 18759 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18760 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 10.663, + "step": 18761 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18762 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18763 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.1003, + "step": 18764 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18765 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18766 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18767 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18768 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.6405, + "step": 18769 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.1016, + "step": 18770 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.8783, + "step": 18771 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 9.2533, + "step": 18772 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 8.0986, + "step": 18773 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18774 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.008, + "step": 18775 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18776 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18777 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.1701, + "step": 18778 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18779 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18780 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 2.95, + "step": 18781 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18782 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.0097, + "step": 18783 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.5536, + "step": 18784 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18785 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18786 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.6723, + "step": 18787 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.8584, + "step": 18788 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.2952, + "step": 18789 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.9371, + "step": 18790 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 8.2689, + "step": 18791 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.6803, + "step": 18792 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.0443, + "step": 18793 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.9493, + "step": 18794 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.0246, + "step": 18795 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.1782, + "step": 18796 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 9.6681, + "step": 18797 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18798 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 10.2085, + "step": 18799 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 2.3694, + "step": 18800 + }, + { + "epoch": 1.57, + "eval_loss": NaN, + "eval_runtime": 41.9978, + "eval_samples_per_second": 59.527, + "eval_steps_per_second": 7.453, + "step": 18800 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.668, + "step": 18801 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18802 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 10.6247, + "step": 18803 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18804 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 2.0495, + "step": 18805 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 8.0628, + "step": 18806 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18807 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.8019, + "step": 18808 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 5.1644, + "step": 18809 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.3305, + "step": 18810 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.9888, + "step": 18811 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.1601, + "step": 18812 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.9792, + "step": 18813 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 8.8643, + "step": 18814 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18815 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.1093, + "step": 18816 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18817 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18818 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18819 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18820 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 7.6562, + "step": 18821 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18822 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.138, + "step": 18823 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.9192, + "step": 18824 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 2.8925, + "step": 18825 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.9799, + "step": 18826 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.0215, + "step": 18827 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.8362, + "step": 18828 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18829 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 5.5705, + "step": 18830 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18831 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 9.697, + "step": 18832 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18833 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 3.6472, + "step": 18834 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 1.0966, + "step": 18835 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18836 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18837 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.3996, + "step": 18838 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.1022, + "step": 18839 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 2.8961, + "step": 18840 + }, + { + "epoch": 1.57, + "learning_rate": 0.00019991083122021904, + "loss": 4.7998, + "step": 18841 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 4.1793, + "step": 18842 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18843 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18844 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18845 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 6.5781, + "step": 18846 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 4.5267, + "step": 18847 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18848 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 6.0143, + "step": 18849 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18850 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 6.7902, + "step": 18851 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18852 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 1.4849, + "step": 18853 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 2.4518, + "step": 18854 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 5.2129, + "step": 18855 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 3.6508, + "step": 18856 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18857 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18858 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 1.0026, + "step": 18859 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.8392, + "step": 18860 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 1.0858, + "step": 18861 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 6.4978, + "step": 18862 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 8.4549, + "step": 18863 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 6.9634, + "step": 18864 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 4.4162, + "step": 18865 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18866 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.8079, + "step": 18867 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18868 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18869 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 2.179, + "step": 18870 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 2.761, + "step": 18871 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18872 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 3.8992, + "step": 18873 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 10.342, + "step": 18874 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 7.9083, + "step": 18875 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 2.0213, + "step": 18876 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18877 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 2.0941, + "step": 18878 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 1.0606, + "step": 18879 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18880 + }, + { + "epoch": 1.58, + "learning_rate": 0.00019991083122021904, + "loss": 0.0, + "step": 18881 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 4.3821, + "step": 18882 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18883 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18884 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18885 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 8.7219, + "step": 18886 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18887 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18888 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18889 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18890 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 3.7551, + "step": 18891 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18892 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 8.604, + "step": 18893 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18894 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 4.5207, + "step": 18895 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 8.559, + "step": 18896 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 2.1149, + "step": 18897 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18898 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18899 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 1.8162, + "step": 18900 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18901 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 1.911, + "step": 18902 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18903 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18904 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18905 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18906 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18907 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 1.5887, + "step": 18908 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 3.6363, + "step": 18909 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18910 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18911 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18912 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18913 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 1.952, + "step": 18914 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18915 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 2.1271, + "step": 18916 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.7749, + "step": 18917 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.5729, + "step": 18918 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 5.0136, + "step": 18919 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 8.1156, + "step": 18920 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.5688, + "step": 18921 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18922 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 2.2053, + "step": 18923 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 2.934, + "step": 18924 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 4.2457, + "step": 18925 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 4.0455, + "step": 18926 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 3.9757, + "step": 18927 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 1.1614, + "step": 18928 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18929 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 9.345, + "step": 18930 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18931 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18932 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18933 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18934 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 6.996, + "step": 18935 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 8.7794, + "step": 18936 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18937 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18938 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 1.9684, + "step": 18939 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 5.7483, + "step": 18940 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 8.2545, + "step": 18941 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18942 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18943 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.9767, + "step": 18944 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18945 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 6.4467, + "step": 18946 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18947 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18948 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 2.2655, + "step": 18949 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.9708, + "step": 18950 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 2.0958, + "step": 18951 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18952 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18953 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 3.9053, + "step": 18954 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18955 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 3.6394, + "step": 18956 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 5.1268, + "step": 18957 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 4.0747, + "step": 18958 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 4.0808, + "step": 18959 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.9821, + "step": 18960 + }, + { + "epoch": 1.58, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18961 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18962 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18963 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.9754, + "step": 18964 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.3889, + "step": 18965 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 6.9735, + "step": 18966 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.7109, + "step": 18967 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 5.5692, + "step": 18968 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18969 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18970 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18971 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18972 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18973 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.0815, + "step": 18974 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.1091, + "step": 18975 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 8.0473, + "step": 18976 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.9396, + "step": 18977 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18978 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 7.3719, + "step": 18979 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.8688, + "step": 18980 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.8018, + "step": 18981 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18982 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.7084, + "step": 18983 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18984 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.1794, + "step": 18985 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.9559, + "step": 18986 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.0091, + "step": 18987 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.4171, + "step": 18988 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18989 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18990 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.9929, + "step": 18991 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.7575, + "step": 18992 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18993 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.394, + "step": 18994 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18995 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.1527, + "step": 18996 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.0538, + "step": 18997 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 9.5121, + "step": 18998 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 18999 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 6.8966, + "step": 19000 + }, + { + "epoch": 1.59, + "eval_loss": NaN, + "eval_runtime": 43.0269, + "eval_samples_per_second": 58.103, + "eval_steps_per_second": 7.275, + "step": 19000 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 8.5457, + "step": 19001 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.3441, + "step": 19002 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.912, + "step": 19003 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19004 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.8526, + "step": 19005 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.6886, + "step": 19006 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19007 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19008 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19009 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.4422, + "step": 19010 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.0291, + "step": 19011 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 7.5637, + "step": 19012 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19013 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19014 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.0408, + "step": 19015 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.9042, + "step": 19016 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19017 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19018 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.9229, + "step": 19019 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19020 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19021 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.0059, + "step": 19022 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19023 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19024 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19025 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 7.8302, + "step": 19026 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 7.8193, + "step": 19027 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 8.9084, + "step": 19028 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 13.4465, + "step": 19029 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.437, + "step": 19030 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19031 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 5.8713, + "step": 19032 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.5264, + "step": 19033 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.0415, + "step": 19034 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19035 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19036 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 6.7926, + "step": 19037 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.7027, + "step": 19038 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19039 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.0664, + "step": 19040 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.2215, + "step": 19041 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19042 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.2373, + "step": 19043 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.319, + "step": 19044 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 7.2634, + "step": 19045 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.1178, + "step": 19046 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.8788, + "step": 19047 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.1682, + "step": 19048 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 5.5697, + "step": 19049 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.8355, + "step": 19050 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 8.3812, + "step": 19051 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19052 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19053 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 4.053, + "step": 19054 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.6612, + "step": 19055 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 14.876, + "step": 19056 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19057 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 7.404, + "step": 19058 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19059 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19060 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19061 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.9662, + "step": 19062 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19063 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19064 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19065 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19066 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19067 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19068 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19069 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.917, + "step": 19070 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.7486, + "step": 19071 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.4739, + "step": 19072 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19073 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 2.045, + "step": 19074 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.5444, + "step": 19075 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19076 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 1.2178, + "step": 19077 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19078 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19079 + }, + { + "epoch": 1.59, + "learning_rate": 0.0001999052581714827, + "loss": 3.6242, + "step": 19080 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19081 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.0428, + "step": 19082 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19083 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19084 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 4.3931, + "step": 19085 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19086 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 6.5878, + "step": 19087 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19088 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19089 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 5.286, + "step": 19090 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.8354, + "step": 19091 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19092 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 4.1541, + "step": 19093 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19094 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19095 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19096 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19097 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 4.2697, + "step": 19098 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 4.1755, + "step": 19099 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19100 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19101 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.8087, + "step": 19102 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.0243, + "step": 19103 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19104 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19105 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 2.0169, + "step": 19106 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 3.6981, + "step": 19107 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.7466, + "step": 19108 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 9.6143, + "step": 19109 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.5402, + "step": 19110 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19111 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19112 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.9234, + "step": 19113 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.6049, + "step": 19114 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19115 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19116 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19117 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19118 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19119 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19120 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 7.342, + "step": 19121 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19122 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 3.1854, + "step": 19123 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19124 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19125 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19126 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19127 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19128 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.4793, + "step": 19129 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.0656, + "step": 19130 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 5.2825, + "step": 19131 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.9525, + "step": 19132 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.134, + "step": 19133 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.8379, + "step": 19134 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 2.0338, + "step": 19135 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 3.8366, + "step": 19136 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.909, + "step": 19137 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.2516, + "step": 19138 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 3.3463, + "step": 19139 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.1063, + "step": 19140 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19141 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19142 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.0731, + "step": 19143 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19144 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.7315, + "step": 19145 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19146 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19147 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 2.9244, + "step": 19148 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19149 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 2.8752, + "step": 19150 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19151 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19152 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19153 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19154 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19155 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 12.8351, + "step": 19156 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19157 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 0.0, + "step": 19158 + }, + { + "epoch": 1.6, + "learning_rate": 0.0001999052581714827, + "loss": 1.6984, + "step": 19159 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 4.0305, + "step": 19160 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.9028, + "step": 19161 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19162 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 1.9382, + "step": 19163 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 8.1106, + "step": 19164 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 1.4514, + "step": 19165 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 5.5454, + "step": 19166 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.7644, + "step": 19167 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 2.1472, + "step": 19168 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19169 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19170 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19171 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19172 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19173 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 8.0723, + "step": 19174 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 7.2112, + "step": 19175 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 1.5765, + "step": 19176 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 6.8709, + "step": 19177 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19178 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 3.3886, + "step": 19179 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 5.4524, + "step": 19180 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19181 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 3.2136, + "step": 19182 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 4.1306, + "step": 19183 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19184 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19185 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 3.6721, + "step": 19186 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19187 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 1.1103, + "step": 19188 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 1.819, + "step": 19189 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 7.7738, + "step": 19190 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19191 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 5.4509, + "step": 19192 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19193 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19194 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19195 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 1.4737, + "step": 19196 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19197 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19198 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 2.8803, + "step": 19199 + }, + { + "epoch": 1.6, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19200 + }, + { + "epoch": 1.6, + "eval_loss": NaN, + "eval_runtime": 43.3099, + "eval_samples_per_second": 57.724, + "eval_steps_per_second": 7.227, + "step": 19200 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.2448, + "step": 19201 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.6939, + "step": 19202 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 7.866, + "step": 19203 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 7.6337, + "step": 19204 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19205 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19206 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.5609, + "step": 19207 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.4699, + "step": 19208 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19209 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.1214, + "step": 19210 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19211 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.3971, + "step": 19212 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.8986, + "step": 19213 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19214 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.9886, + "step": 19215 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19216 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19217 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19218 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.2036, + "step": 19219 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.9757, + "step": 19220 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.7735, + "step": 19221 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 5.1184, + "step": 19222 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19223 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.9913, + "step": 19224 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19225 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.1265, + "step": 19226 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19227 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.1848, + "step": 19228 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19229 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.5599, + "step": 19230 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.9159, + "step": 19231 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.8211, + "step": 19232 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19233 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.2619, + "step": 19234 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 5.2268, + "step": 19235 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19236 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.0632, + "step": 19237 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19238 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.3177, + "step": 19239 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19240 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19241 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.916, + "step": 19242 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.1817, + "step": 19243 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.9657, + "step": 19244 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.7427, + "step": 19245 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 8.1552, + "step": 19246 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.8823, + "step": 19247 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 9.5181, + "step": 19248 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 6.5326, + "step": 19249 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 6.6125, + "step": 19250 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19251 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.3129, + "step": 19252 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 6.9175, + "step": 19253 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.9638, + "step": 19254 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19255 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19256 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19257 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.9946, + "step": 19258 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19259 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.8136, + "step": 19260 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19261 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 12.186, + "step": 19262 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.9797, + "step": 19263 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.0113, + "step": 19264 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 8.3315, + "step": 19265 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 9.9622, + "step": 19266 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.9772, + "step": 19267 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19268 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.6482, + "step": 19269 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19270 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19271 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.5024, + "step": 19272 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19273 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19274 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19275 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19276 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19277 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 9.1614, + "step": 19278 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.1025, + "step": 19279 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19280 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.9694, + "step": 19281 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19282 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.2067, + "step": 19283 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.8452, + "step": 19284 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19285 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.7638, + "step": 19286 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.093, + "step": 19287 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.0631, + "step": 19288 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19289 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19290 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19291 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19292 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19293 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19294 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 7.3769, + "step": 19295 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19296 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.4551, + "step": 19297 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19298 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19299 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.1375, + "step": 19300 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19301 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19302 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 7.8892, + "step": 19303 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 2.7732, + "step": 19304 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 3.8846, + "step": 19305 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19306 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 4.9993, + "step": 19307 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 8.8264, + "step": 19308 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19309 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 8.8387, + "step": 19310 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19311 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 6.8358, + "step": 19312 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 6.519, + "step": 19313 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19314 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19315 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19316 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19317 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19318 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 1.6748, + "step": 19319 + }, + { + "epoch": 1.61, + "learning_rate": 0.00019989968512274642, + "loss": 7.6763, + "step": 19320 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.9018, + "step": 19321 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19322 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19323 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.7118, + "step": 19324 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.9795, + "step": 19325 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.387, + "step": 19326 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.007, + "step": 19327 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19328 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 6.3863, + "step": 19329 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.4671, + "step": 19330 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19331 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19332 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 7.6827, + "step": 19333 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 5.0034, + "step": 19334 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.053, + "step": 19335 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19336 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.9241, + "step": 19337 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19338 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 6.153, + "step": 19339 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.4704, + "step": 19340 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 5.5729, + "step": 19341 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19342 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19343 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.9402, + "step": 19344 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 5.2929, + "step": 19345 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.0426, + "step": 19346 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19347 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 6.8618, + "step": 19348 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19349 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19350 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19351 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19352 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19353 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.9647, + "step": 19354 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.3952, + "step": 19355 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19356 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19357 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 8.4756, + "step": 19358 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.7574, + "step": 19359 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 5.0351, + "step": 19360 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.0789, + "step": 19361 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19362 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.3391, + "step": 19363 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19364 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.6408, + "step": 19365 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19366 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.7703, + "step": 19367 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19368 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.5643, + "step": 19369 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 6.5337, + "step": 19370 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.8687, + "step": 19371 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.7805, + "step": 19372 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 7.3437, + "step": 19373 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 9.9288, + "step": 19374 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.4277, + "step": 19375 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 6.6037, + "step": 19376 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19377 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19378 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.4289, + "step": 19379 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.8163, + "step": 19380 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.487, + "step": 19381 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.5071, + "step": 19382 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19383 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19384 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19385 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.8846, + "step": 19386 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.99, + "step": 19387 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 8.4072, + "step": 19388 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.2782, + "step": 19389 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.2104, + "step": 19390 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.7537, + "step": 19391 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.2524, + "step": 19392 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19393 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19394 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19395 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19396 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.994, + "step": 19397 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19398 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.2887, + "step": 19399 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.1404, + "step": 19400 + }, + { + "epoch": 1.62, + "eval_loss": NaN, + "eval_runtime": 43.5092, + "eval_samples_per_second": 57.459, + "eval_steps_per_second": 7.194, + "step": 19400 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.7893, + "step": 19401 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19402 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.9906, + "step": 19403 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.3201, + "step": 19404 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.96, + "step": 19405 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.1063, + "step": 19406 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.8738, + "step": 19407 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19408 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.2711, + "step": 19409 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19410 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 6.5507, + "step": 19411 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 8.197, + "step": 19412 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19413 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19414 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19415 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 7.8184, + "step": 19416 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.021, + "step": 19417 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.3893, + "step": 19418 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.3632, + "step": 19419 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.2864, + "step": 19420 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 1.7831, + "step": 19421 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 8.3703, + "step": 19422 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19423 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 7.2047, + "step": 19424 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19425 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.2123, + "step": 19426 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19427 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19428 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.3462, + "step": 19429 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19430 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 6.3544, + "step": 19431 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 2.0704, + "step": 19432 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19433 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 3.1841, + "step": 19434 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19435 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19436 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19437 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19438 + }, + { + "epoch": 1.62, + "learning_rate": 0.00019989968512274642, + "loss": 4.2726, + "step": 19439 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19440 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 4.0161, + "step": 19441 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19442 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19443 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19444 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.8937, + "step": 19445 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19446 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19447 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 6.5714, + "step": 19448 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.3345, + "step": 19449 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.0704, + "step": 19450 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 12.1631, + "step": 19451 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 6.124, + "step": 19452 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.879, + "step": 19453 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19454 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19455 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.405, + "step": 19456 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 4.4448, + "step": 19457 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.8458, + "step": 19458 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.4544, + "step": 19459 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.1114, + "step": 19460 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 9.4602, + "step": 19461 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.4283, + "step": 19462 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19463 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.905, + "step": 19464 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.1965, + "step": 19465 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 7.0903, + "step": 19466 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19467 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19468 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.2297, + "step": 19469 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.0157, + "step": 19470 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.0635, + "step": 19471 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19472 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19473 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.4716, + "step": 19474 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19475 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 4.3701, + "step": 19476 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19477 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19478 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.0389, + "step": 19479 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 5.706, + "step": 19480 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.4629, + "step": 19481 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19482 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19483 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.1003, + "step": 19484 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 4.0419, + "step": 19485 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19486 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19487 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19488 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.0435, + "step": 19489 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19490 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19491 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.0564, + "step": 19492 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19493 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 6.8163, + "step": 19494 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19495 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19496 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19497 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19498 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.8647, + "step": 19499 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19500 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.3172, + "step": 19501 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 7.0615, + "step": 19502 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.8578, + "step": 19503 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.4553, + "step": 19504 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19505 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19506 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19507 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 4.868, + "step": 19508 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.0047, + "step": 19509 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19510 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19511 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19512 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 7.5926, + "step": 19513 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.5553, + "step": 19514 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 6.4345, + "step": 19515 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19516 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19517 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19518 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 9.0774, + "step": 19519 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19520 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19521 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 4.1839, + "step": 19522 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 8.3906, + "step": 19523 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.3428, + "step": 19524 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.5732, + "step": 19525 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 7.6355, + "step": 19526 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19527 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19528 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19529 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19530 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19531 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19532 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 5.5338, + "step": 19533 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19534 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19535 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.878, + "step": 19536 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19537 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 2.2505, + "step": 19538 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 4.4496, + "step": 19539 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 8.3081, + "step": 19540 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19541 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.7308, + "step": 19542 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.7605, + "step": 19543 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 8.287, + "step": 19544 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 6.2954, + "step": 19545 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 9.0027, + "step": 19546 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.9931, + "step": 19547 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.1171, + "step": 19548 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 5.0869, + "step": 19549 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.917, + "step": 19550 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 11.6811, + "step": 19551 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19552 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19553 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19554 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.9961, + "step": 19555 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19556 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 6.0597, + "step": 19557 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 1.9721, + "step": 19558 + }, + { + "epoch": 1.63, + "learning_rate": 0.00019989968512274642, + "loss": 3.0198, + "step": 19559 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19560 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19561 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19562 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 7.1105, + "step": 19563 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19564 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19565 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 7.509, + "step": 19566 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.1165, + "step": 19567 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.0447, + "step": 19568 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.2312, + "step": 19569 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 4.0694, + "step": 19570 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19571 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 5.5626, + "step": 19572 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 6.871, + "step": 19573 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.9328, + "step": 19574 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 4.31, + "step": 19575 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 3.7673, + "step": 19576 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19577 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 4.3811, + "step": 19578 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 9.0931, + "step": 19579 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 8.7807, + "step": 19580 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 8.5245, + "step": 19581 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19582 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19583 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19584 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 7.391, + "step": 19585 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.0178, + "step": 19586 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19587 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19588 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19589 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19590 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19591 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19592 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.8951, + "step": 19593 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 4.412, + "step": 19594 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.0096, + "step": 19595 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19596 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19597 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.8261, + "step": 19598 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 3.5078, + "step": 19599 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19600 + }, + { + "epoch": 1.64, + "eval_loss": NaN, + "eval_runtime": 44.2026, + "eval_samples_per_second": 56.558, + "eval_steps_per_second": 7.081, + "step": 19600 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 9.3673, + "step": 19601 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19602 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19603 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.0348, + "step": 19604 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 7.4114, + "step": 19605 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 3.7894, + "step": 19606 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19607 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19608 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19609 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19610 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.1619, + "step": 19611 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 7.8006, + "step": 19612 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.5893, + "step": 19613 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19614 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19615 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19616 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.9437, + "step": 19617 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19618 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.1701, + "step": 19619 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19620 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19621 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.967, + "step": 19622 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 3.3575, + "step": 19623 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19624 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19625 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 6.4092, + "step": 19626 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19627 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.4232, + "step": 19628 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19629 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.204, + "step": 19630 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 4.4076, + "step": 19631 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19632 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19633 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19634 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19635 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 8.3004, + "step": 19636 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.9764, + "step": 19637 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 10.1047, + "step": 19638 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19639 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19640 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19641 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19642 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 3.5264, + "step": 19643 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19644 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19645 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 7.2337, + "step": 19646 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.0038, + "step": 19647 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19648 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 3.9801, + "step": 19649 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.6045, + "step": 19650 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.038, + "step": 19651 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.1325, + "step": 19652 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 3.9624, + "step": 19653 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19654 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19655 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.1034, + "step": 19656 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.7795, + "step": 19657 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19658 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.8116, + "step": 19659 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.8402, + "step": 19660 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19661 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 2.051, + "step": 19662 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19663 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.6604, + "step": 19664 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 6.2054, + "step": 19665 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19666 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.8055, + "step": 19667 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19668 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19669 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.9176, + "step": 19670 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 8.6266, + "step": 19671 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 7.5275, + "step": 19672 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19673 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 4.3958, + "step": 19674 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 5.9327, + "step": 19675 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19676 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 6.0356, + "step": 19677 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19678 + }, + { + "epoch": 1.64, + "learning_rate": 0.00019989968512274642, + "loss": 1.818, + "step": 19679 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.5269, + "step": 19680 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19681 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19682 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 7.3386, + "step": 19683 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 2.191, + "step": 19684 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19685 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19686 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.1803, + "step": 19687 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.8818, + "step": 19688 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.8581, + "step": 19689 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19690 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 8.7115, + "step": 19691 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19692 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19693 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19694 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.9962, + "step": 19695 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19696 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19697 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19698 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.9934, + "step": 19699 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.6841, + "step": 19700 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 5.0197, + "step": 19701 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19702 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19703 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.7552, + "step": 19704 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.9433, + "step": 19705 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.8666, + "step": 19706 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.3154, + "step": 19707 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 7.6488, + "step": 19708 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19709 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19710 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 2.2862, + "step": 19711 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 9.4756, + "step": 19712 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 6.5793, + "step": 19713 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 6.8303, + "step": 19714 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19715 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19716 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.8388, + "step": 19717 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19718 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 2.7142, + "step": 19719 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.3916, + "step": 19720 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19721 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19722 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.7296, + "step": 19723 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19724 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 9.4078, + "step": 19725 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19726 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19727 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.024, + "step": 19728 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.9821, + "step": 19729 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 7.5006, + "step": 19730 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19731 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19732 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.0996, + "step": 19733 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19734 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19735 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 6.7544, + "step": 19736 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19737 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 5.4568, + "step": 19738 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.8781, + "step": 19739 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 5.0454, + "step": 19740 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.7904, + "step": 19741 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 8.7252, + "step": 19742 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19743 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 7.0531, + "step": 19744 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.2199, + "step": 19745 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.3612, + "step": 19746 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19747 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.2372, + "step": 19748 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19749 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19750 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19751 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.6619, + "step": 19752 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.0942, + "step": 19753 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 8.3266, + "step": 19754 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.9721, + "step": 19755 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19756 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 2.4606, + "step": 19757 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 6.0654, + "step": 19758 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19759 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 11.4438, + "step": 19760 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19761 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19762 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.3059, + "step": 19763 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 7.7179, + "step": 19764 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 11.0541, + "step": 19765 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 2.371, + "step": 19766 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.9581, + "step": 19767 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 5.9591, + "step": 19768 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.8043, + "step": 19769 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19770 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19771 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19772 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19773 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 8.5603, + "step": 19774 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.9707, + "step": 19775 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19776 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 8.0618, + "step": 19777 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.301, + "step": 19778 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 5.3378, + "step": 19779 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 7.3946, + "step": 19780 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 3.3511, + "step": 19781 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.9602, + "step": 19782 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 8.1094, + "step": 19783 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 7.0269, + "step": 19784 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19785 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19786 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 1.1397, + "step": 19787 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19788 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19789 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19790 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 8.246, + "step": 19791 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19792 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 5.0843, + "step": 19793 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19794 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19795 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.0599, + "step": 19796 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19797 + }, + { + "epoch": 1.65, + "learning_rate": 0.00019989968512274642, + "loss": 4.0785, + "step": 19798 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 5.8294, + "step": 19799 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.104, + "step": 19800 + }, + { + "epoch": 1.66, + "eval_loss": NaN, + "eval_runtime": 41.7446, + "eval_samples_per_second": 59.888, + "eval_steps_per_second": 7.498, + "step": 19800 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.3868, + "step": 19801 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19802 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19803 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 6.7262, + "step": 19804 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19805 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19806 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 5.1509, + "step": 19807 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19808 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19809 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.6063, + "step": 19810 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 3.3432, + "step": 19811 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 4.6785, + "step": 19812 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19813 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19814 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 8.2794, + "step": 19815 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 9.3363, + "step": 19816 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 4.9416, + "step": 19817 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 7.3678, + "step": 19818 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19819 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 8.1512, + "step": 19820 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 3.9371, + "step": 19821 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19822 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.0099, + "step": 19823 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 7.4658, + "step": 19824 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19825 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19826 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.897, + "step": 19827 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19828 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19829 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 4.6575, + "step": 19830 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19831 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.8243, + "step": 19832 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 9.0961, + "step": 19833 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19834 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.6967, + "step": 19835 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 7.3585, + "step": 19836 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 3.868, + "step": 19837 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19838 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19839 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19840 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.4132, + "step": 19841 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.9209, + "step": 19842 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19843 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 2.053, + "step": 19844 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.1142, + "step": 19845 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 7.7894, + "step": 19846 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19847 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19848 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19849 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 2.8212, + "step": 19850 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 7.0886, + "step": 19851 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 13.1383, + "step": 19852 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19853 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19854 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19855 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.6484, + "step": 19856 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19857 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19858 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19859 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19860 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19861 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19862 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.8469, + "step": 19863 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19864 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 6.1058, + "step": 19865 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 3.7543, + "step": 19866 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19867 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19868 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19869 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.0853, + "step": 19870 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 10.8487, + "step": 19871 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 6.1254, + "step": 19872 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 4.9364, + "step": 19873 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19874 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 9.91, + "step": 19875 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 2.1743, + "step": 19876 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 4.8978, + "step": 19877 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 4.789, + "step": 19878 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.7752, + "step": 19879 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.9462, + "step": 19880 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19881 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.6475, + "step": 19882 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19883 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.9001, + "step": 19884 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19885 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19886 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 8.3907, + "step": 19887 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19888 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19889 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.6541, + "step": 19890 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.0533, + "step": 19891 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 2.8564, + "step": 19892 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19893 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19894 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19895 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.0412, + "step": 19896 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19897 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19898 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19899 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19900 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19901 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 7.9667, + "step": 19902 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19903 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19904 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19905 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19906 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19907 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 6.6377, + "step": 19908 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 8.5696, + "step": 19909 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 3.168, + "step": 19910 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 2.0643, + "step": 19911 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 2.2252, + "step": 19912 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.6437, + "step": 19913 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 1.7215, + "step": 19914 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19915 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19916 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19917 + }, + { + "epoch": 1.66, + "learning_rate": 0.00019989968512274642, + "loss": 2.0605, + "step": 19918 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19919 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.9597, + "step": 19920 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 8.1091, + "step": 19921 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.0468, + "step": 19922 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19923 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 5.895, + "step": 19924 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.0976, + "step": 19925 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19926 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 7.3474, + "step": 19927 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19928 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 7.4672, + "step": 19929 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19930 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.8327, + "step": 19931 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.9811, + "step": 19932 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 9.4237, + "step": 19933 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19934 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19935 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.2761, + "step": 19936 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.0576, + "step": 19937 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.4378, + "step": 19938 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 4.0184, + "step": 19939 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19940 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19941 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.6913, + "step": 19942 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.8145, + "step": 19943 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19944 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.9688, + "step": 19945 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19946 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.2343, + "step": 19947 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 5.0027, + "step": 19948 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 9.1339, + "step": 19949 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19950 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19951 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19952 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 6.948, + "step": 19953 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.035, + "step": 19954 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.1853, + "step": 19955 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19956 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.8546, + "step": 19957 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 6.1432, + "step": 19958 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.2964, + "step": 19959 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19960 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.8765, + "step": 19961 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.6506, + "step": 19962 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 4.2729, + "step": 19963 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.6492, + "step": 19964 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.5461, + "step": 19965 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19966 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 5.1247, + "step": 19967 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19968 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19969 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.1809, + "step": 19970 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 7.1082, + "step": 19971 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 4.0521, + "step": 19972 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19973 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.3826, + "step": 19974 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19975 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 8.9086, + "step": 19976 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19977 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19978 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19979 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.344, + "step": 19980 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.1692, + "step": 19981 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 6.5076, + "step": 19982 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19983 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.5031, + "step": 19984 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19985 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19986 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19987 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.6965, + "step": 19988 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.9466, + "step": 19989 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.7893, + "step": 19990 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 7.4439, + "step": 19991 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 7.8409, + "step": 19992 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19993 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 10.9529, + "step": 19994 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.2835, + "step": 19995 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19996 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19997 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 19998 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.5692, + "step": 19999 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.6578, + "step": 20000 + }, + { + "epoch": 1.67, + "eval_loss": NaN, + "eval_runtime": 43.4103, + "eval_samples_per_second": 57.59, + "eval_steps_per_second": 7.21, + "step": 20000 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.9419, + "step": 20001 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.7857, + "step": 20002 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 8.3572, + "step": 20003 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.0105, + "step": 20004 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.3938, + "step": 20005 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20006 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.0462, + "step": 20007 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 5.5218, + "step": 20008 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.4535, + "step": 20009 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20010 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 1.1602, + "step": 20011 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 15.0598, + "step": 20012 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20013 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20014 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20015 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 4.8574, + "step": 20016 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20017 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20018 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20019 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.9061, + "step": 20020 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20021 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20022 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20023 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 7.3083, + "step": 20024 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.0849, + "step": 20025 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 3.2872, + "step": 20026 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20027 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.3614, + "step": 20028 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20029 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20030 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20031 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20032 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 5.868, + "step": 20033 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20034 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 2.0474, + "step": 20035 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20036 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 4.4919, + "step": 20037 + }, + { + "epoch": 1.67, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20038 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20039 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 4.1537, + "step": 20040 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20041 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.5327, + "step": 20042 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 5.5953, + "step": 20043 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.5742, + "step": 20044 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.7783, + "step": 20045 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.1719, + "step": 20046 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20047 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20048 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.3778, + "step": 20049 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 7.2315, + "step": 20050 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20051 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20052 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 9.8844, + "step": 20053 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 5.1684, + "step": 20054 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20055 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.5227, + "step": 20056 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.1042, + "step": 20057 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.5045, + "step": 20058 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20059 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20060 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.1489, + "step": 20061 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20062 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20063 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20064 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.8902, + "step": 20065 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20066 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20067 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20068 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.1151, + "step": 20069 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20070 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20071 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.4182, + "step": 20072 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.0358, + "step": 20073 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.7527, + "step": 20074 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20075 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.471, + "step": 20076 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20077 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.8252, + "step": 20078 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20079 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.6546, + "step": 20080 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20081 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 5.7295, + "step": 20082 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20083 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 6.8383, + "step": 20084 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20085 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20086 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20087 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20088 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.8878, + "step": 20089 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20090 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20091 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.0111, + "step": 20092 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.87, + "step": 20093 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20094 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 4.5368, + "step": 20095 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20096 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20097 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 4.6396, + "step": 20098 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 4.6993, + "step": 20099 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 8.3702, + "step": 20100 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20101 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20102 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20103 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.072, + "step": 20104 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 9.4588, + "step": 20105 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 5.9326, + "step": 20106 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20107 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.323, + "step": 20108 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 6.85, + "step": 20109 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.0632, + "step": 20110 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.8782, + "step": 20111 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.4101, + "step": 20112 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20113 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20114 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20115 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.7256, + "step": 20116 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 6.2845, + "step": 20117 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20118 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 5.7732, + "step": 20119 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20120 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.2829, + "step": 20121 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.5513, + "step": 20122 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20123 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.7703, + "step": 20124 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20125 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20126 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20127 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.1808, + "step": 20128 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 7.7817, + "step": 20129 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20130 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 8.5618, + "step": 20131 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 7.1414, + "step": 20132 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20133 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 7.5262, + "step": 20134 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.0116, + "step": 20135 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20136 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.1759, + "step": 20137 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20138 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 4.1672, + "step": 20139 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.7354, + "step": 20140 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.0445, + "step": 20141 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20142 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 2.1211, + "step": 20143 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 3.6849, + "step": 20144 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 11.8264, + "step": 20145 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20146 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.958, + "step": 20147 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20148 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20149 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.9573, + "step": 20150 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 1.8121, + "step": 20151 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20152 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20153 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20154 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20155 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.9602, + "step": 20156 + }, + { + "epoch": 1.68, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20157 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.9315, + "step": 20158 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.9946, + "step": 20159 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20160 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.9492, + "step": 20161 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20162 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.1751, + "step": 20163 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20164 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20165 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.7303, + "step": 20166 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.18, + "step": 20167 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.1481, + "step": 20168 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.0744, + "step": 20169 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.1085, + "step": 20170 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 12.5246, + "step": 20171 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20172 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20173 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 6.9137, + "step": 20174 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.9893, + "step": 20175 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.6248, + "step": 20176 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.0473, + "step": 20177 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20178 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.089, + "step": 20179 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20180 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20181 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20182 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20183 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 9.2166, + "step": 20184 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 7.7959, + "step": 20185 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.6731, + "step": 20186 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 5.187, + "step": 20187 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20188 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.9238, + "step": 20189 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.0995, + "step": 20190 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.439, + "step": 20191 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.9849, + "step": 20192 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 5.6618, + "step": 20193 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.7721, + "step": 20194 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.3567, + "step": 20195 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.6525, + "step": 20196 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 6.958, + "step": 20197 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20198 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20199 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.3733, + "step": 20200 + }, + { + "epoch": 1.69, + "eval_loss": NaN, + "eval_runtime": 41.6224, + "eval_samples_per_second": 60.064, + "eval_steps_per_second": 7.52, + "step": 20200 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 6.9603, + "step": 20201 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.4637, + "step": 20202 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.3571, + "step": 20203 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.234, + "step": 20204 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20205 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20206 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20207 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.9941, + "step": 20208 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20209 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.5904, + "step": 20210 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.0462, + "step": 20211 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20212 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.0398, + "step": 20213 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20214 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20215 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.3308, + "step": 20216 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 9.8485, + "step": 20217 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20218 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20219 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.9447, + "step": 20220 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20221 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20222 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 5.0791, + "step": 20223 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.7291, + "step": 20224 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.0621, + "step": 20225 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 7.4706, + "step": 20226 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20227 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20228 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 8.4485, + "step": 20229 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20230 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.8003, + "step": 20231 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20232 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.2452, + "step": 20233 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 8.2072, + "step": 20234 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20235 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20236 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20237 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20238 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20239 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.4077, + "step": 20240 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 8.0251, + "step": 20241 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.9853, + "step": 20242 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20243 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20244 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20245 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.919, + "step": 20246 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20247 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.1737, + "step": 20248 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20249 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 8.2721, + "step": 20250 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.1521, + "step": 20251 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 8.4403, + "step": 20252 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20253 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20254 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20255 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20256 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20257 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 3.325, + "step": 20258 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.0156, + "step": 20259 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.9893, + "step": 20260 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 9.0632, + "step": 20261 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20262 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.9143, + "step": 20263 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20264 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.9975, + "step": 20265 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 1.2484, + "step": 20266 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 2.2796, + "step": 20267 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.5449, + "step": 20268 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 6.6239, + "step": 20269 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 7.4379, + "step": 20270 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20271 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20272 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20273 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 7.6748, + "step": 20274 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20275 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20276 + }, + { + "epoch": 1.69, + "learning_rate": 0.00019989968512274642, + "loss": 4.1361, + "step": 20277 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20278 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 11.5955, + "step": 20279 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20280 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20281 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20282 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.5116, + "step": 20283 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20284 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20285 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.6151, + "step": 20286 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 6.9456, + "step": 20287 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20288 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.8407, + "step": 20289 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20290 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 2.2318, + "step": 20291 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20292 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.5924, + "step": 20293 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20294 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20295 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.9077, + "step": 20296 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20297 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20298 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.5895, + "step": 20299 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20300 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.9117, + "step": 20301 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.762, + "step": 20302 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20303 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20304 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.011, + "step": 20305 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20306 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.0613, + "step": 20307 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.2596, + "step": 20308 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.7206, + "step": 20309 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.8229, + "step": 20310 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.8203, + "step": 20311 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 8.5728, + "step": 20312 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20313 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.9853, + "step": 20314 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20315 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 2.3503, + "step": 20316 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20317 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20318 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 2.3498, + "step": 20319 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.7442, + "step": 20320 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20321 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20322 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 8.9962, + "step": 20323 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.9893, + "step": 20324 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 5.2654, + "step": 20325 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.6093, + "step": 20326 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20327 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20328 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20329 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20330 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20331 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 2.0345, + "step": 20332 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20333 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.3577, + "step": 20334 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 5.1498, + "step": 20335 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.559, + "step": 20336 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20337 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 8.8736, + "step": 20338 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 5.1438, + "step": 20339 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20340 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 5.0572, + "step": 20341 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.6569, + "step": 20342 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20343 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.6407, + "step": 20344 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.7328, + "step": 20345 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 8.4215, + "step": 20346 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20347 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.2491, + "step": 20348 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20349 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20350 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.6777, + "step": 20351 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.7998, + "step": 20352 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.7986, + "step": 20353 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20354 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.9492, + "step": 20355 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 11.4351, + "step": 20356 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.0796, + "step": 20357 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20358 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20359 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.3124, + "step": 20360 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.4649, + "step": 20361 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20362 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 5.1681, + "step": 20363 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20364 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.0321, + "step": 20365 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 1.0762, + "step": 20366 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20367 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 10.0295, + "step": 20368 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20369 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20370 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 5.2887, + "step": 20371 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20372 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20373 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.0746, + "step": 20374 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20375 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.2825, + "step": 20376 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20377 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20378 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20379 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.8316, + "step": 20380 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20381 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20382 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 2.8845, + "step": 20383 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20384 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 2.166, + "step": 20385 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.9301, + "step": 20386 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.5539, + "step": 20387 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 7.5061, + "step": 20388 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 3.2888, + "step": 20389 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20390 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20391 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20392 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 9.4424, + "step": 20393 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20394 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20395 + }, + { + "epoch": 1.7, + "learning_rate": 0.00019989968512274642, + "loss": 4.5107, + "step": 20396 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 6.7593, + "step": 20397 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 8.8719, + "step": 20398 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20399 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.9223, + "step": 20400 + }, + { + "epoch": 1.71, + "eval_loss": NaN, + "eval_runtime": 41.5027, + "eval_samples_per_second": 60.237, + "eval_steps_per_second": 7.542, + "step": 20400 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.8779, + "step": 20401 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.6885, + "step": 20402 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.8934, + "step": 20403 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 10.7634, + "step": 20404 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.4891, + "step": 20405 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 5.8409, + "step": 20406 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.8184, + "step": 20407 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20408 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.1184, + "step": 20409 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 7.4091, + "step": 20410 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.9834, + "step": 20411 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20412 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20413 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.6305, + "step": 20414 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 2.0608, + "step": 20415 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20416 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.1266, + "step": 20417 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20418 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 7.5487, + "step": 20419 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 7.9493, + "step": 20420 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20421 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20422 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20423 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.4826, + "step": 20424 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.717, + "step": 20425 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 5.5507, + "step": 20426 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20427 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.4762, + "step": 20428 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20429 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20430 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.1108, + "step": 20431 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.2281, + "step": 20432 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20433 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.9099, + "step": 20434 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.0047, + "step": 20435 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.7433, + "step": 20436 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.7691, + "step": 20437 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20438 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.2435, + "step": 20439 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 5.2061, + "step": 20440 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.9464, + "step": 20441 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20442 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.6031, + "step": 20443 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 6.7754, + "step": 20444 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20445 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 2.0264, + "step": 20446 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20447 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20448 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.0054, + "step": 20449 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20450 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 6.9769, + "step": 20451 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20452 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.9372, + "step": 20453 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20454 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20455 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20456 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20457 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20458 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 7.7626, + "step": 20459 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20460 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 2.8435, + "step": 20461 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 5.5189, + "step": 20462 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20463 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 6.9643, + "step": 20464 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20465 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20466 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.0976, + "step": 20467 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20468 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 2.8312, + "step": 20469 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20470 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20471 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.9899, + "step": 20472 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.9534, + "step": 20473 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.549, + "step": 20474 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20475 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.135, + "step": 20476 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.0717, + "step": 20477 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 8.0662, + "step": 20478 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 9.5121, + "step": 20479 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20480 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.6395, + "step": 20481 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20482 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20483 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 2.5054, + "step": 20484 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.0178, + "step": 20485 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20486 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20487 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20488 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 2.189, + "step": 20489 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20490 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20491 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.6948, + "step": 20492 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20493 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20494 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 2.0943, + "step": 20495 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.9536, + "step": 20496 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 7.1075, + "step": 20497 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 1.9067, + "step": 20498 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20499 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20500 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20501 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20502 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 8.0938, + "step": 20503 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20504 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20505 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 7.1949, + "step": 20506 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20507 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20508 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20509 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 8.0675, + "step": 20510 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 4.4021, + "step": 20511 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20512 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.8611, + "step": 20513 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20514 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 3.8673, + "step": 20515 + }, + { + "epoch": 1.71, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20516 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20517 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 8.2122, + "step": 20518 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 5.6306, + "step": 20519 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.5573, + "step": 20520 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.0444, + "step": 20521 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 8.71, + "step": 20522 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20523 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.1534, + "step": 20524 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.4996, + "step": 20525 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.7877, + "step": 20526 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20527 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20528 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.7678, + "step": 20529 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.4002, + "step": 20530 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.8952, + "step": 20531 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20532 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.0377, + "step": 20533 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.292, + "step": 20534 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.1091, + "step": 20535 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.7129, + "step": 20536 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 5.8583, + "step": 20537 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 5.2728, + "step": 20538 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.8978, + "step": 20539 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.4589, + "step": 20540 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20541 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.7011, + "step": 20542 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20543 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.0858, + "step": 20544 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.548, + "step": 20545 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20546 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.3448, + "step": 20547 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 5.4256, + "step": 20548 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20549 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.1295, + "step": 20550 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20551 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20552 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.7766, + "step": 20553 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20554 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.1768, + "step": 20555 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.7384, + "step": 20556 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 6.4592, + "step": 20557 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20558 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 6.2702, + "step": 20559 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.9929, + "step": 20560 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.7333, + "step": 20561 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.0547, + "step": 20562 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.2411, + "step": 20563 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.143, + "step": 20564 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.9501, + "step": 20565 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20566 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.9099, + "step": 20567 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20568 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20569 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 6.7708, + "step": 20570 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.2181, + "step": 20571 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20572 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 10.9839, + "step": 20573 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20574 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20575 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.7265, + "step": 20576 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20577 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.4186, + "step": 20578 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20579 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.6183, + "step": 20580 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20581 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.7365, + "step": 20582 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.6861, + "step": 20583 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20584 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.8236, + "step": 20585 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.7322, + "step": 20586 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.3952, + "step": 20587 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20588 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.8518, + "step": 20589 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.5847, + "step": 20590 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 5.0751, + "step": 20591 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20592 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20593 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20594 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 9.1983, + "step": 20595 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.4351, + "step": 20596 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.4296, + "step": 20597 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.0592, + "step": 20598 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.5978, + "step": 20599 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20600 + }, + { + "epoch": 1.72, + "eval_loss": NaN, + "eval_runtime": 41.4235, + "eval_samples_per_second": 60.352, + "eval_steps_per_second": 7.556, + "step": 20600 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.242, + "step": 20601 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20602 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.9398, + "step": 20603 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20604 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.4857, + "step": 20605 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.5605, + "step": 20606 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.5249, + "step": 20607 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.2942, + "step": 20608 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20609 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20610 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.7381, + "step": 20611 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.9342, + "step": 20612 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.7377, + "step": 20613 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20614 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20615 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20616 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 3.9064, + "step": 20617 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.2091, + "step": 20618 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 8.3972, + "step": 20619 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 7.5215, + "step": 20620 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20621 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20622 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 1.9636, + "step": 20623 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20624 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 2.7334, + "step": 20625 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 9.4034, + "step": 20626 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 5.661, + "step": 20627 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.8552, + "step": 20628 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20629 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20630 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 4.3317, + "step": 20631 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20632 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20633 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 5.1683, + "step": 20634 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20635 + }, + { + "epoch": 1.72, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20636 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20637 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.7243, + "step": 20638 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.1043, + "step": 20639 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20640 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20641 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.1496, + "step": 20642 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20643 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.2258, + "step": 20644 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20645 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.1944, + "step": 20646 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20647 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.834, + "step": 20648 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20649 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20650 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20651 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.8234, + "step": 20652 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 5.1485, + "step": 20653 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.3467, + "step": 20654 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 10.6162, + "step": 20655 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 6.8854, + "step": 20656 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20657 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 6.5074, + "step": 20658 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.4116, + "step": 20659 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.5492, + "step": 20660 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.0348, + "step": 20661 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.8619, + "step": 20662 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20663 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 6.7999, + "step": 20664 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 6.1861, + "step": 20665 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.677, + "step": 20666 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20667 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20668 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20669 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20670 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20671 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.2439, + "step": 20672 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20673 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.6933, + "step": 20674 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.136, + "step": 20675 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20676 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20677 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.902, + "step": 20678 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 5.4887, + "step": 20679 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20680 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.9974, + "step": 20681 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20682 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20683 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20684 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.9915, + "step": 20685 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 8.6148, + "step": 20686 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20687 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20688 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.9742, + "step": 20689 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 7.3624, + "step": 20690 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.9942, + "step": 20691 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.9676, + "step": 20692 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 2.299, + "step": 20693 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20694 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20695 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 6.9769, + "step": 20696 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 8.9129, + "step": 20697 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20698 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.017, + "step": 20699 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.8982, + "step": 20700 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 6.423, + "step": 20701 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20702 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.9204, + "step": 20703 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20704 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 2.5666, + "step": 20705 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.7252, + "step": 20706 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20707 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20708 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20709 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.2831, + "step": 20710 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20711 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20712 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20713 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20714 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20715 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20716 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20717 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20718 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.2043, + "step": 20719 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 2.6171, + "step": 20720 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20721 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.9032, + "step": 20722 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20723 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20724 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.9209, + "step": 20725 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 8.2409, + "step": 20726 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 7.7539, + "step": 20727 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20728 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.4714, + "step": 20729 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20730 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.7692, + "step": 20731 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 8.1883, + "step": 20732 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 2.0346, + "step": 20733 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 8.6649, + "step": 20734 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.7897, + "step": 20735 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.843, + "step": 20736 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20737 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 2.0813, + "step": 20738 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 8.7233, + "step": 20739 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20740 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 2.0229, + "step": 20741 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.5069, + "step": 20742 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 7.4306, + "step": 20743 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.2362, + "step": 20744 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20745 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 2.1102, + "step": 20746 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20747 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 1.1276, + "step": 20748 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20749 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 3.4646, + "step": 20750 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.8468, + "step": 20751 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20752 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20753 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 5.3344, + "step": 20754 + }, + { + "epoch": 1.73, + "learning_rate": 0.00019989968512274642, + "loss": 4.8492, + "step": 20755 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 8.6744, + "step": 20756 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 2.354, + "step": 20757 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20758 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20759 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20760 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20761 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 0.0, + "step": 20762 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 9.3965, + "step": 20763 + }, + { + "epoch": 1.74, + "learning_rate": 0.00019989968512274642, + "loss": 4.9548, + "step": 20764 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.3656, + "step": 20765 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 4.6991, + "step": 20766 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20767 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20768 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 5.3864, + "step": 20769 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 2.3209, + "step": 20770 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20771 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20772 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.6516, + "step": 20773 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 2.0733, + "step": 20774 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 6.996, + "step": 20775 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.6912, + "step": 20776 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20777 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20778 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 4.3827, + "step": 20779 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.8251, + "step": 20780 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.0012, + "step": 20781 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.0287, + "step": 20782 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 4.3597, + "step": 20783 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20784 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.2812, + "step": 20785 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20786 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20787 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20788 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 4.3505, + "step": 20789 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20790 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 7.6923, + "step": 20791 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20792 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20793 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 5.5796, + "step": 20794 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.8621, + "step": 20795 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 4.2431, + "step": 20796 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 13.9402, + "step": 20797 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.1679, + "step": 20798 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20799 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20800 + }, + { + "epoch": 1.74, + "eval_loss": NaN, + "eval_runtime": 41.5268, + "eval_samples_per_second": 60.202, + "eval_steps_per_second": 7.537, + "step": 20800 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 2.676, + "step": 20801 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20802 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 5.7348, + "step": 20803 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20804 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.7906, + "step": 20805 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 2.9276, + "step": 20806 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20807 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 8.1645, + "step": 20808 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20809 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20810 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.8506, + "step": 20811 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20812 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 4.1129, + "step": 20813 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20814 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20815 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 6.5941, + "step": 20816 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 8.2611, + "step": 20817 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20818 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20819 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20820 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.7469, + "step": 20821 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.9798, + "step": 20822 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20823 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.642, + "step": 20824 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 9.2554, + "step": 20825 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20826 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20827 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.0598, + "step": 20828 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.6654, + "step": 20829 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20830 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.6733, + "step": 20831 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20832 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20833 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20834 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20835 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 5.3579, + "step": 20836 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 5.6201, + "step": 20837 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.4302, + "step": 20838 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.9448, + "step": 20839 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 9.4419, + "step": 20840 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20841 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20842 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.919, + "step": 20843 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20844 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 3.666, + "step": 20845 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20846 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20847 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 5.0067, + "step": 20848 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20849 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20850 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.8701, + "step": 20851 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 2.3319, + "step": 20852 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.6515, + "step": 20853 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20854 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20855 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 7.4172, + "step": 20856 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20857 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.8185, + "step": 20858 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 7.0053, + "step": 20859 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20860 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 9.2696, + "step": 20861 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 5.6219, + "step": 20862 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.0713, + "step": 20863 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 1.4187, + "step": 20864 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20865 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20866 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.8972, + "step": 20867 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20868 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.8303, + "step": 20869 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20870 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20871 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20872 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 8.5686, + "step": 20873 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20874 + }, + { + "epoch": 1.74, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20875 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20876 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 2.2554, + "step": 20877 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 5.6988, + "step": 20878 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20879 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 2.3276, + "step": 20880 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20881 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20882 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20883 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20884 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20885 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 5.4064, + "step": 20886 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20887 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 7.6121, + "step": 20888 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 2.0556, + "step": 20889 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 4.0196, + "step": 20890 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20891 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20892 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20893 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 8.2245, + "step": 20894 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 6.3766, + "step": 20895 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20896 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 7.1033, + "step": 20897 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 4.1067, + "step": 20898 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 10.4582, + "step": 20899 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.9983, + "step": 20900 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20901 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 6.4059, + "step": 20902 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.9961, + "step": 20903 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.9669, + "step": 20904 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 6.5374, + "step": 20905 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20906 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20907 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20908 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20909 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 6.5399, + "step": 20910 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 10.2699, + "step": 20911 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 3.6295, + "step": 20912 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20913 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 1.1213, + "step": 20914 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 3.8365, + "step": 20915 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 5.6936, + "step": 20916 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 1.8314, + "step": 20917 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 5.1414, + "step": 20918 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20919 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 1.0341, + "step": 20920 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20921 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20922 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 2.6841, + "step": 20923 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20924 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 2.233, + "step": 20925 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20926 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20927 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20928 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20929 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 1.9876, + "step": 20930 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 2.2335, + "step": 20931 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.9125, + "step": 20932 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20933 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20934 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 7.7323, + "step": 20935 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 1.1958, + "step": 20936 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20937 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20938 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 0.0, + "step": 20939 + }, + { + "epoch": 1.75, + "learning_rate": 0.0001998941120740101, + "loss": 1.9288, + "step": 20940 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 5.134, + "step": 20941 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20942 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20943 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 8.2178, + "step": 20944 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 3.4636, + "step": 20945 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 1.1347, + "step": 20946 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20947 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20948 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 1.8322, + "step": 20949 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 8.0757, + "step": 20950 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20951 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20952 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 1.8844, + "step": 20953 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.886, + "step": 20954 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 3.9457, + "step": 20955 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 4.1192, + "step": 20956 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 5.9407, + "step": 20957 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20958 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 2.5604, + "step": 20959 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20960 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 4.5142, + "step": 20961 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 8.8567, + "step": 20962 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 8.5649, + "step": 20963 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 9.3628, + "step": 20964 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20965 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20966 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 2.1778, + "step": 20967 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 8.7406, + "step": 20968 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 4.5677, + "step": 20969 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20970 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20971 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 4.4657, + "step": 20972 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 7.684, + "step": 20973 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 1.4921, + "step": 20974 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20975 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 3.6685, + "step": 20976 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20977 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 3.7298, + "step": 20978 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.9785, + "step": 20979 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20980 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.8869, + "step": 20981 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.9863, + "step": 20982 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 7.6131, + "step": 20983 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 1.6643, + "step": 20984 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20985 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 3.1293, + "step": 20986 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 2.0261, + "step": 20987 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20988 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20989 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20990 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20991 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20992 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20993 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 20994 + }, + { + "epoch": 1.75, + "learning_rate": 0.00019988853902527378, + "loss": 4.0762, + "step": 20995 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.5526, + "step": 20996 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 7.8194, + "step": 20997 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.2654, + "step": 20998 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.8842, + "step": 20999 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.7585, + "step": 21000 + }, + { + "epoch": 1.76, + "eval_loss": NaN, + "eval_runtime": 41.5062, + "eval_samples_per_second": 60.232, + "eval_steps_per_second": 7.541, + "step": 21000 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.8824, + "step": 21001 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21002 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 2.0509, + "step": 21003 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.1188, + "step": 21004 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21005 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 2.8523, + "step": 21006 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 6.2178, + "step": 21007 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21008 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 2.1397, + "step": 21009 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21010 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.8037, + "step": 21011 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21012 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.7649, + "step": 21013 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21014 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21015 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21016 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21017 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.4123, + "step": 21018 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.6815, + "step": 21019 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21020 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.9504, + "step": 21021 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21022 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.3868, + "step": 21023 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21024 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 7.0835, + "step": 21025 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21026 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 8.3696, + "step": 21027 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.9354, + "step": 21028 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21029 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21030 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 8.9254, + "step": 21031 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21032 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21033 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21034 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.0981, + "step": 21035 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.9783, + "step": 21036 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21037 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21038 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21039 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21040 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 5.6732, + "step": 21041 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21042 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21043 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21044 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.1607, + "step": 21045 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 10.4298, + "step": 21046 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 2.129, + "step": 21047 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.7691, + "step": 21048 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21049 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21050 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21051 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21052 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21053 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 6.1638, + "step": 21054 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21055 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21056 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 8.6665, + "step": 21057 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.0431, + "step": 21058 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21059 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21060 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.7315, + "step": 21061 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21062 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 2.1708, + "step": 21063 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.5066, + "step": 21064 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 8.1397, + "step": 21065 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 10.6825, + "step": 21066 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21067 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21068 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21069 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 6.7645, + "step": 21070 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21071 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21072 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 8.9292, + "step": 21073 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21074 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.9868, + "step": 21075 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 5.1781, + "step": 21076 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21077 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21078 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21079 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.5986, + "step": 21080 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.9749, + "step": 21081 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21082 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.9937, + "step": 21083 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21084 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 5.4704, + "step": 21085 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.5559, + "step": 21086 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.0877, + "step": 21087 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.7521, + "step": 21088 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 11.9519, + "step": 21089 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21090 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21091 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.8462, + "step": 21092 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.6147, + "step": 21093 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21094 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 9.7311, + "step": 21095 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21096 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21097 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21098 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.3678, + "step": 21099 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.9963, + "step": 21100 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21101 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21102 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21103 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 5.1941, + "step": 21104 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.1492, + "step": 21105 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.3598, + "step": 21106 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.1113, + "step": 21107 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21108 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21109 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 6.4581, + "step": 21110 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 3.0448, + "step": 21111 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 1.6526, + "step": 21112 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21113 + }, + { + "epoch": 1.76, + "learning_rate": 0.00019988853902527378, + "loss": 4.1649, + "step": 21114 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.9794, + "step": 21115 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21116 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 4.616, + "step": 21117 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 9.4652, + "step": 21118 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21119 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 5.1299, + "step": 21120 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21121 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21122 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21123 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 7.1355, + "step": 21124 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 9.0731, + "step": 21125 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21126 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.9853, + "step": 21127 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21128 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.1027, + "step": 21129 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 3.9711, + "step": 21130 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.087, + "step": 21131 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21132 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 5.3529, + "step": 21133 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.0637, + "step": 21134 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21135 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 3.9866, + "step": 21136 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.1861, + "step": 21137 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21138 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21139 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 5.0594, + "step": 21140 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21141 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21142 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.4313, + "step": 21143 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 4.8817, + "step": 21144 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 7.71, + "step": 21145 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21146 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21147 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.8626, + "step": 21148 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21149 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21150 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 4.1494, + "step": 21151 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.3718, + "step": 21152 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.1488, + "step": 21153 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.171, + "step": 21154 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21155 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21156 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21157 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21158 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.5977, + "step": 21159 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 10.9105, + "step": 21160 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 5.8356, + "step": 21161 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21162 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 6.163, + "step": 21163 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.9646, + "step": 21164 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21165 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21166 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21167 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21168 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 7.2945, + "step": 21169 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.7841, + "step": 21170 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21171 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 7.7827, + "step": 21172 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 7.9513, + "step": 21173 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 3.3587, + "step": 21174 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 7.8132, + "step": 21175 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21176 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.0143, + "step": 21177 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 9.9168, + "step": 21178 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21179 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.3365, + "step": 21180 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 4.5145, + "step": 21181 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.8888, + "step": 21182 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21183 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 6.5793, + "step": 21184 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21185 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.5997, + "step": 21186 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21187 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.6473, + "step": 21188 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21189 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21190 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21191 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21192 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21193 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21194 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21195 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 8.1492, + "step": 21196 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21197 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21198 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 8.7157, + "step": 21199 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21200 + }, + { + "epoch": 1.77, + "eval_loss": NaN, + "eval_runtime": 44.6285, + "eval_samples_per_second": 56.018, + "eval_steps_per_second": 7.013, + "step": 21200 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 2.0153, + "step": 21201 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21202 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21203 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 8.9898, + "step": 21204 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21205 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 4.8496, + "step": 21206 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21207 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21208 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.9981, + "step": 21209 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21210 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21211 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 3.8601, + "step": 21212 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 4.2094, + "step": 21213 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21214 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 6.1061, + "step": 21215 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21216 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.9561, + "step": 21217 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.8343, + "step": 21218 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21219 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 8.8436, + "step": 21220 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.02, + "step": 21221 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.6644, + "step": 21222 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 5.1856, + "step": 21223 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21224 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21225 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21226 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.3699, + "step": 21227 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21228 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21229 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21230 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 4.5785, + "step": 21231 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.9288, + "step": 21232 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.9043, + "step": 21233 + }, + { + "epoch": 1.77, + "learning_rate": 0.00019988853902527378, + "loss": 1.9503, + "step": 21234 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21235 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21236 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21237 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21238 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.0274, + "step": 21239 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.2978, + "step": 21240 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.5789, + "step": 21241 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21242 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 2.1151, + "step": 21243 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21244 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 5.0121, + "step": 21245 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.8148, + "step": 21246 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21247 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 11.1686, + "step": 21248 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 9.2185, + "step": 21249 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 8.6298, + "step": 21250 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21251 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 9.0522, + "step": 21252 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21253 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.7937, + "step": 21254 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 2.1893, + "step": 21255 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 7.251, + "step": 21256 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21257 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21258 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 6.7734, + "step": 21259 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.9587, + "step": 21260 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.6019, + "step": 21261 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21262 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 6.779, + "step": 21263 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 5.3271, + "step": 21264 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21265 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21266 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 5.2094, + "step": 21267 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21268 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21269 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 7.8387, + "step": 21270 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21271 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.9908, + "step": 21272 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21273 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.9947, + "step": 21274 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21275 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.9761, + "step": 21276 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21277 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 6.8412, + "step": 21278 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21279 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21280 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 4.1344, + "step": 21281 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21282 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21283 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.9687, + "step": 21284 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21285 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.9379, + "step": 21286 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 2.2361, + "step": 21287 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 7.3109, + "step": 21288 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 8.457, + "step": 21289 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21290 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21291 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.0495, + "step": 21292 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 8.3338, + "step": 21293 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21294 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.0099, + "step": 21295 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21296 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.9641, + "step": 21297 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.9894, + "step": 21298 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21299 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 4.1398, + "step": 21300 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21301 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 9.8351, + "step": 21302 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21303 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.6462, + "step": 21304 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 4.1663, + "step": 21305 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.9414, + "step": 21306 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 7.4116, + "step": 21307 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21308 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 7.0457, + "step": 21309 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.3497, + "step": 21310 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.948, + "step": 21311 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21312 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21313 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 2.5827, + "step": 21314 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21315 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 5.7641, + "step": 21316 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21317 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21318 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.1218, + "step": 21319 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.9854, + "step": 21320 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21321 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.1328, + "step": 21322 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21323 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 6.8433, + "step": 21324 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.139, + "step": 21325 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 4.8916, + "step": 21326 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.2802, + "step": 21327 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21328 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21329 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.8121, + "step": 21330 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21331 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21332 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.1833, + "step": 21333 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.8366, + "step": 21334 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.6331, + "step": 21335 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21336 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 1.0192, + "step": 21337 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 9.023, + "step": 21338 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 6.0921, + "step": 21339 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.7282, + "step": 21340 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21341 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21342 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 7.1633, + "step": 21343 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21344 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 3.8897, + "step": 21345 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 5.2364, + "step": 21346 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21347 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.7312, + "step": 21348 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 5.8439, + "step": 21349 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21350 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21351 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 6.8916, + "step": 21352 + }, + { + "epoch": 1.78, + "learning_rate": 0.00019988853902527378, + "loss": 4.9725, + "step": 21353 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.3415, + "step": 21354 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 7.8378, + "step": 21355 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.9726, + "step": 21356 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21357 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.9457, + "step": 21358 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.0131, + "step": 21359 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21360 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21361 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 9.9351, + "step": 21362 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21363 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.8426, + "step": 21364 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 7.612, + "step": 21365 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 12.7774, + "step": 21366 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21367 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.8092, + "step": 21368 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 5.4282, + "step": 21369 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21370 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 8.4313, + "step": 21371 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21372 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21373 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 6.5278, + "step": 21374 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.883, + "step": 21375 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21376 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21377 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21378 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.7437, + "step": 21379 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21380 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.0359, + "step": 21381 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21382 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21383 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.8283, + "step": 21384 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.8285, + "step": 21385 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 4.0685, + "step": 21386 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21387 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21388 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21389 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21390 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.0835, + "step": 21391 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.4319, + "step": 21392 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.6568, + "step": 21393 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21394 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21395 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 4.6861, + "step": 21396 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21397 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.1347, + "step": 21398 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21399 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 4.9593, + "step": 21400 + }, + { + "epoch": 1.79, + "eval_loss": NaN, + "eval_runtime": 44.3606, + "eval_samples_per_second": 56.356, + "eval_steps_per_second": 7.056, + "step": 21400 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21401 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.7751, + "step": 21402 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21403 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21404 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 9.3733, + "step": 21405 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.3555, + "step": 21406 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21407 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21408 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.9473, + "step": 21409 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21410 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.7387, + "step": 21411 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21412 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21413 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.9082, + "step": 21414 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.7455, + "step": 21415 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21416 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 8.608, + "step": 21417 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21418 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.9421, + "step": 21419 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21420 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21421 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.8814, + "step": 21422 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 9.1353, + "step": 21423 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.9058, + "step": 21424 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 5.0443, + "step": 21425 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 5.7322, + "step": 21426 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 3.7419, + "step": 21427 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 5.2542, + "step": 21428 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.8185, + "step": 21429 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 7.6219, + "step": 21430 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.0065, + "step": 21431 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21432 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 5.1362, + "step": 21433 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21434 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21435 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 7.3674, + "step": 21436 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 4.5989, + "step": 21437 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 4.0138, + "step": 21438 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.1477, + "step": 21439 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.0972, + "step": 21440 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.0732, + "step": 21441 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 4.2012, + "step": 21442 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21443 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21444 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21445 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21446 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 4.0127, + "step": 21447 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21448 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21449 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21450 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 5.6096, + "step": 21451 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21452 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21453 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21454 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21455 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 8.479, + "step": 21456 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21457 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21458 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 7.2511, + "step": 21459 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21460 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.976, + "step": 21461 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 8.2598, + "step": 21462 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21463 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 1.0234, + "step": 21464 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21465 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21466 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21467 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 2.9118, + "step": 21468 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21469 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 8.4653, + "step": 21470 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 5.8289, + "step": 21471 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21472 + }, + { + "epoch": 1.79, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21473 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 0.7913, + "step": 21474 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21475 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 6.8273, + "step": 21476 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 6.0655, + "step": 21477 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21478 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21479 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 5.6262, + "step": 21480 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 5.3667, + "step": 21481 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 2.4681, + "step": 21482 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21483 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 7.5565, + "step": 21484 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21485 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 0.0, + "step": 21486 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 3.1976, + "step": 21487 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 1.0041, + "step": 21488 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 1.0916, + "step": 21489 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 6.8256, + "step": 21490 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 6.1819, + "step": 21491 + }, + { + "epoch": 1.8, + "learning_rate": 0.00019988853902527378, + "loss": 2.0082, + "step": 21492 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 4.1962, + "step": 21493 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21494 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21495 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21496 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21497 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 5.3989, + "step": 21498 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 3.8809, + "step": 21499 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21500 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.8122, + "step": 21501 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21502 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 2.0747, + "step": 21503 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21504 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.8545, + "step": 21505 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 3.8156, + "step": 21506 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 6.8325, + "step": 21507 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.8216, + "step": 21508 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21509 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21510 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21511 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21512 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21513 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21514 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.9803, + "step": 21515 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.0958, + "step": 21516 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21517 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 5.9759, + "step": 21518 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21519 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21520 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 9.0113, + "step": 21521 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21522 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21523 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21524 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 6.8003, + "step": 21525 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 5.4352, + "step": 21526 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.9817, + "step": 21527 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21528 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21529 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21530 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.0007, + "step": 21531 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21532 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 7.6439, + "step": 21533 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 6.0469, + "step": 21534 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21535 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21536 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21537 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21538 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21539 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 5.3023, + "step": 21540 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21541 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.9877, + "step": 21542 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21543 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 8.5304, + "step": 21544 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21545 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.8933, + "step": 21546 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21547 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.8196, + "step": 21548 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.9309, + "step": 21549 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 4.1437, + "step": 21550 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21551 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 8.1147, + "step": 21552 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.0085, + "step": 21553 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21554 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.2301, + "step": 21555 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 7.1416, + "step": 21556 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 4.358, + "step": 21557 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 4.8408, + "step": 21558 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21559 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21560 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21561 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.015, + "step": 21562 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.9415, + "step": 21563 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.8292, + "step": 21564 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 2.0864, + "step": 21565 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21566 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21567 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 7.8603, + "step": 21568 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21569 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21570 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.0079, + "step": 21571 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21572 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21573 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21574 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.8305, + "step": 21575 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.0198, + "step": 21576 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 4.6423, + "step": 21577 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.8583, + "step": 21578 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 3.097, + "step": 21579 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.9322, + "step": 21580 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21581 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21582 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21583 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 3.9283, + "step": 21584 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 2.0293, + "step": 21585 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 7.6678, + "step": 21586 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21587 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21588 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.5916, + "step": 21589 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 2.2025, + "step": 21590 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 1.9858, + "step": 21591 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 2.9941, + "step": 21592 + }, + { + "epoch": 1.8, + "learning_rate": 0.0001998829659765375, + "loss": 2.0339, + "step": 21593 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21594 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 9.3717, + "step": 21595 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.2629, + "step": 21596 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 5.2118, + "step": 21597 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.9259, + "step": 21598 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.8464, + "step": 21599 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21600 + }, + { + "epoch": 1.81, + "eval_loss": NaN, + "eval_runtime": 45.1498, + "eval_samples_per_second": 55.371, + "eval_steps_per_second": 6.932, + "step": 21600 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21601 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 5.1491, + "step": 21602 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21603 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21604 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.3661, + "step": 21605 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 5.2327, + "step": 21606 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 8.7683, + "step": 21607 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 8.3109, + "step": 21608 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21609 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21610 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.6941, + "step": 21611 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.1213, + "step": 21612 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.876, + "step": 21613 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.0646, + "step": 21614 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21615 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.627, + "step": 21616 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21617 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21618 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21619 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21620 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.9334, + "step": 21621 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21622 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21623 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21624 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.2696, + "step": 21625 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21626 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.9884, + "step": 21627 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21628 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21629 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.5243, + "step": 21630 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.7145, + "step": 21631 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21632 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21633 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.9761, + "step": 21634 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.2314, + "step": 21635 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.6374, + "step": 21636 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.7729, + "step": 21637 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.4348, + "step": 21638 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21639 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.2446, + "step": 21640 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21641 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.4889, + "step": 21642 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.8684, + "step": 21643 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.7278, + "step": 21644 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21645 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.99, + "step": 21646 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.6799, + "step": 21647 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21648 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 5.7023, + "step": 21649 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21650 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21651 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21652 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.7389, + "step": 21653 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21654 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 6.8469, + "step": 21655 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21656 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21657 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21658 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.0984, + "step": 21659 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.8919, + "step": 21660 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.153, + "step": 21661 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.8872, + "step": 21662 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21663 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21664 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.0289, + "step": 21665 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21666 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21667 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.6064, + "step": 21668 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 5.2323, + "step": 21669 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21670 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21671 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21672 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.0714, + "step": 21673 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21674 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21675 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.1978, + "step": 21676 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 6.5231, + "step": 21677 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.9116, + "step": 21678 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21679 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 6.2716, + "step": 21680 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 11.0933, + "step": 21681 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 9.6106, + "step": 21682 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 7.8248, + "step": 21683 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 7.0382, + "step": 21684 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.9947, + "step": 21685 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21686 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.7405, + "step": 21687 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 1.8874, + "step": 21688 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21689 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.9442, + "step": 21690 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21691 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 6.0665, + "step": 21692 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21693 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.8251, + "step": 21694 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 9.202, + "step": 21695 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21696 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 3.2324, + "step": 21697 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21698 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.1448, + "step": 21699 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21700 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21701 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21702 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.7074, + "step": 21703 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 2.0819, + "step": 21704 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 4.2308, + "step": 21705 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21706 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 8.0065, + "step": 21707 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21708 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21709 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21710 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21711 + }, + { + "epoch": 1.81, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21712 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21713 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21714 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21715 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.8532, + "step": 21716 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21717 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.0666, + "step": 21718 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.8535, + "step": 21719 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 8.6621, + "step": 21720 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21721 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.3373, + "step": 21722 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.7819, + "step": 21723 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.2348, + "step": 21724 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.3072, + "step": 21725 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.7267, + "step": 21726 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21727 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.1166, + "step": 21728 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21729 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21730 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.5465, + "step": 21731 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 4.5351, + "step": 21732 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21733 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21734 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.9661, + "step": 21735 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.0652, + "step": 21736 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21737 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21738 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21739 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21740 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.9678, + "step": 21741 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21742 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 4.8021, + "step": 21743 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.7523, + "step": 21744 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 6.9637, + "step": 21745 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21746 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21747 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21748 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 4.9622, + "step": 21749 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 7.0911, + "step": 21750 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21751 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.9332, + "step": 21752 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.5651, + "step": 21753 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.0763, + "step": 21754 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.4555, + "step": 21755 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.8155, + "step": 21756 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 9.4533, + "step": 21757 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 7.7825, + "step": 21758 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21759 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.761, + "step": 21760 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21761 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21762 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21763 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21764 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21765 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21766 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.0527, + "step": 21767 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21768 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.9899, + "step": 21769 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.1919, + "step": 21770 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 7.3576, + "step": 21771 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21772 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.9754, + "step": 21773 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21774 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21775 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21776 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21777 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 4.8121, + "step": 21778 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21779 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 8.9123, + "step": 21780 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.0707, + "step": 21781 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.1919, + "step": 21782 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.9564, + "step": 21783 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21784 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21785 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21786 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21787 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21788 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.3167, + "step": 21789 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.8626, + "step": 21790 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 5.9918, + "step": 21791 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.1994, + "step": 21792 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.0241, + "step": 21793 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21794 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 8.881, + "step": 21795 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.7987, + "step": 21796 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21797 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21798 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 6.6036, + "step": 21799 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.8335, + "step": 21800 + }, + { + "epoch": 1.82, + "eval_loss": NaN, + "eval_runtime": 42.9384, + "eval_samples_per_second": 58.223, + "eval_steps_per_second": 7.29, + "step": 21800 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21801 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21802 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21803 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.8108, + "step": 21804 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 1.4823, + "step": 21805 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21806 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 4.3947, + "step": 21807 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21808 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21809 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 6.0632, + "step": 21810 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.7514, + "step": 21811 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.3174, + "step": 21812 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 5.8681, + "step": 21813 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 7.3763, + "step": 21814 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.9991, + "step": 21815 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21816 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21817 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21818 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.7056, + "step": 21819 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 2.3531, + "step": 21820 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21821 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21822 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21823 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 8.2867, + "step": 21824 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.8125, + "step": 21825 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 9.1216, + "step": 21826 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 3.7613, + "step": 21827 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21828 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21829 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.7761, + "step": 21830 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21831 + }, + { + "epoch": 1.82, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21832 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21833 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21834 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21835 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21836 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.0233, + "step": 21837 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.6376, + "step": 21838 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.8701, + "step": 21839 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.0204, + "step": 21840 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 2.3133, + "step": 21841 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.665, + "step": 21842 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21843 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21844 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21845 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.3315, + "step": 21846 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21847 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21848 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21849 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21850 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21851 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.857, + "step": 21852 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21853 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 4.3146, + "step": 21854 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 6.8383, + "step": 21855 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 2.6733, + "step": 21856 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21857 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21858 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.4787, + "step": 21859 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21860 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21861 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.8758, + "step": 21862 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 4.647, + "step": 21863 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.8768, + "step": 21864 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21865 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21866 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21867 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21868 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.5275, + "step": 21869 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.7309, + "step": 21870 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 4.7039, + "step": 21871 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.87, + "step": 21872 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21873 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.4851, + "step": 21874 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.8312, + "step": 21875 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.2545, + "step": 21876 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.4057, + "step": 21877 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.1527, + "step": 21878 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21879 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.9239, + "step": 21880 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.7517, + "step": 21881 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 4.3792, + "step": 21882 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21883 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21884 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21885 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 9.127, + "step": 21886 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.9626, + "step": 21887 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21888 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.8103, + "step": 21889 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21890 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21891 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 6.1942, + "step": 21892 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21893 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21894 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21895 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21896 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21897 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21898 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21899 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21900 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.8139, + "step": 21901 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21902 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21903 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 8.72, + "step": 21904 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21905 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21906 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21907 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21908 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21909 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.1052, + "step": 21910 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.4972, + "step": 21911 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 2.8927, + "step": 21912 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.386, + "step": 21913 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21914 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21915 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.11, + "step": 21916 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21917 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21918 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.5791, + "step": 21919 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21920 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21921 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21922 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 6.4528, + "step": 21923 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.5341, + "step": 21924 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21925 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 6.7344, + "step": 21926 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 6.688, + "step": 21927 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 4.7189, + "step": 21928 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21929 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21930 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.8984, + "step": 21931 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.9027, + "step": 21932 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21933 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21934 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.0095, + "step": 21935 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21936 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.8053, + "step": 21937 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21938 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.0518, + "step": 21939 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 8.4701, + "step": 21940 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21941 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 3.4604, + "step": 21942 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.7899, + "step": 21943 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.4858, + "step": 21944 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21945 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 7.555, + "step": 21946 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21947 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 4.6587, + "step": 21948 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.7, + "step": 21949 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 1.792, + "step": 21950 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 5.1139, + "step": 21951 + }, + { + "epoch": 1.83, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21952 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.937, + "step": 21953 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21954 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21955 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 7.045, + "step": 21956 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.0581, + "step": 21957 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21958 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21959 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 6.4515, + "step": 21960 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21961 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.9819, + "step": 21962 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.7445, + "step": 21963 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21964 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21965 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21966 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21967 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21968 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.5947, + "step": 21969 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.3134, + "step": 21970 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21971 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21972 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21973 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.9077, + "step": 21974 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21975 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.3195, + "step": 21976 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21977 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 5.6041, + "step": 21978 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21979 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.8839, + "step": 21980 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.0914, + "step": 21981 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.8608, + "step": 21982 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.9082, + "step": 21983 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 9.5341, + "step": 21984 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.9448, + "step": 21985 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21986 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.9837, + "step": 21987 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.9749, + "step": 21988 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21989 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21990 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.2545, + "step": 21991 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 7.5603, + "step": 21992 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.291, + "step": 21993 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21994 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21995 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21996 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 5.7882, + "step": 21997 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.9103, + "step": 21998 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 21999 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.8908, + "step": 22000 + }, + { + "epoch": 1.84, + "eval_loss": NaN, + "eval_runtime": 47.0782, + "eval_samples_per_second": 53.103, + "eval_steps_per_second": 6.649, + "step": 22000 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22001 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 6.2917, + "step": 22002 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22003 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.3641, + "step": 22004 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22005 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.9721, + "step": 22006 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.3786, + "step": 22007 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 7.9635, + "step": 22008 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.8945, + "step": 22009 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.0003, + "step": 22010 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.0137, + "step": 22011 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.2548, + "step": 22012 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.4797, + "step": 22013 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.674, + "step": 22014 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22015 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22016 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.6894, + "step": 22017 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22018 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22019 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 5.5743, + "step": 22020 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22021 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22022 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 8.6645, + "step": 22023 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 5.6193, + "step": 22024 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22025 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.1794, + "step": 22026 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22027 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.1445, + "step": 22028 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.1074, + "step": 22029 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 11.4671, + "step": 22030 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.6569, + "step": 22031 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22032 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 7.9459, + "step": 22033 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.0773, + "step": 22034 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.1452, + "step": 22035 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22036 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22037 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22038 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22039 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.1546, + "step": 22040 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.9535, + "step": 22041 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22042 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.9305, + "step": 22043 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.2608, + "step": 22044 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.9585, + "step": 22045 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22046 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.7185, + "step": 22047 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 5.4064, + "step": 22048 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22049 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.6305, + "step": 22050 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.4103, + "step": 22051 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.1104, + "step": 22052 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 8.857, + "step": 22053 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 1.7545, + "step": 22054 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.0101, + "step": 22055 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 4.6314, + "step": 22056 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.9737, + "step": 22057 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22058 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.2668, + "step": 22059 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 5.3399, + "step": 22060 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22061 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22062 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.3858, + "step": 22063 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22064 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.3013, + "step": 22065 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22066 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22067 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22068 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 2.8772, + "step": 22069 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 3.4982, + "step": 22070 + }, + { + "epoch": 1.84, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22071 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 3.4453, + "step": 22072 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 3.6179, + "step": 22073 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.9546, + "step": 22074 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 5.1775, + "step": 22075 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 3.1568, + "step": 22076 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.9084, + "step": 22077 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22078 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22079 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.9588, + "step": 22080 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.9346, + "step": 22081 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.2413, + "step": 22082 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.213, + "step": 22083 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.9962, + "step": 22084 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.3421, + "step": 22085 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.7988, + "step": 22086 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22087 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22088 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22089 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22090 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.1018, + "step": 22091 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.2926, + "step": 22092 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22093 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.1712, + "step": 22094 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.2961, + "step": 22095 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22096 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.6906, + "step": 22097 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22098 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.3553, + "step": 22099 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 8.9255, + "step": 22100 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22101 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22102 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.8219, + "step": 22103 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22104 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22105 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.9089, + "step": 22106 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.2516, + "step": 22107 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.7361, + "step": 22108 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22109 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.2153, + "step": 22110 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.1022, + "step": 22111 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.3382, + "step": 22112 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22113 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.9451, + "step": 22114 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22115 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22116 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.9222, + "step": 22117 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22118 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.2003, + "step": 22119 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.0154, + "step": 22120 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22121 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.3429, + "step": 22122 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22123 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22124 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22125 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22126 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.776, + "step": 22127 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22128 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.3492, + "step": 22129 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 3.2795, + "step": 22130 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22131 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22132 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22133 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22134 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.3484, + "step": 22135 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22136 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 3.1088, + "step": 22137 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22138 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 8.1389, + "step": 22139 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.812, + "step": 22140 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.3825, + "step": 22141 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.2538, + "step": 22142 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.9738, + "step": 22143 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.0686, + "step": 22144 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.0845, + "step": 22145 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22146 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22147 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.0371, + "step": 22148 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22149 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.6661, + "step": 22150 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22151 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.8269, + "step": 22152 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 8.2332, + "step": 22153 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22154 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 3.9174, + "step": 22155 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22156 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.5466, + "step": 22157 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.3223, + "step": 22158 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 7.1834, + "step": 22159 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22160 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22161 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.2414, + "step": 22162 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 8.6602, + "step": 22163 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.7521, + "step": 22164 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.6035, + "step": 22165 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22166 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22167 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22168 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22169 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22170 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 1.9775, + "step": 22171 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22172 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 8.5261, + "step": 22173 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.4216, + "step": 22174 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.0218, + "step": 22175 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 2.5288, + "step": 22176 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 6.6452, + "step": 22177 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22178 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22179 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.3314, + "step": 22180 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22181 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22182 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22183 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22184 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 8.3118, + "step": 22185 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 4.6228, + "step": 22186 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22187 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22188 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22189 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 6.9585, + "step": 22190 + }, + { + "epoch": 1.85, + "learning_rate": 0.0001998829659765375, + "loss": 10.5814, + "step": 22191 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 9.9805, + "step": 22192 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.0644, + "step": 22193 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22194 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22195 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22196 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.2557, + "step": 22197 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.2359, + "step": 22198 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 4.7844, + "step": 22199 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22200 + }, + { + "epoch": 1.86, + "eval_loss": NaN, + "eval_runtime": 43.4337, + "eval_samples_per_second": 57.559, + "eval_steps_per_second": 7.206, + "step": 22200 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22201 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22202 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22203 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.1201, + "step": 22204 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22205 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22206 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 8.8319, + "step": 22207 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 6.4734, + "step": 22208 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22209 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22210 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.8396, + "step": 22211 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.1948, + "step": 22212 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.3581, + "step": 22213 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22214 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.0472, + "step": 22215 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.0425, + "step": 22216 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22217 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.3619, + "step": 22218 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22219 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.944, + "step": 22220 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.8629, + "step": 22221 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22222 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22223 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22224 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 6.8705, + "step": 22225 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22226 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22227 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22228 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.9556, + "step": 22229 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.9691, + "step": 22230 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22231 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 5.6872, + "step": 22232 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 6.691, + "step": 22233 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22234 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22235 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.7924, + "step": 22236 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22237 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.7927, + "step": 22238 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 8.4058, + "step": 22239 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 4.3083, + "step": 22240 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.0873, + "step": 22241 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22242 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.1645, + "step": 22243 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22244 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22245 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.9227, + "step": 22246 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.6557, + "step": 22247 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22248 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22249 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.0544, + "step": 22250 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 5.6839, + "step": 22251 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.0515, + "step": 22252 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.3676, + "step": 22253 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.8062, + "step": 22254 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.41, + "step": 22255 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22256 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.1493, + "step": 22257 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22258 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 5.049, + "step": 22259 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.6419, + "step": 22260 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.0428, + "step": 22261 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22262 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.3555, + "step": 22263 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.3532, + "step": 22264 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22265 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22266 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22267 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22268 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 4.1333, + "step": 22269 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22270 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.964, + "step": 22271 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.0825, + "step": 22272 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 8.4897, + "step": 22273 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22274 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.7927, + "step": 22275 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.8468, + "step": 22276 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.7954, + "step": 22277 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22278 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22279 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.73, + "step": 22280 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22281 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.0962, + "step": 22282 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.9997, + "step": 22283 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 7.826, + "step": 22284 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.2514, + "step": 22285 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.7857, + "step": 22286 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22287 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 3.9981, + "step": 22288 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 1.0017, + "step": 22289 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.5728, + "step": 22290 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 4.4525, + "step": 22291 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22292 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22293 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22294 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22295 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.8924, + "step": 22296 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22297 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.502, + "step": 22298 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22299 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 7.1168, + "step": 22300 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22301 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22302 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22303 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.8484, + "step": 22304 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22305 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 2.1082, + "step": 22306 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 5.2547, + "step": 22307 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22308 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22309 + }, + { + "epoch": 1.86, + "learning_rate": 0.0001998829659765375, + "loss": 4.3416, + "step": 22310 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.0217, + "step": 22311 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.757, + "step": 22312 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22313 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.6678, + "step": 22314 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22315 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22316 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.8883, + "step": 22317 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22318 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.2934, + "step": 22319 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22320 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.998, + "step": 22321 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22322 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22323 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22324 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.6448, + "step": 22325 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.082, + "step": 22326 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.9556, + "step": 22327 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22328 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.9483, + "step": 22329 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22330 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 8.1693, + "step": 22331 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.2749, + "step": 22332 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.7758, + "step": 22333 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22334 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.0358, + "step": 22335 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.083, + "step": 22336 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22337 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.354, + "step": 22338 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22339 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22340 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22341 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22342 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.8035, + "step": 22343 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22344 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.7962, + "step": 22345 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22346 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22347 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.0009, + "step": 22348 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 9.0514, + "step": 22349 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.0862, + "step": 22350 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 8.2523, + "step": 22351 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 9.1276, + "step": 22352 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22353 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 8.0365, + "step": 22354 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22355 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22356 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.8706, + "step": 22357 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22358 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22359 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.0, + "step": 22360 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 7.4915, + "step": 22361 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.8514, + "step": 22362 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.1122, + "step": 22363 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22364 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.4749, + "step": 22365 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22366 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 8.1278, + "step": 22367 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22368 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22369 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.9226, + "step": 22370 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.2884, + "step": 22371 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22372 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.9221, + "step": 22373 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.5439, + "step": 22374 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.8867, + "step": 22375 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 5.2307, + "step": 22376 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22377 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22378 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.0759, + "step": 22379 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.9351, + "step": 22380 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22381 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.2518, + "step": 22382 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.8393, + "step": 22383 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 6.9077, + "step": 22384 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 9.019, + "step": 22385 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22386 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 12.4519, + "step": 22387 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 7.5548, + "step": 22388 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 6.0675, + "step": 22389 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.536, + "step": 22390 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.1393, + "step": 22391 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.8091, + "step": 22392 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22393 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.3559, + "step": 22394 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.0376, + "step": 22395 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22396 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 7.7652, + "step": 22397 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.9582, + "step": 22398 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22399 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 5.0536, + "step": 22400 + }, + { + "epoch": 1.87, + "eval_loss": NaN, + "eval_runtime": 44.4341, + "eval_samples_per_second": 56.263, + "eval_steps_per_second": 7.044, + "step": 22400 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.7979, + "step": 22401 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 12.0747, + "step": 22402 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 5.6634, + "step": 22403 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22404 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 2.1137, + "step": 22405 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22406 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22407 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22408 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.7018, + "step": 22409 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22410 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22411 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22412 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 3.9273, + "step": 22413 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.4786, + "step": 22414 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22415 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22416 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22417 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 9.6577, + "step": 22418 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.2393, + "step": 22419 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.0421, + "step": 22420 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22421 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 7.8981, + "step": 22422 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 1.6786, + "step": 22423 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22424 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22425 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22426 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22427 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22428 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 4.2882, + "step": 22429 + }, + { + "epoch": 1.87, + "learning_rate": 0.0001998829659765375, + "loss": 8.3392, + "step": 22430 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22431 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 6.3483, + "step": 22432 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22433 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.1912, + "step": 22434 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 9.2076, + "step": 22435 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22436 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.1315, + "step": 22437 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.2848, + "step": 22438 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22439 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.2074, + "step": 22440 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 6.1686, + "step": 22441 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 5.0201, + "step": 22442 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22443 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.8323, + "step": 22444 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.5907, + "step": 22445 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 5.9515, + "step": 22446 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 6.3988, + "step": 22447 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22448 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22449 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22450 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.3395, + "step": 22451 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.3645, + "step": 22452 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22453 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.0516, + "step": 22454 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.0991, + "step": 22455 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22456 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22457 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.7181, + "step": 22458 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.9879, + "step": 22459 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 9.3049, + "step": 22460 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.8524, + "step": 22461 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 3.2529, + "step": 22462 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 7.8654, + "step": 22463 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.0906, + "step": 22464 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22465 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.0593, + "step": 22466 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22467 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22468 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.9581, + "step": 22469 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 8.4044, + "step": 22470 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22471 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.182, + "step": 22472 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22473 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22474 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22475 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.2073, + "step": 22476 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 5.7488, + "step": 22477 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.3007, + "step": 22478 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.9913, + "step": 22479 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22480 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 3.1504, + "step": 22481 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22482 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22483 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.932, + "step": 22484 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.9918, + "step": 22485 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.8942, + "step": 22486 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 7.2565, + "step": 22487 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22488 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22489 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.9992, + "step": 22490 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22491 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.8689, + "step": 22492 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 5.4702, + "step": 22493 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 7.4922, + "step": 22494 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22495 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.5516, + "step": 22496 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22497 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.9771, + "step": 22498 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22499 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 8.3225, + "step": 22500 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22501 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22502 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 7.4696, + "step": 22503 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22504 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.3751, + "step": 22505 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22506 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22507 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 9.2668, + "step": 22508 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 6.6026, + "step": 22509 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.1981, + "step": 22510 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.3438, + "step": 22511 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 3.9079, + "step": 22512 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.0339, + "step": 22513 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22514 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 7.7023, + "step": 22515 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22516 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22517 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22518 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.9748, + "step": 22519 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22520 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 2.968, + "step": 22521 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.1793, + "step": 22522 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 5.8065, + "step": 22523 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 8.3491, + "step": 22524 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22525 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22526 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22527 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 3.9038, + "step": 22528 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22529 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 3.7788, + "step": 22530 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22531 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 3.7606, + "step": 22532 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.9757, + "step": 22533 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22534 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22535 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22536 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.0624, + "step": 22537 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22538 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22539 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22540 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22541 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 1.0261, + "step": 22542 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 4.4652, + "step": 22543 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22544 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 3.4756, + "step": 22545 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22546 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 7.5608, + "step": 22547 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 5.8151, + "step": 22548 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22549 + }, + { + "epoch": 1.88, + "learning_rate": 0.0001998829659765375, + "loss": 0.9011, + "step": 22550 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 8.0434, + "step": 22551 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22552 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22553 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 2.0166, + "step": 22554 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.2283, + "step": 22555 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.812, + "step": 22556 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 2.0941, + "step": 22557 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 2.8142, + "step": 22558 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.8302, + "step": 22559 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22560 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22561 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 10.1853, + "step": 22562 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.2813, + "step": 22563 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22564 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 4.8737, + "step": 22565 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.1671, + "step": 22566 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.5891, + "step": 22567 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 5.7214, + "step": 22568 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22569 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22570 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22571 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 9.9659, + "step": 22572 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.207, + "step": 22573 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.1803, + "step": 22574 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 8.9186, + "step": 22575 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22576 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22577 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.7404, + "step": 22578 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22579 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22580 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.0229, + "step": 22581 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.3571, + "step": 22582 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22583 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22584 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.1737, + "step": 22585 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.5845, + "step": 22586 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 4.2616, + "step": 22587 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22588 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22589 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22590 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.032, + "step": 22591 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.8699, + "step": 22592 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.6449, + "step": 22593 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22594 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.9627, + "step": 22595 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 8.3661, + "step": 22596 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 4.0344, + "step": 22597 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.0789, + "step": 22598 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22599 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22600 + }, + { + "epoch": 1.89, + "eval_loss": NaN, + "eval_runtime": 44.0383, + "eval_samples_per_second": 56.769, + "eval_steps_per_second": 7.107, + "step": 22600 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.6116, + "step": 22601 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22602 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22603 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22604 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22605 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22606 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.1828, + "step": 22607 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 5.5988, + "step": 22608 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.7738, + "step": 22609 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.0172, + "step": 22610 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 4.6299, + "step": 22611 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22612 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 5.4916, + "step": 22613 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 6.6216, + "step": 22614 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.9924, + "step": 22615 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 8.2657, + "step": 22616 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 5.3753, + "step": 22617 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.0899, + "step": 22618 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22619 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22620 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 5.7863, + "step": 22621 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22622 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22623 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22624 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22625 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.8156, + "step": 22626 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22627 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22628 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.5049, + "step": 22629 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.3573, + "step": 22630 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22631 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.7919, + "step": 22632 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22633 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22634 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22635 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.9892, + "step": 22636 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 4.7188, + "step": 22637 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22638 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22639 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22640 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 10.2227, + "step": 22641 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.0792, + "step": 22642 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22643 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22644 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 4.4096, + "step": 22645 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 4.1805, + "step": 22646 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 2.0754, + "step": 22647 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.9975, + "step": 22648 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22649 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.8355, + "step": 22650 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22651 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22652 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 7.8852, + "step": 22653 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 3.0636, + "step": 22654 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22655 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 8.9847, + "step": 22656 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.9332, + "step": 22657 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22658 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 18.2224, + "step": 22659 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22660 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22661 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 8.577, + "step": 22662 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22663 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 1.0947, + "step": 22664 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 2.0716, + "step": 22665 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 2.5832, + "step": 22666 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22667 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22668 + }, + { + "epoch": 1.89, + "learning_rate": 0.0001998829659765375, + "loss": 2.2963, + "step": 22669 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.1568, + "step": 22670 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.911, + "step": 22671 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.6471, + "step": 22672 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22673 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 8.1666, + "step": 22674 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22675 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 5.9375, + "step": 22676 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 5.7039, + "step": 22677 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 9.4276, + "step": 22678 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 5.2646, + "step": 22679 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.0132, + "step": 22680 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22681 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.968, + "step": 22682 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.6261, + "step": 22683 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 8.3372, + "step": 22684 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22685 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22686 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.8978, + "step": 22687 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.1382, + "step": 22688 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.6118, + "step": 22689 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.1599, + "step": 22690 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22691 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.8597, + "step": 22692 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.0774, + "step": 22693 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.6656, + "step": 22694 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22695 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22696 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.5025, + "step": 22697 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.5807, + "step": 22698 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.2161, + "step": 22699 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.6617, + "step": 22700 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22701 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.9947, + "step": 22702 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22703 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22704 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22705 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22706 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.312, + "step": 22707 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 2.3723, + "step": 22708 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 2.1224, + "step": 22709 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22710 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 6.2829, + "step": 22711 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.7748, + "step": 22712 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22713 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 8.7134, + "step": 22714 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.9638, + "step": 22715 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22716 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.2741, + "step": 22717 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.2915, + "step": 22718 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.416, + "step": 22719 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22720 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22721 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22722 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22723 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22724 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.5182, + "step": 22725 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.9375, + "step": 22726 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22727 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22728 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.4342, + "step": 22729 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.1504, + "step": 22730 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22731 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.6085, + "step": 22732 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.9403, + "step": 22733 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22734 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.85, + "step": 22735 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 5.0667, + "step": 22736 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.6902, + "step": 22737 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.4677, + "step": 22738 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 2.0724, + "step": 22739 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22740 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22741 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.133, + "step": 22742 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22743 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 5.3111, + "step": 22744 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.5563, + "step": 22745 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22746 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.7209, + "step": 22747 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.562, + "step": 22748 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22749 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.5315, + "step": 22750 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22751 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22752 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22753 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.0993, + "step": 22754 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22755 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22756 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22757 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.2193, + "step": 22758 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 2.744, + "step": 22759 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.0621, + "step": 22760 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.1267, + "step": 22761 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.315, + "step": 22762 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22763 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 5.3382, + "step": 22764 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22765 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22766 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22767 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 2.2245, + "step": 22768 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 10.4891, + "step": 22769 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22770 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22771 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 7.137, + "step": 22772 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 1.0149, + "step": 22773 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 11.1678, + "step": 22774 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22775 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22776 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.4408, + "step": 22777 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.9778, + "step": 22778 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.5697, + "step": 22779 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22780 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22781 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22782 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.8957, + "step": 22783 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.9004, + "step": 22784 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.7388, + "step": 22785 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22786 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.2571, + "step": 22787 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 4.5249, + "step": 22788 + }, + { + "epoch": 1.9, + "learning_rate": 0.0001998829659765375, + "loss": 3.8061, + "step": 22789 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22790 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22791 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 1.1734, + "step": 22792 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22793 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22794 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22795 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22796 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 5.5226, + "step": 22797 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 4.2495, + "step": 22798 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 13.0341, + "step": 22799 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22800 + }, + { + "epoch": 1.91, + "eval_loss": NaN, + "eval_runtime": 44.1374, + "eval_samples_per_second": 56.641, + "eval_steps_per_second": 7.091, + "step": 22800 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22801 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 4.4909, + "step": 22802 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 2.7302, + "step": 22803 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 2.1622, + "step": 22804 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.385, + "step": 22805 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22806 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.9617, + "step": 22807 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 6.4016, + "step": 22808 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 3.9384, + "step": 22809 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22810 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 1.6925, + "step": 22811 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 2.0979, + "step": 22812 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 1.8453, + "step": 22813 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22814 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22815 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22816 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22817 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22818 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 8.3178, + "step": 22819 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22820 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.7748, + "step": 22821 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 2.3261, + "step": 22822 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22823 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 3.1232, + "step": 22824 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 2.7825, + "step": 22825 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22826 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 8.2046, + "step": 22827 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22828 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 4.2361, + "step": 22829 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22830 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22831 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 9.1596, + "step": 22832 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 4.9918, + "step": 22833 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 8.7878, + "step": 22834 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22835 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 4.7334, + "step": 22836 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 4.5898, + "step": 22837 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 7.7278, + "step": 22838 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22839 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22840 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22841 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 4.5628, + "step": 22842 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22843 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 3.7899, + "step": 22844 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22845 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22846 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22847 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22848 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 8.1612, + "step": 22849 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22850 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.9022, + "step": 22851 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 1.9007, + "step": 22852 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 5.6926, + "step": 22853 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22854 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 3.7008, + "step": 22855 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 8.5343, + "step": 22856 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22857 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 3.749, + "step": 22858 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.8288, + "step": 22859 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 1.5419, + "step": 22860 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 2.2728, + "step": 22861 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 1.87, + "step": 22862 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 6.4798, + "step": 22863 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.8621, + "step": 22864 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 1.0461, + "step": 22865 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22866 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22867 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 3.794, + "step": 22868 + }, + { + "epoch": 1.91, + "learning_rate": 0.0001998829659765375, + "loss": 0.0, + "step": 22869 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 3.709, + "step": 22870 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 2.172, + "step": 22871 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22872 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22873 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 4.4427, + "step": 22874 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 4.359, + "step": 22875 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22876 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22877 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22878 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.9325, + "step": 22879 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22880 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 8.3965, + "step": 22881 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 4.3749, + "step": 22882 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 9.9455, + "step": 22883 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22884 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22885 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22886 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22887 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22888 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 6.3905, + "step": 22889 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22890 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22891 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 4.3711, + "step": 22892 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22893 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 1.0501, + "step": 22894 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 1.0147, + "step": 22895 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22896 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22897 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22898 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 2.3979, + "step": 22899 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 1.803, + "step": 22900 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22901 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 1.9779, + "step": 22902 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22903 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22904 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 3.2829, + "step": 22905 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 7.3204, + "step": 22906 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22907 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 4.5094, + "step": 22908 + }, + { + "epoch": 1.91, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22909 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.0665, + "step": 22910 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.548, + "step": 22911 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22912 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.7261, + "step": 22913 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.1084, + "step": 22914 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22915 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22916 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.973, + "step": 22917 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.8373, + "step": 22918 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.9427, + "step": 22919 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.8426, + "step": 22920 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.5612, + "step": 22921 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.9382, + "step": 22922 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 5.1455, + "step": 22923 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22924 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 2.0326, + "step": 22925 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22926 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.0736, + "step": 22927 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.8997, + "step": 22928 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 4.198, + "step": 22929 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.7588, + "step": 22930 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.1164, + "step": 22931 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.0768, + "step": 22932 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22933 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22934 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.1691, + "step": 22935 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22936 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22937 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 6.8315, + "step": 22938 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22939 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 11.0339, + "step": 22940 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.8157, + "step": 22941 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.7994, + "step": 22942 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 2.0413, + "step": 22943 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.737, + "step": 22944 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 5.274, + "step": 22945 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22946 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.6883, + "step": 22947 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 4.8823, + "step": 22948 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 2.0956, + "step": 22949 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.3017, + "step": 22950 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22951 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22952 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 2.0708, + "step": 22953 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 5.9923, + "step": 22954 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.9749, + "step": 22955 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 6.0315, + "step": 22956 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.575, + "step": 22957 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.5334, + "step": 22958 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22959 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 6.1966, + "step": 22960 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 2.101, + "step": 22961 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22962 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22963 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22964 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.8043, + "step": 22965 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 4.0581, + "step": 22966 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22967 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.5868, + "step": 22968 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 7.3731, + "step": 22969 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 5.0176, + "step": 22970 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 9.0517, + "step": 22971 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.1004, + "step": 22972 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 7.7242, + "step": 22973 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22974 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.9655, + "step": 22975 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22976 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22977 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.9405, + "step": 22978 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.4959, + "step": 22979 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22980 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22981 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.7977, + "step": 22982 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22983 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22984 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22985 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22986 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.5199, + "step": 22987 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22988 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22989 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22990 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.125, + "step": 22991 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22992 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22993 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22994 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.7996, + "step": 22995 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22996 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22997 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22998 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 22999 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23000 + }, + { + "epoch": 1.92, + "eval_loss": NaN, + "eval_runtime": 44.1562, + "eval_samples_per_second": 56.617, + "eval_steps_per_second": 7.088, + "step": 23000 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.9093, + "step": 23001 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23002 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23003 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23004 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23005 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 7.5208, + "step": 23006 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23007 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 2.7503, + "step": 23008 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23009 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23010 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 1.0337, + "step": 23011 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 8.2727, + "step": 23012 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23013 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 7.0998, + "step": 23014 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.4994, + "step": 23015 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 4.4475, + "step": 23016 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 9.819, + "step": 23017 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 4.3444, + "step": 23018 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23019 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 6.9185, + "step": 23020 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23021 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23022 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23023 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 2.1242, + "step": 23024 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23025 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 3.6792, + "step": 23026 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 16.5056, + "step": 23027 + }, + { + "epoch": 1.92, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23028 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 8.619, + "step": 23029 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.0081, + "step": 23030 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.9402, + "step": 23031 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23032 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23033 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.999, + "step": 23034 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23035 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23036 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 6.2784, + "step": 23037 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23038 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 3.1259, + "step": 23039 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.0369, + "step": 23040 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23041 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 7.0796, + "step": 23042 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.9358, + "step": 23043 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23044 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23045 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 3.9539, + "step": 23046 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23047 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23048 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.0231, + "step": 23049 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 3.1803, + "step": 23050 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23051 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23052 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 3.0841, + "step": 23053 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 4.9805, + "step": 23054 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23055 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 6.7643, + "step": 23056 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23057 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23058 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23059 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 7.8712, + "step": 23060 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 4.3201, + "step": 23061 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23062 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 5.2972, + "step": 23063 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 6.1276, + "step": 23064 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23065 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 4.068, + "step": 23066 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.9878, + "step": 23067 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.6839, + "step": 23068 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 5.2085, + "step": 23069 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 7.9497, + "step": 23070 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 4.1905, + "step": 23071 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.0455, + "step": 23072 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23073 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23074 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.2571, + "step": 23075 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 7.0747, + "step": 23076 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 5.9119, + "step": 23077 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23078 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 8.4304, + "step": 23079 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23080 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.6266, + "step": 23081 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 4.3326, + "step": 23082 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23083 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23084 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 2.1473, + "step": 23085 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23086 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 2.2702, + "step": 23087 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 4.2631, + "step": 23088 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 2.0187, + "step": 23089 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.0026, + "step": 23090 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 3.8221, + "step": 23091 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 8.1622, + "step": 23092 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.0544, + "step": 23093 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.8903, + "step": 23094 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23095 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 3.4291, + "step": 23096 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 3.5745, + "step": 23097 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23098 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.9212, + "step": 23099 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23100 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.0102, + "step": 23101 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 2.3049, + "step": 23102 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23103 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.9504, + "step": 23104 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23105 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 7.8927, + "step": 23106 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 1.4446, + "step": 23107 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23108 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23109 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987739292780116, + "loss": 0.0, + "step": 23110 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 3.6973, + "step": 23111 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 2.0406, + "step": 23112 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 1.9006, + "step": 23113 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 3.4241, + "step": 23114 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 1.8515, + "step": 23115 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 4.15, + "step": 23116 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 8.9236, + "step": 23117 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 5.3466, + "step": 23118 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 1.1497, + "step": 23119 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 8.3219, + "step": 23120 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 3.7486, + "step": 23121 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 2.3863, + "step": 23122 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23123 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23124 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23125 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 6.3876, + "step": 23126 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23127 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 2.1463, + "step": 23128 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23129 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.9587, + "step": 23130 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 4.2742, + "step": 23131 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 2.277, + "step": 23132 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 1.6785, + "step": 23133 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23134 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 1.9681, + "step": 23135 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 6.3261, + "step": 23136 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 6.571, + "step": 23137 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23138 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23139 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 7.2257, + "step": 23140 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23141 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 4.6205, + "step": 23142 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23143 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23144 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23145 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23146 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23147 + }, + { + "epoch": 1.93, + "learning_rate": 0.00019987181987906485, + "loss": 8.3406, + "step": 23148 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23149 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23150 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23151 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23152 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.0853, + "step": 23153 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23154 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23155 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.9539, + "step": 23156 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23157 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 5.0326, + "step": 23158 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.385, + "step": 23159 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23160 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23161 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.6161, + "step": 23162 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.3649, + "step": 23163 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 5.7436, + "step": 23164 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.1177, + "step": 23165 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23166 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23167 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23168 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23169 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23170 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 10.4738, + "step": 23171 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23172 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 5.9588, + "step": 23173 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.2001, + "step": 23174 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23175 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.5652, + "step": 23176 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.2765, + "step": 23177 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.7485, + "step": 23178 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.7616, + "step": 23179 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.8233, + "step": 23180 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23181 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23182 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.5759, + "step": 23183 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.4752, + "step": 23184 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 9.1195, + "step": 23185 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23186 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23187 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.039, + "step": 23188 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 5.1587, + "step": 23189 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.6767, + "step": 23190 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23191 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.8385, + "step": 23192 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.9865, + "step": 23193 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 9.2076, + "step": 23194 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23195 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23196 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23197 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.7981, + "step": 23198 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23199 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.6875, + "step": 23200 + }, + { + "epoch": 1.94, + "eval_loss": NaN, + "eval_runtime": 46.42, + "eval_samples_per_second": 53.856, + "eval_steps_per_second": 6.743, + "step": 23200 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23201 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23202 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23203 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 8.0627, + "step": 23204 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 5.5464, + "step": 23205 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.8701, + "step": 23206 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.6335, + "step": 23207 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23208 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.5847, + "step": 23209 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23210 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.7817, + "step": 23211 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.1471, + "step": 23212 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.3592, + "step": 23213 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.9138, + "step": 23214 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.4531, + "step": 23215 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 12.2359, + "step": 23216 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.9542, + "step": 23217 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23218 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23219 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23220 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23221 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.9806, + "step": 23222 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.5032, + "step": 23223 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23224 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.7177, + "step": 23225 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23226 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 6.5403, + "step": 23227 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 7.8821, + "step": 23228 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.488, + "step": 23229 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 5.504, + "step": 23230 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 10.3447, + "step": 23231 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23232 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23233 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.4243, + "step": 23234 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 5.3497, + "step": 23235 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23236 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 6.8166, + "step": 23237 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23238 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23239 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.9722, + "step": 23240 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23241 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 1.8956, + "step": 23242 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 10.6034, + "step": 23243 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.1488, + "step": 23244 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.5677, + "step": 23245 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23246 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23247 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23248 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23249 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 2.9474, + "step": 23250 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23251 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23252 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 4.8841, + "step": 23253 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 3.8116, + "step": 23254 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23255 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23256 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23257 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23258 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.9609, + "step": 23259 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 7.6288, + "step": 23260 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23261 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 7.3817, + "step": 23262 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 8.6965, + "step": 23263 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 6.4822, + "step": 23264 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23265 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23266 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 7.0467, + "step": 23267 + }, + { + "epoch": 1.94, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23268 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.9381, + "step": 23269 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.7699, + "step": 23270 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 6.2286, + "step": 23271 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.0685, + "step": 23272 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 7.6932, + "step": 23273 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.1864, + "step": 23274 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23275 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.0504, + "step": 23276 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 2.4496, + "step": 23277 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.6619, + "step": 23278 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23279 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.9713, + "step": 23280 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23281 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.8791, + "step": 23282 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23283 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 4.286, + "step": 23284 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 8.0528, + "step": 23285 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.0738, + "step": 23286 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 7.4526, + "step": 23287 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.9066, + "step": 23288 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23289 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23290 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.9991, + "step": 23291 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.9526, + "step": 23292 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23293 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23294 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23295 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 6.6223, + "step": 23296 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23297 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23298 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23299 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.8178, + "step": 23300 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23301 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23302 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 15.1814, + "step": 23303 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23304 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23305 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 7.3659, + "step": 23306 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 4.4277, + "step": 23307 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23308 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23309 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23310 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 3.8689, + "step": 23311 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.0383, + "step": 23312 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 3.743, + "step": 23313 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.9893, + "step": 23314 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.0696, + "step": 23315 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 6.1337, + "step": 23316 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23317 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 2.1657, + "step": 23318 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 3.7107, + "step": 23319 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23320 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23321 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 8.4825, + "step": 23322 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23323 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23324 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23325 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23326 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 9.118, + "step": 23327 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23328 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 4.4625, + "step": 23329 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23330 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.6196, + "step": 23331 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 2.6283, + "step": 23332 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23333 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 4.4007, + "step": 23334 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 8.4416, + "step": 23335 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 9.8792, + "step": 23336 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.8699, + "step": 23337 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23338 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23339 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.0785, + "step": 23340 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23341 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.9765, + "step": 23342 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.8815, + "step": 23343 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.4299, + "step": 23344 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 6.236, + "step": 23345 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 2.4002, + "step": 23346 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 16.0593, + "step": 23347 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 4.5616, + "step": 23348 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 4.3768, + "step": 23349 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23350 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23351 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23352 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23353 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 3.3257, + "step": 23354 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 9.6892, + "step": 23355 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23356 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 8.7844, + "step": 23357 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.2091, + "step": 23358 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23359 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23360 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.7123, + "step": 23361 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.9429, + "step": 23362 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23363 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23364 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.1484, + "step": 23365 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 7.4764, + "step": 23366 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23367 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 3.8238, + "step": 23368 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 5.3263, + "step": 23369 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 2.0053, + "step": 23370 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23371 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23372 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 3.2587, + "step": 23373 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23374 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.2584, + "step": 23375 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23376 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23377 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 7.5598, + "step": 23378 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.5989, + "step": 23379 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.92, + "step": 23380 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23381 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23382 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23383 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23384 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 1.0346, + "step": 23385 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 2.1931, + "step": 23386 + }, + { + "epoch": 1.95, + "learning_rate": 0.00019987181987906485, + "loss": 0.8693, + "step": 23387 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23388 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23389 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23390 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.9237, + "step": 23391 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23392 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23393 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.9058, + "step": 23394 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23395 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.7374, + "step": 23396 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.9645, + "step": 23397 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 8.671, + "step": 23398 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 7.65, + "step": 23399 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23400 + }, + { + "epoch": 1.96, + "eval_loss": NaN, + "eval_runtime": 44.7763, + "eval_samples_per_second": 55.833, + "eval_steps_per_second": 6.99, + "step": 23400 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.5116, + "step": 23401 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23402 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 11.3176, + "step": 23403 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.4984, + "step": 23404 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.4604, + "step": 23405 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.014, + "step": 23406 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23407 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 2.2189, + "step": 23408 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.7067, + "step": 23409 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23410 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 3.9689, + "step": 23411 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.6086, + "step": 23412 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 3.6737, + "step": 23413 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.9252, + "step": 23414 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 3.1445, + "step": 23415 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23416 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 5.199, + "step": 23417 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23418 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.7623, + "step": 23419 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.6162, + "step": 23420 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23421 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 7.5783, + "step": 23422 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23423 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 6.1662, + "step": 23424 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 9.7187, + "step": 23425 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23426 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.0982, + "step": 23427 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 5.8634, + "step": 23428 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23429 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 5.0795, + "step": 23430 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23431 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 7.643, + "step": 23432 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23433 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 8.4541, + "step": 23434 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.9091, + "step": 23435 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 2.091, + "step": 23436 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23437 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23438 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23439 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23440 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.8933, + "step": 23441 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 2.4928, + "step": 23442 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23443 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23444 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.4103, + "step": 23445 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23446 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23447 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.1467, + "step": 23448 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23449 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23450 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23451 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 6.2649, + "step": 23452 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23453 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.5561, + "step": 23454 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23455 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23456 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23457 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.7631, + "step": 23458 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23459 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.2765, + "step": 23460 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 2.1806, + "step": 23461 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23462 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23463 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23464 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 2.2608, + "step": 23465 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 2.2279, + "step": 23466 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.0074, + "step": 23467 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23468 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.581, + "step": 23469 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 6.432, + "step": 23470 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.1416, + "step": 23471 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23472 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23473 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 7.7218, + "step": 23474 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.7813, + "step": 23475 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23476 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23477 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23478 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23479 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 3.2522, + "step": 23480 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 6.8297, + "step": 23481 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23482 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.9848, + "step": 23483 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23484 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 8.8525, + "step": 23485 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 3.3262, + "step": 23486 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23487 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23488 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 6.676, + "step": 23489 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23490 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23491 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.6302, + "step": 23492 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23493 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23494 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 3.0965, + "step": 23495 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23496 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 4.505, + "step": 23497 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 9.5395, + "step": 23498 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23499 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23500 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23501 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23502 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23503 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23504 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23505 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 1.8056, + "step": 23506 + }, + { + "epoch": 1.96, + "learning_rate": 0.00019987181987906485, + "loss": 7.6714, + "step": 23507 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23508 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23509 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.9744, + "step": 23510 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.8445, + "step": 23511 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.073, + "step": 23512 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 8.9765, + "step": 23513 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.7259, + "step": 23514 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 5.3775, + "step": 23515 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.463, + "step": 23516 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23517 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23518 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23519 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.7746, + "step": 23520 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23521 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 5.2797, + "step": 23522 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.0376, + "step": 23523 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 5.4581, + "step": 23524 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.2308, + "step": 23525 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23526 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23527 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 7.1896, + "step": 23528 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.614, + "step": 23529 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.1964, + "step": 23530 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.8371, + "step": 23531 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23532 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.8467, + "step": 23533 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.3399, + "step": 23534 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.0269, + "step": 23535 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23536 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23537 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 7.2585, + "step": 23538 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 6.7376, + "step": 23539 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.2865, + "step": 23540 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.877, + "step": 23541 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.0812, + "step": 23542 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23543 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23544 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.2228, + "step": 23545 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23546 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23547 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23548 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.4148, + "step": 23549 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.1315, + "step": 23550 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23551 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23552 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23553 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.7272, + "step": 23554 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.228, + "step": 23555 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23556 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.9062, + "step": 23557 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.6917, + "step": 23558 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.6435, + "step": 23559 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23560 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.1695, + "step": 23561 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23562 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23563 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.9719, + "step": 23564 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.2315, + "step": 23565 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23566 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23567 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23568 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23569 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23570 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23571 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23572 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.9819, + "step": 23573 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 8.5546, + "step": 23574 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.2919, + "step": 23575 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.7615, + "step": 23576 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23577 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23578 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.7981, + "step": 23579 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23580 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 7.2688, + "step": 23581 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23582 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23583 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.9988, + "step": 23584 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.4982, + "step": 23585 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23586 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23587 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 9.3085, + "step": 23588 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23589 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.7652, + "step": 23590 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.5724, + "step": 23591 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23592 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.0712, + "step": 23593 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.3681, + "step": 23594 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23595 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 9.0947, + "step": 23596 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 7.7079, + "step": 23597 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 8.8805, + "step": 23598 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.9785, + "step": 23599 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.9013, + "step": 23600 + }, + { + "epoch": 1.97, + "eval_loss": NaN, + "eval_runtime": 45.7377, + "eval_samples_per_second": 54.66, + "eval_steps_per_second": 6.843, + "step": 23600 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 2.2232, + "step": 23601 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23602 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 10.602, + "step": 23603 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 6.6647, + "step": 23604 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23605 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 8.2111, + "step": 23606 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23607 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.451, + "step": 23608 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 7.1814, + "step": 23609 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 3.942, + "step": 23610 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 7.1849, + "step": 23611 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23612 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23613 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23614 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23615 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 4.6993, + "step": 23616 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.98, + "step": 23617 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23618 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23619 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.9604, + "step": 23620 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23621 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.3295, + "step": 23622 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 6.7883, + "step": 23623 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23624 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 1.1674, + "step": 23625 + }, + { + "epoch": 1.97, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23626 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23627 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.0527, + "step": 23628 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23629 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23630 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23631 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23632 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23633 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23634 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 6.9242, + "step": 23635 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23636 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.859, + "step": 23637 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 6.1997, + "step": 23638 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.9749, + "step": 23639 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.1324, + "step": 23640 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 4.8544, + "step": 23641 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 7.5212, + "step": 23642 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 7.4484, + "step": 23643 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23644 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23645 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.9815, + "step": 23646 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23647 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.9415, + "step": 23648 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23649 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.9618, + "step": 23650 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23651 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23652 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.0026, + "step": 23653 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23654 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23655 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 4.8197, + "step": 23656 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.0635, + "step": 23657 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.7666, + "step": 23658 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.9487, + "step": 23659 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 10.0894, + "step": 23660 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23661 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23662 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23663 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.3955, + "step": 23664 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 6.5755, + "step": 23665 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23666 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23667 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.4715, + "step": 23668 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23669 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23670 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23671 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 9.1695, + "step": 23672 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.4113, + "step": 23673 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23674 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23675 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.6746, + "step": 23676 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 8.2454, + "step": 23677 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.9843, + "step": 23678 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.1746, + "step": 23679 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23680 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23681 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23682 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 6.0782, + "step": 23683 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23684 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.0849, + "step": 23685 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23686 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23687 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23688 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.1516, + "step": 23689 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 4.7101, + "step": 23690 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23691 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23692 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 6.9809, + "step": 23693 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23694 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23695 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.1086, + "step": 23696 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 9.1228, + "step": 23697 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.095, + "step": 23698 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23699 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23700 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 6.1824, + "step": 23701 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 7.8053, + "step": 23702 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23703 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.8429, + "step": 23704 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.9516, + "step": 23705 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.0864, + "step": 23706 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.0446, + "step": 23707 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23708 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.9203, + "step": 23709 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.909, + "step": 23710 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.8237, + "step": 23711 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.2163, + "step": 23712 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 4.4952, + "step": 23713 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23714 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23715 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.2512, + "step": 23716 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.7896, + "step": 23717 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 9.1259, + "step": 23718 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23719 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 5.8526, + "step": 23720 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 8.019, + "step": 23721 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23722 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23723 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.4326, + "step": 23724 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.6962, + "step": 23725 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 6.9075, + "step": 23726 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.8697, + "step": 23727 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.2622, + "step": 23728 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23729 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.8022, + "step": 23730 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 7.899, + "step": 23731 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 8.0731, + "step": 23732 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 3.7716, + "step": 23733 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23734 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23735 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 5.1156, + "step": 23736 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23737 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23738 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.1401, + "step": 23739 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23740 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.0341, + "step": 23741 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 2.0794, + "step": 23742 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 8.1456, + "step": 23743 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23744 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23745 + }, + { + "epoch": 1.98, + "learning_rate": 0.00019987181987906485, + "loss": 1.9275, + "step": 23746 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 4.1336, + "step": 23747 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23748 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 7.0912, + "step": 23749 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23750 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23751 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 5.8696, + "step": 23752 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.1679, + "step": 23753 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23754 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 2.0254, + "step": 23755 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.8674, + "step": 23756 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 2.8683, + "step": 23757 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.9529, + "step": 23758 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.9968, + "step": 23759 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 2.2221, + "step": 23760 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23761 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 4.1282, + "step": 23762 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.9848, + "step": 23763 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23764 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 9.3877, + "step": 23765 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 6.805, + "step": 23766 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23767 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 4.1777, + "step": 23768 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 7.5817, + "step": 23769 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.3696, + "step": 23770 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 5.2713, + "step": 23771 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23772 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23773 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23774 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23775 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23776 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23777 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.8659, + "step": 23778 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23779 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23780 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23781 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23782 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.885, + "step": 23783 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23784 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23785 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.8491, + "step": 23786 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.8667, + "step": 23787 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23788 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.2263, + "step": 23789 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23790 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23791 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23792 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.9235, + "step": 23793 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 7.5565, + "step": 23794 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.7969, + "step": 23795 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.7087, + "step": 23796 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.7379, + "step": 23797 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23798 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23799 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23800 + }, + { + "epoch": 1.99, + "eval_loss": NaN, + "eval_runtime": 45.4548, + "eval_samples_per_second": 55.0, + "eval_steps_per_second": 6.886, + "step": 23800 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 11.6615, + "step": 23801 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.2129, + "step": 23802 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23803 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 5.4969, + "step": 23804 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.7882, + "step": 23805 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.057, + "step": 23806 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 2.0046, + "step": 23807 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23808 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23809 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23810 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23811 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 8.1285, + "step": 23812 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.7462, + "step": 23813 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23814 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23815 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23816 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.3104, + "step": 23817 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 5.8747, + "step": 23818 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.9653, + "step": 23819 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.9106, + "step": 23820 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.223, + "step": 23821 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23822 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.7254, + "step": 23823 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23824 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 8.6519, + "step": 23825 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.7978, + "step": 23826 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 7.7412, + "step": 23827 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23828 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.1712, + "step": 23829 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.788, + "step": 23830 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.3526, + "step": 23831 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 4.068, + "step": 23832 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23833 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.1035, + "step": 23834 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.8981, + "step": 23835 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 2.194, + "step": 23836 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 7.8508, + "step": 23837 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 9.0321, + "step": 23838 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 8.9404, + "step": 23839 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23840 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23841 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 4.9097, + "step": 23842 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23843 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 3.1697, + "step": 23844 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23845 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23846 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23847 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23848 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.0625, + "step": 23849 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 7.9775, + "step": 23850 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23851 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23852 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23853 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23854 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.1116, + "step": 23855 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23856 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23857 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23858 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23859 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 4.7976, + "step": 23860 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23861 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 2.134, + "step": 23862 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 4.6065, + "step": 23863 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 7.3723, + "step": 23864 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 1.0182, + "step": 23865 + }, + { + "epoch": 1.99, + "learning_rate": 0.00019987181987906485, + "loss": 5.6, + "step": 23866 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23867 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23868 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23869 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 8.6857, + "step": 23870 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23871 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23872 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23873 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 6.4996, + "step": 23874 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.9046, + "step": 23875 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 2.9494, + "step": 23876 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23877 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23878 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23879 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23880 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 4.3892, + "step": 23881 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 2.9322, + "step": 23882 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 9.8371, + "step": 23883 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23884 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23885 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23886 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23887 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.0927, + "step": 23888 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 9.0766, + "step": 23889 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 2.0692, + "step": 23890 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.0011, + "step": 23891 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.9046, + "step": 23892 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 3.0291, + "step": 23893 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23894 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 4.013, + "step": 23895 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.9308, + "step": 23896 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.0129, + "step": 23897 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23898 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23899 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23900 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 7.3026, + "step": 23901 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 5.9985, + "step": 23902 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23903 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 8.4127, + "step": 23904 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 5.9515, + "step": 23905 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23906 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23907 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23908 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 4.1238, + "step": 23909 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.9242, + "step": 23910 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 3.9726, + "step": 23911 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 3.3427, + "step": 23912 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23913 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.607, + "step": 23914 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.008, + "step": 23915 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23916 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.97, + "step": 23917 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.0582, + "step": 23918 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23919 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23920 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 3.9237, + "step": 23921 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.4189, + "step": 23922 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23923 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 7.424, + "step": 23924 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23925 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 4.459, + "step": 23926 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 3.5616, + "step": 23927 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23928 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23929 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23930 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.177, + "step": 23931 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 4.2285, + "step": 23932 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 2.4965, + "step": 23933 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23934 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23935 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 8.5233, + "step": 23936 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.8281, + "step": 23937 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 7.6769, + "step": 23938 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23939 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.9023, + "step": 23940 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23941 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23942 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 7.474, + "step": 23943 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.881, + "step": 23944 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23945 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23946 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23947 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23948 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23949 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23950 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23951 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23952 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.0418, + "step": 23953 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 10.8251, + "step": 23954 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 6.2034, + "step": 23955 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23956 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23957 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.7874, + "step": 23958 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23959 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 2.0522, + "step": 23960 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23961 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23962 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 6.9707, + "step": 23963 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23964 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23965 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 8.0908, + "step": 23966 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23967 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23968 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 4.1893, + "step": 23969 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 2.8235, + "step": 23970 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.937, + "step": 23971 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23972 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.9379, + "step": 23973 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 3.5718, + "step": 23974 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 8.805, + "step": 23975 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 5.9787, + "step": 23976 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23977 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 9.2462, + "step": 23978 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.9732, + "step": 23979 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 5.8149, + "step": 23980 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 2.1484, + "step": 23981 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 3.8195, + "step": 23982 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 4.6127, + "step": 23983 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23984 + }, + { + "epoch": 2.0, + "learning_rate": 0.00019987181987906485, + "loss": 1.0626, + "step": 23985 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23986 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.9723, + "step": 23987 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23988 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23989 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23990 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23991 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 9.1885, + "step": 23992 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 10.3395, + "step": 23993 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 3.2351, + "step": 23994 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23995 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.3491, + "step": 23996 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 23997 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.7905, + "step": 23998 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 12.8089, + "step": 23999 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24000 + }, + { + "epoch": 2.01, + "eval_loss": NaN, + "eval_runtime": 41.7958, + "eval_samples_per_second": 59.815, + "eval_steps_per_second": 7.489, + "step": 24000 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 8.017, + "step": 24001 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 8.096, + "step": 24002 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24003 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24004 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24005 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24006 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 11.7357, + "step": 24007 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24008 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24009 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.9495, + "step": 24010 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 3.2359, + "step": 24011 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24012 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24013 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 3.0086, + "step": 24014 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.9855, + "step": 24015 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 2.9573, + "step": 24016 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.6533, + "step": 24017 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24018 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24019 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24020 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.3751, + "step": 24021 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 3.9181, + "step": 24022 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24023 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 5.1719, + "step": 24024 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.5276, + "step": 24025 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24026 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 3.1077, + "step": 24027 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24028 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 10.9689, + "step": 24029 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.7455, + "step": 24030 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24031 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24032 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 2.52, + "step": 24033 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.7507, + "step": 24034 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24035 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.9867, + "step": 24036 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 6.447, + "step": 24037 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24038 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.8022, + "step": 24039 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24040 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24041 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.9834, + "step": 24042 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.7947, + "step": 24043 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24044 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 2.0942, + "step": 24045 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 9.4613, + "step": 24046 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24047 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24048 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.0875, + "step": 24049 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 6.2285, + "step": 24050 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24051 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.4304, + "step": 24052 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24053 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 3.0067, + "step": 24054 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 6.4988, + "step": 24055 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.7314, + "step": 24056 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.9837, + "step": 24057 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24058 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 6.7332, + "step": 24059 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 3.3423, + "step": 24060 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.858, + "step": 24061 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 2.0756, + "step": 24062 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24063 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 6.1897, + "step": 24064 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.985, + "step": 24065 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.9445, + "step": 24066 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24067 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24068 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 2.0064, + "step": 24069 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.9461, + "step": 24070 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24071 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24072 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24073 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24074 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.9314, + "step": 24075 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 5.1423, + "step": 24076 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.047, + "step": 24077 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.8729, + "step": 24078 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.7326, + "step": 24079 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.7967, + "step": 24080 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24081 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 11.6768, + "step": 24082 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24083 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.6956, + "step": 24084 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24085 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24086 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24087 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 7.2348, + "step": 24088 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 2.1895, + "step": 24089 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24090 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24091 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24092 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24093 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.0044, + "step": 24094 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 1.1018, + "step": 24095 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.7859, + "step": 24096 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24097 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.6986, + "step": 24098 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24099 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 2.1731, + "step": 24100 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.99, + "step": 24101 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 10.4489, + "step": 24102 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24103 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24104 + }, + { + "epoch": 2.01, + "learning_rate": 0.00019987181987906485, + "loss": 4.9493, + "step": 24105 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 8.254, + "step": 24106 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.8839, + "step": 24107 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24108 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24109 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.4028, + "step": 24110 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24111 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.9432, + "step": 24112 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24113 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 9.7908, + "step": 24114 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 12.2087, + "step": 24115 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.8333, + "step": 24116 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.5276, + "step": 24117 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24118 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 8.2092, + "step": 24119 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24120 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.0162, + "step": 24121 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.3863, + "step": 24122 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24123 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24124 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.8607, + "step": 24125 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24126 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24127 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24128 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24129 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24130 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24131 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24132 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24133 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.831, + "step": 24134 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 8.2354, + "step": 24135 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 7.1793, + "step": 24136 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24137 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.9543, + "step": 24138 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.2489, + "step": 24139 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.8285, + "step": 24140 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 8.3048, + "step": 24141 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 5.5858, + "step": 24142 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.0256, + "step": 24143 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.8258, + "step": 24144 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24145 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24146 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24147 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24148 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.0138, + "step": 24149 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24150 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.0443, + "step": 24151 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24152 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24153 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24154 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24155 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.6453, + "step": 24156 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.1437, + "step": 24157 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.1378, + "step": 24158 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.7251, + "step": 24159 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24160 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24161 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24162 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.8495, + "step": 24163 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24164 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24165 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 7.4661, + "step": 24166 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24167 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 10.4334, + "step": 24168 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24169 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.0319, + "step": 24170 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.8687, + "step": 24171 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 7.6475, + "step": 24172 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24173 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24174 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 3.125, + "step": 24175 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24176 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.9836, + "step": 24177 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24178 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24179 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 6.9533, + "step": 24180 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 5.9309, + "step": 24181 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.8779, + "step": 24182 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.4825, + "step": 24183 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 8.5009, + "step": 24184 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.9948, + "step": 24185 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 8.4224, + "step": 24186 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24187 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.6795, + "step": 24188 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 6.8754, + "step": 24189 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.2944, + "step": 24190 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24191 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.8719, + "step": 24192 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 3.3801, + "step": 24193 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24194 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24195 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.9111, + "step": 24196 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24197 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 8.3862, + "step": 24198 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24199 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24200 + }, + { + "epoch": 2.02, + "eval_loss": NaN, + "eval_runtime": 43.0506, + "eval_samples_per_second": 58.071, + "eval_steps_per_second": 7.271, + "step": 24200 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.7859, + "step": 24201 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24202 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24203 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.4485, + "step": 24204 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.0778, + "step": 24205 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24206 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.8619, + "step": 24207 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 9.0601, + "step": 24208 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.0234, + "step": 24209 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.9495, + "step": 24210 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24211 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.0203, + "step": 24212 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24213 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24214 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24215 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.0431, + "step": 24216 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 1.8599, + "step": 24217 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24218 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 5.4272, + "step": 24219 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24220 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 7.7239, + "step": 24221 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 2.0282, + "step": 24222 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 3.4166, + "step": 24223 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 4.0835, + "step": 24224 + }, + { + "epoch": 2.02, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24225 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24226 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.9779, + "step": 24227 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24228 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.048, + "step": 24229 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24230 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24231 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 5.3601, + "step": 24232 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24233 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24234 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.0688, + "step": 24235 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.912, + "step": 24236 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.1145, + "step": 24237 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.8316, + "step": 24238 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.5944, + "step": 24239 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24240 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.2225, + "step": 24241 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.721, + "step": 24242 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.9551, + "step": 24243 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.8284, + "step": 24244 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.9168, + "step": 24245 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24246 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24247 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.954, + "step": 24248 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24249 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24250 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 9.0521, + "step": 24251 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.4793, + "step": 24252 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.4593, + "step": 24253 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24254 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.0034, + "step": 24255 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.0724, + "step": 24256 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24257 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.1179, + "step": 24258 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24259 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24260 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.0927, + "step": 24261 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.3758, + "step": 24262 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 7.8069, + "step": 24263 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24264 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.0323, + "step": 24265 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24266 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.0405, + "step": 24267 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.4362, + "step": 24268 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24269 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24270 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24271 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24272 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.985, + "step": 24273 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 6.5967, + "step": 24274 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24275 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24276 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 6.6227, + "step": 24277 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.053, + "step": 24278 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.7854, + "step": 24279 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.899, + "step": 24280 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24281 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.1731, + "step": 24282 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.8345, + "step": 24283 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24284 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.0964, + "step": 24285 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.0414, + "step": 24286 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24287 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24288 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 5.2199, + "step": 24289 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24290 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24291 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24292 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24293 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.8477, + "step": 24294 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24295 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24296 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 7.3755, + "step": 24297 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.5791, + "step": 24298 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 5.5964, + "step": 24299 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.2204, + "step": 24300 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24301 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24302 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 5.7756, + "step": 24303 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 7.2003, + "step": 24304 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24305 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24306 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.7884, + "step": 24307 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 7.65, + "step": 24308 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.5525, + "step": 24309 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24310 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24311 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 2.1755, + "step": 24312 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24313 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 6.8702, + "step": 24314 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.4483, + "step": 24315 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24316 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.4093, + "step": 24317 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 4.1493, + "step": 24318 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24319 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 3.6726, + "step": 24320 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.9568, + "step": 24321 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24322 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.9969, + "step": 24323 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24324 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24325 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24326 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 9.1011, + "step": 24327 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24328 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.9136, + "step": 24329 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 6.4772, + "step": 24330 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 7.3489, + "step": 24331 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.8839, + "step": 24332 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24333 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 6.1994, + "step": 24334 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 6.0431, + "step": 24335 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24336 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24337 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24338 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24339 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24340 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 7.5762, + "step": 24341 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24342 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24343 + }, + { + "epoch": 2.03, + "learning_rate": 0.00019987181987906485, + "loss": 1.4363, + "step": 24344 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.5452, + "step": 24345 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24346 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24347 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.9883, + "step": 24348 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.9536, + "step": 24349 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24350 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.2496, + "step": 24351 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24352 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.4753, + "step": 24353 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.4032, + "step": 24354 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24355 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.5435, + "step": 24356 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 2.0426, + "step": 24357 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24358 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 2.3213, + "step": 24359 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24360 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.0689, + "step": 24361 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24362 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24363 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.6833, + "step": 24364 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24365 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24366 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.2797, + "step": 24367 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24368 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24369 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24370 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 5.2646, + "step": 24371 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.0883, + "step": 24372 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24373 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24374 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24375 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.7074, + "step": 24376 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24377 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.2344, + "step": 24378 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24379 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.707, + "step": 24380 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.9241, + "step": 24381 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24382 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.8718, + "step": 24383 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.3685, + "step": 24384 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24385 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24386 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.9845, + "step": 24387 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24388 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.1337, + "step": 24389 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 2.2471, + "step": 24390 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 4.931, + "step": 24391 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24392 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24393 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24394 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24395 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.8142, + "step": 24396 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 4.1031, + "step": 24397 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.8318, + "step": 24398 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 6.5018, + "step": 24399 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 5.2363, + "step": 24400 + }, + { + "epoch": 2.04, + "eval_loss": NaN, + "eval_runtime": 43.4217, + "eval_samples_per_second": 57.575, + "eval_steps_per_second": 7.208, + "step": 24400 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 8.4594, + "step": 24401 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24402 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.8193, + "step": 24403 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24404 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 6.4536, + "step": 24405 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24406 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 9.4826, + "step": 24407 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24408 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24409 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.3585, + "step": 24410 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24411 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24412 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.9886, + "step": 24413 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24414 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.1369, + "step": 24415 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 5.5329, + "step": 24416 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24417 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24418 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24419 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24420 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.31, + "step": 24421 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24422 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 6.1602, + "step": 24423 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.2452, + "step": 24424 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24425 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24426 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24427 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 4.5507, + "step": 24428 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24429 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 4.4603, + "step": 24430 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.7437, + "step": 24431 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 1.0434, + "step": 24432 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24433 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24434 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 10.1033, + "step": 24435 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24436 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24437 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24438 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24439 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24440 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 2.984, + "step": 24441 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24442 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.0749, + "step": 24443 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.1186, + "step": 24444 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 6.4388, + "step": 24445 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.4943, + "step": 24446 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24447 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.4019, + "step": 24448 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24449 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 2.1132, + "step": 24450 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 5.0343, + "step": 24451 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24452 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 10.1529, + "step": 24453 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24454 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.2491, + "step": 24455 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 2.7298, + "step": 24456 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 2.4881, + "step": 24457 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 7.3192, + "step": 24458 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 3.9063, + "step": 24459 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 9.6607, + "step": 24460 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24461 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 4.0238, + "step": 24462 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 4.4152, + "step": 24463 + }, + { + "epoch": 2.04, + "learning_rate": 0.00019987181987906485, + "loss": 4.6431, + "step": 24464 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24465 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24466 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24467 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.8819, + "step": 24468 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.647, + "step": 24469 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.9702, + "step": 24470 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.9528, + "step": 24471 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.6201, + "step": 24472 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24473 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24474 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.1966, + "step": 24475 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24476 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.4438, + "step": 24477 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24478 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24479 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.0259, + "step": 24480 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24481 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.2507, + "step": 24482 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24483 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24484 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.936, + "step": 24485 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24486 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24487 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24488 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24489 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.6904, + "step": 24490 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.8679, + "step": 24491 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24492 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24493 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.6114, + "step": 24494 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.6628, + "step": 24495 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24496 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.0202, + "step": 24497 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.3967, + "step": 24498 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.3394, + "step": 24499 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 7.8799, + "step": 24500 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24501 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24502 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.9167, + "step": 24503 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24504 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.057, + "step": 24505 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24506 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.8725, + "step": 24507 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.2222, + "step": 24508 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 7.5848, + "step": 24509 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24510 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.8016, + "step": 24511 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.7877, + "step": 24512 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 5.1082, + "step": 24513 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.5219, + "step": 24514 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24515 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.8216, + "step": 24516 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24517 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24518 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.4287, + "step": 24519 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 9.3006, + "step": 24520 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24521 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24522 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24523 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24524 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.9021, + "step": 24525 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24526 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24527 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.7137, + "step": 24528 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24529 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.3258, + "step": 24530 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.9188, + "step": 24531 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 6.6537, + "step": 24532 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24533 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.0025, + "step": 24534 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24535 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.0504, + "step": 24536 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24537 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.5248, + "step": 24538 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24539 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.312, + "step": 24540 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.688, + "step": 24541 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24542 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24543 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 8.9589, + "step": 24544 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.0188, + "step": 24545 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.8016, + "step": 24546 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24547 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 5.6563, + "step": 24548 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.8647, + "step": 24549 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.1528, + "step": 24550 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.912, + "step": 24551 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.872, + "step": 24552 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.7723, + "step": 24553 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24554 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.9799, + "step": 24555 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24556 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24557 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24558 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.8733, + "step": 24559 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24560 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.8536, + "step": 24561 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24562 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.7971, + "step": 24563 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24564 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.8557, + "step": 24565 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 4.1239, + "step": 24566 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.8484, + "step": 24567 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.2004, + "step": 24568 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 4.1463, + "step": 24569 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24570 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 4.3863, + "step": 24571 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24572 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 5.5703, + "step": 24573 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24574 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.9448, + "step": 24575 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24576 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24577 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24578 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 4.1812, + "step": 24579 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 2.2217, + "step": 24580 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 3.1546, + "step": 24581 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 1.6696, + "step": 24582 + }, + { + "epoch": 2.05, + "learning_rate": 0.00019987181987906485, + "loss": 4.3558, + "step": 24583 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 10.9201, + "step": 24584 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24585 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.4666, + "step": 24586 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.2205, + "step": 24587 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24588 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24589 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 7.6913, + "step": 24590 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24591 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 5.4128, + "step": 24592 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24593 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 6.2703, + "step": 24594 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.7916, + "step": 24595 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 2.0064, + "step": 24596 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.7484, + "step": 24597 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24598 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 8.5631, + "step": 24599 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.1541, + "step": 24600 + }, + { + "epoch": 2.06, + "eval_loss": NaN, + "eval_runtime": 44.6101, + "eval_samples_per_second": 56.041, + "eval_steps_per_second": 7.016, + "step": 24600 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.2796, + "step": 24601 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.0146, + "step": 24602 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24603 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 5.0515, + "step": 24604 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24605 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24606 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 6.9846, + "step": 24607 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.8208, + "step": 24608 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.6751, + "step": 24609 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24610 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24611 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.1434, + "step": 24612 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 9.8343, + "step": 24613 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24614 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.8927, + "step": 24615 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24616 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24617 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.948, + "step": 24618 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24619 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 7.3641, + "step": 24620 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24621 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24622 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.8423, + "step": 24623 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 6.9732, + "step": 24624 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24625 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24626 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24627 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24628 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24629 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24630 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24631 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 6.1482, + "step": 24632 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 7.6298, + "step": 24633 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24634 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.351, + "step": 24635 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24636 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 8.4057, + "step": 24637 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24638 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 2.6484, + "step": 24639 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.0335, + "step": 24640 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24641 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.9792, + "step": 24642 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.2824, + "step": 24643 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24644 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 2.2207, + "step": 24645 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24646 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.3571, + "step": 24647 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.656, + "step": 24648 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.7378, + "step": 24649 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 5.1116, + "step": 24650 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24651 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.8509, + "step": 24652 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24653 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24654 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.3631, + "step": 24655 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.8459, + "step": 24656 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 9.4669, + "step": 24657 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24658 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24659 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 8.1235, + "step": 24660 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24661 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24662 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.4127, + "step": 24663 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24664 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.2752, + "step": 24665 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24666 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24667 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24668 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24669 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24670 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24671 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24672 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24673 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24674 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.8613, + "step": 24675 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 7.406, + "step": 24676 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24677 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24678 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24679 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 2.9889, + "step": 24680 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 2.0032, + "step": 24681 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24682 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.283, + "step": 24683 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.7485, + "step": 24684 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.4015, + "step": 24685 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.936, + "step": 24686 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 3.1453, + "step": 24687 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 4.5894, + "step": 24688 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 1.7541, + "step": 24689 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24690 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 6.8799, + "step": 24691 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 5.2938, + "step": 24692 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24693 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24694 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 5.2244, + "step": 24695 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24696 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24697 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24698 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24699 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24700 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 9.1445, + "step": 24701 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 2.6512, + "step": 24702 + }, + { + "epoch": 2.06, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24703 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.7035, + "step": 24704 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 9.4085, + "step": 24705 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.2911, + "step": 24706 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24707 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24708 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.0442, + "step": 24709 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24710 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24711 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24712 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.7264, + "step": 24713 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24714 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.8242, + "step": 24715 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 2.4618, + "step": 24716 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24717 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 6.7784, + "step": 24718 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 2.9051, + "step": 24719 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 2.8543, + "step": 24720 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.644, + "step": 24721 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 7.4307, + "step": 24722 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.8811, + "step": 24723 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 6.6954, + "step": 24724 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.9822, + "step": 24725 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24726 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24727 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24728 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 2.2679, + "step": 24729 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 3.7433, + "step": 24730 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24731 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.6224, + "step": 24732 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.0864, + "step": 24733 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24734 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.6412, + "step": 24735 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24736 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 2.1109, + "step": 24737 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24738 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24739 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24740 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.7927, + "step": 24741 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.0101, + "step": 24742 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.9952, + "step": 24743 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 7.1468, + "step": 24744 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 7.0581, + "step": 24745 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 9.7904, + "step": 24746 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.9963, + "step": 24747 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 2.0694, + "step": 24748 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 3.9248, + "step": 24749 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 8.2483, + "step": 24750 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.468, + "step": 24751 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24752 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24753 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.7047, + "step": 24754 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24755 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 6.493, + "step": 24756 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.0363, + "step": 24757 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24758 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.9657, + "step": 24759 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.2726, + "step": 24760 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24761 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.8605, + "step": 24762 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.395, + "step": 24763 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24764 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24765 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24766 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24767 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 3.5486, + "step": 24768 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.0653, + "step": 24769 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 8.8976, + "step": 24770 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.9809, + "step": 24771 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.0079, + "step": 24772 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.9331, + "step": 24773 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.0539, + "step": 24774 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 3.4408, + "step": 24775 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24776 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 2.2507, + "step": 24777 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.9888, + "step": 24778 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24779 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 3.4828, + "step": 24780 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24781 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 7.9523, + "step": 24782 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24783 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24784 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.9428, + "step": 24785 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.8796, + "step": 24786 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.5188, + "step": 24787 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24788 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 8.1947, + "step": 24789 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.228, + "step": 24790 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24791 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24792 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.5515, + "step": 24793 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24794 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24795 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24796 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.1855, + "step": 24797 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24798 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24799 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24800 + }, + { + "epoch": 2.07, + "eval_loss": NaN, + "eval_runtime": 45.3364, + "eval_samples_per_second": 55.143, + "eval_steps_per_second": 6.904, + "step": 24800 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.0798, + "step": 24801 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 3.6833, + "step": 24802 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24803 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 7.3138, + "step": 24804 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24805 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 7.3533, + "step": 24806 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24807 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 7.3928, + "step": 24808 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24809 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 6.1549, + "step": 24810 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.9479, + "step": 24811 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.8086, + "step": 24812 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24813 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.3847, + "step": 24814 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.9468, + "step": 24815 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.8683, + "step": 24816 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 5.8995, + "step": 24817 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 3.9117, + "step": 24818 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24819 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 4.8434, + "step": 24820 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24821 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.7743, + "step": 24822 + }, + { + "epoch": 2.07, + "learning_rate": 0.00019987181987906485, + "loss": 1.0093, + "step": 24823 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.0612, + "step": 24824 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.979, + "step": 24825 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.7081, + "step": 24826 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24827 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.1589, + "step": 24828 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24829 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.9322, + "step": 24830 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24831 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 5.7945, + "step": 24832 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.9573, + "step": 24833 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.5727, + "step": 24834 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24835 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.7054, + "step": 24836 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24837 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.1045, + "step": 24838 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 6.9107, + "step": 24839 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24840 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24841 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24842 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24843 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24844 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24845 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24846 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24847 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24848 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 7.378, + "step": 24849 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.0993, + "step": 24850 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.2862, + "step": 24851 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24852 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.8153, + "step": 24853 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.1964, + "step": 24854 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24855 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.2603, + "step": 24856 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.9592, + "step": 24857 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24858 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.1295, + "step": 24859 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24860 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24861 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.9192, + "step": 24862 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24863 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24864 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24865 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.1332, + "step": 24866 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24867 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.3125, + "step": 24868 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24869 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24870 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24871 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24872 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 3.7421, + "step": 24873 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24874 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24875 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24876 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.1087, + "step": 24877 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24878 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24879 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 6.586, + "step": 24880 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.8158, + "step": 24881 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24882 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24883 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.2892, + "step": 24884 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.8267, + "step": 24885 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24886 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 3.6701, + "step": 24887 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.9193, + "step": 24888 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.1771, + "step": 24889 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24890 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 8.2485, + "step": 24891 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24892 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24893 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.1037, + "step": 24894 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 6.6061, + "step": 24895 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.0159, + "step": 24896 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24897 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.6523, + "step": 24898 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24899 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.8216, + "step": 24900 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24901 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.7917, + "step": 24902 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24903 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24904 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.9726, + "step": 24905 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.8958, + "step": 24906 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.6498, + "step": 24907 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24908 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24909 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 1.1041, + "step": 24910 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24911 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24912 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24913 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24914 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24915 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24916 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24917 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24918 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24919 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24920 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.0824, + "step": 24921 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 6.5025, + "step": 24922 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24923 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.7458, + "step": 24924 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 7.8659, + "step": 24925 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24926 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.0666, + "step": 24927 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24928 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.095, + "step": 24929 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 2.5467, + "step": 24930 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 3.9429, + "step": 24931 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24932 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24933 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24934 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.9628, + "step": 24935 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.8264, + "step": 24936 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24937 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24938 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24939 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24940 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 4.477, + "step": 24941 + }, + { + "epoch": 2.08, + "learning_rate": 0.00019987181987906485, + "loss": 8.363, + "step": 24942 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.8789, + "step": 24943 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24944 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.6814, + "step": 24945 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.9724, + "step": 24946 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.6441, + "step": 24947 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24948 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24949 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24950 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 5.0594, + "step": 24951 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24952 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 5.861, + "step": 24953 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24954 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.8679, + "step": 24955 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.5528, + "step": 24956 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 5.7859, + "step": 24957 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24958 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 7.7234, + "step": 24959 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24960 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 9.3989, + "step": 24961 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24962 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.2718, + "step": 24963 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.011, + "step": 24964 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 6.6935, + "step": 24965 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 5.734, + "step": 24966 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.4305, + "step": 24967 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 7.2945, + "step": 24968 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24969 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.8282, + "step": 24970 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.8801, + "step": 24971 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24972 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24973 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24974 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.5944, + "step": 24975 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.8711, + "step": 24976 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.2598, + "step": 24977 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.4025, + "step": 24978 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24979 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24980 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24981 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 7.4302, + "step": 24982 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.2407, + "step": 24983 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24984 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 6.0369, + "step": 24985 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24986 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.4747, + "step": 24987 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24988 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.906, + "step": 24989 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24990 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.7822, + "step": 24991 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.1014, + "step": 24992 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 5.6804, + "step": 24993 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 6.9052, + "step": 24994 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24995 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24996 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.8747, + "step": 24997 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 24998 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.2427, + "step": 24999 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25000 + }, + { + "epoch": 2.09, + "eval_loss": NaN, + "eval_runtime": 42.4204, + "eval_samples_per_second": 58.934, + "eval_steps_per_second": 7.379, + "step": 25000 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25001 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25002 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25003 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25004 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 6.5929, + "step": 25005 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.681, + "step": 25006 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25007 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.8259, + "step": 25008 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.6118, + "step": 25009 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25010 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25011 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25012 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.9312, + "step": 25013 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25014 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25015 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25016 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25017 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 9.9186, + "step": 25018 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25019 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25020 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.9924, + "step": 25021 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.2416, + "step": 25022 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 9.5776, + "step": 25023 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25024 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25025 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25026 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25027 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 6.3618, + "step": 25028 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25029 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25030 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25031 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.7035, + "step": 25032 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.952, + "step": 25033 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.2803, + "step": 25034 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 5.2489, + "step": 25035 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 6.8592, + "step": 25036 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.6905, + "step": 25037 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25038 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25039 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.9587, + "step": 25040 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.8334, + "step": 25041 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.6117, + "step": 25042 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.2477, + "step": 25043 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25044 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25045 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.24, + "step": 25046 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.0764, + "step": 25047 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25048 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25049 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.5748, + "step": 25050 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 4.0276, + "step": 25051 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25052 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 5.2277, + "step": 25053 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.9467, + "step": 25054 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25055 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 6.5735, + "step": 25056 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25057 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25058 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 1.0112, + "step": 25059 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 8.0118, + "step": 25060 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 2.9157, + "step": 25061 + }, + { + "epoch": 2.09, + "learning_rate": 0.00019987181987906485, + "loss": 3.6696, + "step": 25062 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25063 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 7.08, + "step": 25064 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 5.9558, + "step": 25065 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25066 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 6.9699, + "step": 25067 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.7471, + "step": 25068 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25069 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.4863, + "step": 25070 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.0774, + "step": 25071 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25072 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 7.2962, + "step": 25073 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25074 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 5.9199, + "step": 25075 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.0574, + "step": 25076 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.0943, + "step": 25077 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25078 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.7538, + "step": 25079 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25080 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25081 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25082 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25083 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.8538, + "step": 25084 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 8.5671, + "step": 25085 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 4.1081, + "step": 25086 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25087 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25088 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 9.7945, + "step": 25089 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.9385, + "step": 25090 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.7398, + "step": 25091 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25092 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.2536, + "step": 25093 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 4.0, + "step": 25094 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 4.9365, + "step": 25095 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 4.9194, + "step": 25096 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25097 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.1708, + "step": 25098 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25099 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25100 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25101 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 10.1985, + "step": 25102 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.2791, + "step": 25103 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25104 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 7.5859, + "step": 25105 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25106 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25107 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.9247, + "step": 25108 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.1756, + "step": 25109 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25110 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25111 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25112 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 5.7306, + "step": 25113 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25114 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 4.2755, + "step": 25115 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25116 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.957, + "step": 25117 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25118 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 6.4279, + "step": 25119 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25120 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.7776, + "step": 25121 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 6.6652, + "step": 25122 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25123 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25124 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 5.157, + "step": 25125 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25126 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.1825, + "step": 25127 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 8.061, + "step": 25128 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 5.8001, + "step": 25129 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25130 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.4046, + "step": 25131 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.0163, + "step": 25132 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25133 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25134 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25135 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25136 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.0815, + "step": 25137 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.3975, + "step": 25138 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 5.0551, + "step": 25139 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25140 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25141 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25142 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25143 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25144 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25145 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 8.5397, + "step": 25146 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25147 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.0008, + "step": 25148 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25149 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25150 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25151 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25152 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 4.0818, + "step": 25153 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25154 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25155 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 6.6791, + "step": 25156 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25157 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25158 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25159 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 9.6153, + "step": 25160 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25161 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.3324, + "step": 25162 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 2.133, + "step": 25163 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25164 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 8.7968, + "step": 25165 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.0111, + "step": 25166 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25167 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25168 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25169 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25170 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25171 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 1.6562, + "step": 25172 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 8.7347, + "step": 25173 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25174 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25175 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25176 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 7.5343, + "step": 25177 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25178 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25179 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25180 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 0.9295, + "step": 25181 + }, + { + "epoch": 2.1, + "learning_rate": 0.00019987181987906485, + "loss": 3.064, + "step": 25182 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 5.7085, + "step": 25183 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 2.1757, + "step": 25184 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25185 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 8.0502, + "step": 25186 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.0327, + "step": 25187 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25188 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.5436, + "step": 25189 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25190 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25191 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.6961, + "step": 25192 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25193 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25194 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25195 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 6.0494, + "step": 25196 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.8111, + "step": 25197 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 5.951, + "step": 25198 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25199 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.1342, + "step": 25200 + }, + { + "epoch": 2.11, + "eval_loss": NaN, + "eval_runtime": 42.7978, + "eval_samples_per_second": 58.414, + "eval_steps_per_second": 7.313, + "step": 25200 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.2519, + "step": 25201 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25202 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25203 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.0851, + "step": 25204 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.8987, + "step": 25205 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 6.8789, + "step": 25206 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.5882, + "step": 25207 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.7577, + "step": 25208 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25209 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.8665, + "step": 25210 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.9815, + "step": 25211 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25212 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 2.0379, + "step": 25213 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25214 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25215 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25216 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25217 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25218 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 5.8791, + "step": 25219 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.8051, + "step": 25220 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25221 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.3607, + "step": 25222 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.9571, + "step": 25223 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25224 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.4636, + "step": 25225 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25226 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25227 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25228 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.922, + "step": 25229 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 6.701, + "step": 25230 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 4.3441, + "step": 25231 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25232 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25233 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.9199, + "step": 25234 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25235 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.4429, + "step": 25236 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.9526, + "step": 25237 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 4.0497, + "step": 25238 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25239 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25240 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 4.1571, + "step": 25241 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 9.5348, + "step": 25242 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25243 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.9325, + "step": 25244 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 4.323, + "step": 25245 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25246 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 8.6981, + "step": 25247 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25248 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25249 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25250 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25251 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 4.0473, + "step": 25252 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25253 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25254 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25255 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25256 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25257 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.7852, + "step": 25258 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25259 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25260 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 6.8638, + "step": 25261 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 1.1441, + "step": 25262 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.7174, + "step": 25263 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 2.0537, + "step": 25264 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25265 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25266 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.931, + "step": 25267 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25268 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 3.7715, + "step": 25269 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 2.0778, + "step": 25270 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019987181987906485, + "loss": 0.0, + "step": 25271 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 4.1021, + "step": 25272 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25273 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 7.8949, + "step": 25274 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25275 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25276 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 3.2569, + "step": 25277 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 2.3366, + "step": 25278 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25279 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 2.051, + "step": 25280 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25281 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 6.0815, + "step": 25282 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 3.8486, + "step": 25283 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25284 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 7.6828, + "step": 25285 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25286 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25287 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25288 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25289 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 3.1938, + "step": 25290 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25291 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 4.2897, + "step": 25292 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25293 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25294 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 2.2527, + "step": 25295 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.9575, + "step": 25296 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 5.1511, + "step": 25297 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25298 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 4.8152, + "step": 25299 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 1.9547, + "step": 25300 + }, + { + "epoch": 2.11, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25301 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.9632, + "step": 25302 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 7.556, + "step": 25303 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25304 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25305 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25306 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.0372, + "step": 25307 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25308 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.1437, + "step": 25309 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25310 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.4788, + "step": 25311 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25312 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25313 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25314 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25315 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 2.015, + "step": 25316 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 5.0436, + "step": 25317 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.7606, + "step": 25318 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 2.1653, + "step": 25319 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.826, + "step": 25320 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.0813, + "step": 25321 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.8379, + "step": 25322 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.8474, + "step": 25323 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 2.7284, + "step": 25324 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.8898, + "step": 25325 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.8025, + "step": 25326 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25327 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25328 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25329 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25330 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.1551, + "step": 25331 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.967, + "step": 25332 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 7.0327, + "step": 25333 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.6457, + "step": 25334 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.9035, + "step": 25335 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 9.0933, + "step": 25336 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.9266, + "step": 25337 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25338 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25339 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.3902, + "step": 25340 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.6573, + "step": 25341 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.8623, + "step": 25342 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25343 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25344 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.8129, + "step": 25345 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25346 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.8993, + "step": 25347 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 9.1563, + "step": 25348 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25349 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25350 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25351 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25352 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25353 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 2.2517, + "step": 25354 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.3057, + "step": 25355 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25356 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25357 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.2243, + "step": 25358 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.2729, + "step": 25359 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.8907, + "step": 25360 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.9324, + "step": 25361 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25362 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25363 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25364 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.879, + "step": 25365 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.9046, + "step": 25366 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25367 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.2091, + "step": 25368 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25369 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.5699, + "step": 25370 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25371 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 7.9166, + "step": 25372 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 17.445, + "step": 25373 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25374 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25375 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.858, + "step": 25376 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 7.1207, + "step": 25377 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.8485, + "step": 25378 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 7.1413, + "step": 25379 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.6369, + "step": 25380 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.1884, + "step": 25381 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.8911, + "step": 25382 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25383 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25384 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25385 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.9755, + "step": 25386 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 2.7605, + "step": 25387 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25388 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25389 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 2.1449, + "step": 25390 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 7.8518, + "step": 25391 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 9.5879, + "step": 25392 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25393 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 6.4874, + "step": 25394 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.0399, + "step": 25395 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25396 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25397 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25398 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25399 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 5.1403, + "step": 25400 + }, + { + "epoch": 2.12, + "eval_loss": NaN, + "eval_runtime": 44.1409, + "eval_samples_per_second": 56.637, + "eval_steps_per_second": 7.091, + "step": 25400 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.9629, + "step": 25401 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25402 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 2.0345, + "step": 25403 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 6.8321, + "step": 25404 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25405 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 8.0054, + "step": 25406 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25407 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 3.0852, + "step": 25408 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25409 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.1387, + "step": 25410 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.9644, + "step": 25411 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25412 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25413 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25414 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25415 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25416 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25417 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25418 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 4.156, + "step": 25419 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 1.0809, + "step": 25420 + }, + { + "epoch": 2.12, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25421 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 7.0544, + "step": 25422 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25423 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25424 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.477, + "step": 25425 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25426 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 7.4345, + "step": 25427 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25428 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25429 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 8.0714, + "step": 25430 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.7734, + "step": 25431 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25432 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.0794, + "step": 25433 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25434 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25435 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25436 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25437 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.8587, + "step": 25438 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25439 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25440 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 9.007, + "step": 25441 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25442 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25443 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.8467, + "step": 25444 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25445 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.2545, + "step": 25446 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.8897, + "step": 25447 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 7.5129, + "step": 25448 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 5.4401, + "step": 25449 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25450 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25451 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25452 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25453 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.683, + "step": 25454 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25455 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25456 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 2.8668, + "step": 25457 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.9564, + "step": 25458 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25459 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 2.2956, + "step": 25460 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 8.0278, + "step": 25461 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.7854, + "step": 25462 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25463 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.9692, + "step": 25464 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25465 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.9388, + "step": 25466 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25467 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25468 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.9485, + "step": 25469 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.2053, + "step": 25470 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25471 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.4591, + "step": 25472 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.0376, + "step": 25473 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.5937, + "step": 25474 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25475 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25476 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25477 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.2659, + "step": 25478 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25479 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25480 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 5.6446, + "step": 25481 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25482 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 2.2861, + "step": 25483 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.3771, + "step": 25484 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 2.5647, + "step": 25485 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.7797, + "step": 25486 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.8612, + "step": 25487 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25488 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25489 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.2956, + "step": 25490 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.5289, + "step": 25491 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.2982, + "step": 25492 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.0265, + "step": 25493 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 8.811, + "step": 25494 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25495 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.7859, + "step": 25496 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 8.0227, + "step": 25497 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25498 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.543, + "step": 25499 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.7693, + "step": 25500 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25501 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 6.2223, + "step": 25502 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25503 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 8.6017, + "step": 25504 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 7.9473, + "step": 25505 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.5641, + "step": 25506 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25507 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25508 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.991, + "step": 25509 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.7486, + "step": 25510 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.0813, + "step": 25511 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25512 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.0114, + "step": 25513 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25514 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.0936, + "step": 25515 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25516 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.0398, + "step": 25517 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25518 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25519 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 1.1261, + "step": 25520 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25521 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 5.7192, + "step": 25522 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.7595, + "step": 25523 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.9902, + "step": 25524 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 2.2286, + "step": 25525 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 5.833, + "step": 25526 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25527 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.0293, + "step": 25528 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.3433, + "step": 25529 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25530 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 2.5269, + "step": 25531 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.5007, + "step": 25532 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25533 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25534 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 4.1435, + "step": 25535 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.8235, + "step": 25536 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.8402, + "step": 25537 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 3.705, + "step": 25538 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25539 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25540 + }, + { + "epoch": 2.13, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25541 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25542 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 4.0567, + "step": 25543 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 4.0418, + "step": 25544 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.0715, + "step": 25545 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.0004, + "step": 25546 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.4577, + "step": 25547 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.905, + "step": 25548 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.2344, + "step": 25549 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25550 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 5.1024, + "step": 25551 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.8555, + "step": 25552 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.1443, + "step": 25553 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25554 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25555 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 9.5248, + "step": 25556 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25557 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.9707, + "step": 25558 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 6.9918, + "step": 25559 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25560 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 10.7858, + "step": 25561 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.5726, + "step": 25562 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25563 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25564 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25565 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25566 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 4.6379, + "step": 25567 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25568 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25569 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.7669, + "step": 25570 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.7281, + "step": 25571 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25572 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.1909, + "step": 25573 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.9801, + "step": 25574 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.243, + "step": 25575 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25576 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.8363, + "step": 25577 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25578 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 5.2539, + "step": 25579 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25580 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25581 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25582 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25583 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.6632, + "step": 25584 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25585 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25586 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 4.4551, + "step": 25587 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25588 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25589 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.7111, + "step": 25590 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25591 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25592 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25593 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25594 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25595 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 8.4998, + "step": 25596 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25597 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25598 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25599 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 4.9769, + "step": 25600 + }, + { + "epoch": 2.14, + "eval_loss": NaN, + "eval_runtime": 43.4282, + "eval_samples_per_second": 57.566, + "eval_steps_per_second": 7.207, + "step": 25600 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.847, + "step": 25601 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25602 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 6.152, + "step": 25603 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 7.0311, + "step": 25604 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.3421, + "step": 25605 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25606 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25607 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.812, + "step": 25608 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 7.704, + "step": 25609 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 8.4328, + "step": 25610 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.86, + "step": 25611 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25612 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.1049, + "step": 25613 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25614 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25615 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25616 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.0085, + "step": 25617 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25618 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.0432, + "step": 25619 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25620 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 9.4281, + "step": 25621 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.7324, + "step": 25622 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25623 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 7.0057, + "step": 25624 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25625 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 4.8191, + "step": 25626 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.0246, + "step": 25627 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 9.0605, + "step": 25628 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.6545, + "step": 25629 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.7631, + "step": 25630 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25631 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25632 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 1.5912, + "step": 25633 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25634 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.8726, + "step": 25635 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25636 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.1981, + "step": 25637 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25638 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.6929, + "step": 25639 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.0989, + "step": 25640 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 8.9354, + "step": 25641 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25642 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.3204, + "step": 25643 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 5.0463, + "step": 25644 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25645 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 4.4113, + "step": 25646 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25647 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 6.4547, + "step": 25648 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.6571, + "step": 25649 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.9729, + "step": 25650 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 5.4439, + "step": 25651 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25652 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.8599, + "step": 25653 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25654 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 2.178, + "step": 25655 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25656 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25657 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25658 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25659 + }, + { + "epoch": 2.14, + "learning_rate": 0.00019986624683032854, + "loss": 3.9625, + "step": 25660 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.581, + "step": 25661 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 5.0528, + "step": 25662 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.8982, + "step": 25663 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.5957, + "step": 25664 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 8.7229, + "step": 25665 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.3418, + "step": 25666 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 9.0906, + "step": 25667 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25668 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.9584, + "step": 25669 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.3286, + "step": 25670 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 8.7705, + "step": 25671 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.7197, + "step": 25672 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.4389, + "step": 25673 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25674 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.0654, + "step": 25675 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25676 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 11.1539, + "step": 25677 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.9799, + "step": 25678 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25679 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.1802, + "step": 25680 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.4607, + "step": 25681 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25682 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25683 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.072, + "step": 25684 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.0871, + "step": 25685 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25686 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.2559, + "step": 25687 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.7811, + "step": 25688 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25689 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.9241, + "step": 25690 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.0833, + "step": 25691 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.1371, + "step": 25692 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.081, + "step": 25693 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.4129, + "step": 25694 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 8.0252, + "step": 25695 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.3956, + "step": 25696 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25697 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 5.0623, + "step": 25698 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.7894, + "step": 25699 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.8083, + "step": 25700 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25701 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25702 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25703 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25704 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.2302, + "step": 25705 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25706 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25707 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 5.5208, + "step": 25708 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.8293, + "step": 25709 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25710 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25711 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.581, + "step": 25712 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25713 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25714 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25715 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25716 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.9862, + "step": 25717 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.6691, + "step": 25718 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.1977, + "step": 25719 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.0924, + "step": 25720 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 5.5412, + "step": 25721 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 8.5825, + "step": 25722 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25723 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.0445, + "step": 25724 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.1072, + "step": 25725 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.0373, + "step": 25726 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 5.1463, + "step": 25727 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25728 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.1267, + "step": 25729 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.7335, + "step": 25730 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.735, + "step": 25731 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25732 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25733 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25734 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25735 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.1574, + "step": 25736 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.1605, + "step": 25737 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25738 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.043, + "step": 25739 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25740 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 3.7492, + "step": 25741 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25742 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25743 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25744 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25745 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.7352, + "step": 25746 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25747 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25748 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25749 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25750 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.6629, + "step": 25751 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25752 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25753 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25754 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 7.0814, + "step": 25755 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.9951, + "step": 25756 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.1345, + "step": 25757 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 5.2079, + "step": 25758 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.6444, + "step": 25759 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25760 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.1, + "step": 25761 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 4.9888, + "step": 25762 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.7794, + "step": 25763 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25764 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25765 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.9647, + "step": 25766 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.0271, + "step": 25767 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25768 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.0091, + "step": 25769 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25770 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25771 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 8.0768, + "step": 25772 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 8.7588, + "step": 25773 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25774 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25775 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25776 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25777 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 1.6824, + "step": 25778 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25779 + }, + { + "epoch": 2.15, + "learning_rate": 0.00019986624683032854, + "loss": 2.0786, + "step": 25780 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.8369, + "step": 25781 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25782 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25783 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.916, + "step": 25784 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25785 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25786 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.1685, + "step": 25787 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.9114, + "step": 25788 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25789 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25790 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.8557, + "step": 25791 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25792 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.1835, + "step": 25793 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25794 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25795 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 9.7191, + "step": 25796 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 6.9952, + "step": 25797 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25798 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.5187, + "step": 25799 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 3.7523, + "step": 25800 + }, + { + "epoch": 2.16, + "eval_loss": NaN, + "eval_runtime": 42.9621, + "eval_samples_per_second": 58.191, + "eval_steps_per_second": 7.285, + "step": 25800 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 5.522, + "step": 25801 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25802 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 7.1545, + "step": 25803 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.0706, + "step": 25804 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.3014, + "step": 25805 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25806 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25807 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25808 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25809 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 5.6711, + "step": 25810 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25811 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25812 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25813 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.0617, + "step": 25814 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.3957, + "step": 25815 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.9256, + "step": 25816 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25817 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25818 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25819 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.299, + "step": 25820 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.144, + "step": 25821 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25822 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.0756, + "step": 25823 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25824 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25825 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 5.9547, + "step": 25826 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25827 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25828 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 2.0696, + "step": 25829 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25830 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25831 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.8605, + "step": 25832 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25833 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 3.0507, + "step": 25834 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25835 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 3.1524, + "step": 25836 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25837 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25838 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 5.9028, + "step": 25839 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 3.8535, + "step": 25840 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 5.9244, + "step": 25841 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 2.3025, + "step": 25842 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.9827, + "step": 25843 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.8422, + "step": 25844 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.1157, + "step": 25845 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25846 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 5.3736, + "step": 25847 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 8.4009, + "step": 25848 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.8145, + "step": 25849 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 2.5427, + "step": 25850 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 3.8843, + "step": 25851 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 5.5462, + "step": 25852 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25853 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25854 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25855 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25856 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.9057, + "step": 25857 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25858 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25859 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25860 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.9611, + "step": 25861 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25862 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25863 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 7.8074, + "step": 25864 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.5757, + "step": 25865 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 2.2347, + "step": 25866 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25867 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25868 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 6.9758, + "step": 25869 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25870 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25871 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 3.5082, + "step": 25872 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25873 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.925, + "step": 25874 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 2.9334, + "step": 25875 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25876 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25877 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 4.0005, + "step": 25878 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 7.7377, + "step": 25879 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25880 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25881 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 3.9191, + "step": 25882 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.2626, + "step": 25883 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 7.5407, + "step": 25884 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 6.6163, + "step": 25885 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 14.6142, + "step": 25886 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25887 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25888 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.7603, + "step": 25889 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25890 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25891 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25892 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25893 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.6872, + "step": 25894 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25895 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 1.0397, + "step": 25896 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25897 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25898 + }, + { + "epoch": 2.16, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25899 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25900 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25901 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.5138, + "step": 25902 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.507, + "step": 25903 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25904 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25905 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.2781, + "step": 25906 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25907 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.9947, + "step": 25908 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25909 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25910 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 7.4293, + "step": 25911 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 8.6343, + "step": 25912 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.3173, + "step": 25913 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25914 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25915 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25916 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 3.3435, + "step": 25917 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.007, + "step": 25918 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25919 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25920 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25921 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25922 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 5.019, + "step": 25923 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25924 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 7.8901, + "step": 25925 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25926 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25927 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.0439, + "step": 25928 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 9.91, + "step": 25929 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25930 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 8.7535, + "step": 25931 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25932 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.6981, + "step": 25933 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25934 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25935 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 7.5158, + "step": 25936 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 5.3105, + "step": 25937 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.8907, + "step": 25938 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25939 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.3578, + "step": 25940 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25941 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.8042, + "step": 25942 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.7983, + "step": 25943 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25944 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25945 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25946 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25947 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.9888, + "step": 25948 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25949 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25950 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.0675, + "step": 25951 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25952 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 3.0079, + "step": 25953 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25954 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.0792, + "step": 25955 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25956 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 8.5746, + "step": 25957 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25958 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.6679, + "step": 25959 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 6.5388, + "step": 25960 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25961 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 7.0568, + "step": 25962 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.8507, + "step": 25963 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25964 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25965 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25966 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.9838, + "step": 25967 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25968 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 8.542, + "step": 25969 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.312, + "step": 25970 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 5.2444, + "step": 25971 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25972 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25973 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25974 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 7.1709, + "step": 25975 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25976 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25977 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.8192, + "step": 25978 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.8417, + "step": 25979 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25980 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.9599, + "step": 25981 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25982 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.881, + "step": 25983 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25984 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.1342, + "step": 25985 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25986 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 7.6073, + "step": 25987 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25988 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25989 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.9626, + "step": 25990 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.8407, + "step": 25991 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25992 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.2568, + "step": 25993 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.88, + "step": 25994 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.703, + "step": 25995 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25996 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 25997 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 2.9985, + "step": 25998 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.9495, + "step": 25999 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 6.4014, + "step": 26000 + }, + { + "epoch": 2.17, + "eval_loss": NaN, + "eval_runtime": 44.6373, + "eval_samples_per_second": 56.007, + "eval_steps_per_second": 7.012, + "step": 26000 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26001 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 3.4565, + "step": 26002 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26003 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26004 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26005 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26006 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26007 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26008 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.264, + "step": 26009 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.9196, + "step": 26010 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26011 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.1516, + "step": 26012 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26013 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 1.0957, + "step": 26014 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 8.5499, + "step": 26015 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 3.2055, + "step": 26016 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26017 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26018 + }, + { + "epoch": 2.17, + "learning_rate": 0.00019986624683032854, + "loss": 4.779, + "step": 26019 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.7633, + "step": 26020 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26021 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26022 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26023 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26024 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26025 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26026 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26027 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 8.3853, + "step": 26028 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.0496, + "step": 26029 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26030 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.2435, + "step": 26031 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 8.1026, + "step": 26032 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.7331, + "step": 26033 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26034 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 3.6448, + "step": 26035 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26036 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 4.6035, + "step": 26037 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26038 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26039 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 3.1723, + "step": 26040 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.0329, + "step": 26041 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26042 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.844, + "step": 26043 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.0146, + "step": 26044 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26045 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.891, + "step": 26046 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.109, + "step": 26047 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26048 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26049 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.388, + "step": 26050 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26051 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 4.1982, + "step": 26052 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26053 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26054 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26055 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26056 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26057 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26058 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26059 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26060 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26061 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.925, + "step": 26062 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26063 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.6368, + "step": 26064 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26065 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26066 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 8.2108, + "step": 26067 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.0543, + "step": 26068 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.6371, + "step": 26069 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 6.4113, + "step": 26070 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26071 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.9315, + "step": 26072 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26073 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26074 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26075 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 6.9455, + "step": 26076 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26077 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26078 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26079 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26080 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.99, + "step": 26081 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 3.1642, + "step": 26082 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.3247, + "step": 26083 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 8.2922, + "step": 26084 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 3.2467, + "step": 26085 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26086 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.9359, + "step": 26087 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26088 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26089 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 3.8799, + "step": 26090 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.6843, + "step": 26091 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26092 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.1471, + "step": 26093 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.9224, + "step": 26094 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 6.993, + "step": 26095 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26096 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26097 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26098 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 5.328, + "step": 26099 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26100 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.1305, + "step": 26101 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.0433, + "step": 26102 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26103 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 4.3403, + "step": 26104 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26105 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26106 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.2859, + "step": 26107 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 3.3473, + "step": 26108 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26109 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26110 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.2748, + "step": 26111 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.7447, + "step": 26112 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26113 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26114 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.5174, + "step": 26115 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26116 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26117 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26118 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26119 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26120 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.0633, + "step": 26121 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 4.3846, + "step": 26122 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26123 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26124 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26125 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26126 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.5067, + "step": 26127 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26128 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26129 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 9.7615, + "step": 26130 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 1.23, + "step": 26131 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 2.3424, + "step": 26132 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.7831, + "step": 26133 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 5.2803, + "step": 26134 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 8.4408, + "step": 26135 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26136 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 3.2298, + "step": 26137 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 7.6571, + "step": 26138 + }, + { + "epoch": 2.18, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26139 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26140 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.4761, + "step": 26141 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26142 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.7154, + "step": 26143 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26144 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26145 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.7994, + "step": 26146 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.8579, + "step": 26147 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26148 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26149 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26150 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.6726, + "step": 26151 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26152 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26153 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.9777, + "step": 26154 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26155 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.2656, + "step": 26156 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26157 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 9.8666, + "step": 26158 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26159 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26160 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.2071, + "step": 26161 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26162 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26163 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.0551, + "step": 26164 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26165 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26166 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26167 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.744, + "step": 26168 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26169 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 5.9678, + "step": 26170 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.5675, + "step": 26171 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.8866, + "step": 26172 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.1139, + "step": 26173 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 6.2757, + "step": 26174 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.2884, + "step": 26175 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26176 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.766, + "step": 26177 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 8.2195, + "step": 26178 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26179 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26180 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26181 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26182 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26183 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.1322, + "step": 26184 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.0236, + "step": 26185 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26186 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 6.8363, + "step": 26187 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26188 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26189 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26190 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26191 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26192 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26193 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.889, + "step": 26194 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.5196, + "step": 26195 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26196 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26197 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.5511, + "step": 26198 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.9104, + "step": 26199 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26200 + }, + { + "epoch": 2.19, + "eval_loss": NaN, + "eval_runtime": 44.9064, + "eval_samples_per_second": 55.671, + "eval_steps_per_second": 6.97, + "step": 26200 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26201 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26202 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 2.9839, + "step": 26203 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26204 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 9.378, + "step": 26205 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26206 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 10.7108, + "step": 26207 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 8.5288, + "step": 26208 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26209 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 8.0344, + "step": 26210 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 2.1371, + "step": 26211 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.8815, + "step": 26212 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26213 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.9963, + "step": 26214 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26215 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.7181, + "step": 26216 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26217 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26218 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 2.4133, + "step": 26219 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26220 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.1201, + "step": 26221 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26222 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 7.1313, + "step": 26223 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26224 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.0191, + "step": 26225 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26226 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26227 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26228 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26229 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26230 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.9276, + "step": 26231 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 7.4249, + "step": 26232 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 9.8476, + "step": 26233 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26234 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.6528, + "step": 26235 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.2266, + "step": 26236 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 4.8944, + "step": 26237 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 2.5151, + "step": 26238 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26239 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.1874, + "step": 26240 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 7.4782, + "step": 26241 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26242 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 6.1633, + "step": 26243 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26244 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 8.0601, + "step": 26245 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 1.529, + "step": 26246 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 2.0182, + "step": 26247 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 8.7044, + "step": 26248 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26249 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.2925, + "step": 26250 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26251 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.9052, + "step": 26252 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26253 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 8.4483, + "step": 26254 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26255 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26256 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26257 + }, + { + "epoch": 2.19, + "learning_rate": 0.00019986624683032854, + "loss": 3.5361, + "step": 26258 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26259 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26260 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 5.3249, + "step": 26261 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26262 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.2155, + "step": 26263 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26264 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 4.5605, + "step": 26265 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 8.3867, + "step": 26266 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26267 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26268 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 4.6053, + "step": 26269 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 4.2815, + "step": 26270 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 9.1022, + "step": 26271 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 8.3891, + "step": 26272 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26273 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26274 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.8284, + "step": 26275 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26276 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26277 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.142, + "step": 26278 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26279 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.0062, + "step": 26280 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 4.7835, + "step": 26281 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26282 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26283 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.932, + "step": 26284 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.1125, + "step": 26285 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.8178, + "step": 26286 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26287 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.0285, + "step": 26288 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26289 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26290 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 7.2127, + "step": 26291 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26292 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.3311, + "step": 26293 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.0009, + "step": 26294 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.8332, + "step": 26295 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26296 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26297 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26298 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26299 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.8923, + "step": 26300 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26301 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26302 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26303 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26304 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.9173, + "step": 26305 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.2546, + "step": 26306 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26307 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.4459, + "step": 26308 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 4.4798, + "step": 26309 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26310 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.9121, + "step": 26311 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26312 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26313 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26314 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26315 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26316 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.661, + "step": 26317 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26318 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.9405, + "step": 26319 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.4674, + "step": 26320 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26321 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 8.6813, + "step": 26322 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 8.6111, + "step": 26323 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 6.4514, + "step": 26324 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26325 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.0526, + "step": 26326 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.6495, + "step": 26327 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26328 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26329 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.759, + "step": 26330 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26331 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.3148, + "step": 26332 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26333 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.111, + "step": 26334 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26335 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.1184, + "step": 26336 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 7.1001, + "step": 26337 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26338 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 7.585, + "step": 26339 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26340 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26341 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26342 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.4033, + "step": 26343 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26344 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26345 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.5748, + "step": 26346 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.7993, + "step": 26347 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 8.3588, + "step": 26348 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 7.0627, + "step": 26349 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.7316, + "step": 26350 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26351 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 9.3713, + "step": 26352 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26353 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.8247, + "step": 26354 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26355 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.419, + "step": 26356 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26357 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26358 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.9104, + "step": 26359 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.7843, + "step": 26360 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.9284, + "step": 26361 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 14.1512, + "step": 26362 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.87, + "step": 26363 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.729, + "step": 26364 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26365 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 1.8331, + "step": 26366 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26367 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 9.4531, + "step": 26368 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26369 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26370 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 6.4878, + "step": 26371 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26372 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 2.3979, + "step": 26373 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26374 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 6.2789, + "step": 26375 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 3.5933, + "step": 26376 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26377 + }, + { + "epoch": 2.2, + "learning_rate": 0.00019986624683032854, + "loss": 4.1838, + "step": 26378 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 6.1699, + "step": 26379 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26380 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.2964, + "step": 26381 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 7.593, + "step": 26382 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26383 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 9.6402, + "step": 26384 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.8051, + "step": 26385 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.6192, + "step": 26386 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26387 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.9122, + "step": 26388 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26389 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.8919, + "step": 26390 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 8.1849, + "step": 26391 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.9289, + "step": 26392 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26393 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 5.0429, + "step": 26394 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26395 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26396 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 4.3242, + "step": 26397 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26398 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.2209, + "step": 26399 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 6.6699, + "step": 26400 + }, + { + "epoch": 2.21, + "eval_loss": NaN, + "eval_runtime": 46.6907, + "eval_samples_per_second": 53.544, + "eval_steps_per_second": 6.704, + "step": 26400 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26401 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 4.6262, + "step": 26402 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 8.8352, + "step": 26403 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.028, + "step": 26404 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.875, + "step": 26405 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26406 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.2593, + "step": 26407 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 2.675, + "step": 26408 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 6.8227, + "step": 26409 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 5.2837, + "step": 26410 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 7.1255, + "step": 26411 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26412 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.914, + "step": 26413 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26414 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26415 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.179, + "step": 26416 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26417 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.857, + "step": 26418 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26419 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26420 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26421 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.4246, + "step": 26422 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26423 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26424 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26425 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 4.6721, + "step": 26426 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26427 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26428 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26429 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.9951, + "step": 26430 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26431 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 7.1083, + "step": 26432 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 6.1681, + "step": 26433 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26434 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.8115, + "step": 26435 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 8.8485, + "step": 26436 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.3236, + "step": 26437 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26438 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26439 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26440 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26441 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 5.1836, + "step": 26442 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26443 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26444 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 9.7541, + "step": 26445 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 4.4106, + "step": 26446 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26447 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26448 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26449 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.8696, + "step": 26450 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26451 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 8.454, + "step": 26452 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26453 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 8.9227, + "step": 26454 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.6696, + "step": 26455 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26456 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26457 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26458 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26459 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26460 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.8553, + "step": 26461 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26462 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 2.9921, + "step": 26463 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26464 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.798, + "step": 26465 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.8979, + "step": 26466 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.6732, + "step": 26467 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 7.0705, + "step": 26468 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26469 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 7.6478, + "step": 26470 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 7.7606, + "step": 26471 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26472 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 4.0163, + "step": 26473 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26474 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 7.5481, + "step": 26475 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.7519, + "step": 26476 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 2.3329, + "step": 26477 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.57, + "step": 26478 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 5.9657, + "step": 26479 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.2448, + "step": 26480 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26481 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26482 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26483 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26484 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.8339, + "step": 26485 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.9836, + "step": 26486 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 2.5815, + "step": 26487 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 12.2708, + "step": 26488 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 5.2626, + "step": 26489 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 3.4936, + "step": 26490 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 8.6572, + "step": 26491 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26492 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26493 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 1.1134, + "step": 26494 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26495 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26496 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26497 + }, + { + "epoch": 2.21, + "learning_rate": 0.00019986624683032854, + "loss": 2.1191, + "step": 26498 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.0131, + "step": 26499 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.9828, + "step": 26500 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 2.4216, + "step": 26501 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.7884, + "step": 26502 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26503 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26504 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26505 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26506 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26507 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26508 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.6705, + "step": 26509 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26510 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26511 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26512 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.8322, + "step": 26513 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 8.3252, + "step": 26514 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 6.0153, + "step": 26515 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26516 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 4.6529, + "step": 26517 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.0081, + "step": 26518 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26519 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26520 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 8.2359, + "step": 26521 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 8.2583, + "step": 26522 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 6.8357, + "step": 26523 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 9.6085, + "step": 26524 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26525 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26526 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26527 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26528 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26529 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.9293, + "step": 26530 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26531 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26532 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26533 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26534 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 6.6925, + "step": 26535 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26536 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26537 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 5.0362, + "step": 26538 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.126, + "step": 26539 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.9386, + "step": 26540 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.9173, + "step": 26541 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.8162, + "step": 26542 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.0567, + "step": 26543 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26544 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.2332, + "step": 26545 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 8.2951, + "step": 26546 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26547 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26548 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.9479, + "step": 26549 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 4.1591, + "step": 26550 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 2.1605, + "step": 26551 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26552 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 2.0328, + "step": 26553 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26554 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.9823, + "step": 26555 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26556 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 4.9243, + "step": 26557 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.7314, + "step": 26558 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.8717, + "step": 26559 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.8317, + "step": 26560 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.2727, + "step": 26561 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.8064, + "step": 26562 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.7719, + "step": 26563 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.7578, + "step": 26564 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26565 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.0758, + "step": 26566 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 8.3554, + "step": 26567 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.8968, + "step": 26568 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 8.9896, + "step": 26569 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.8242, + "step": 26570 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26571 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 4.3831, + "step": 26572 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26573 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 2.4429, + "step": 26574 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.0109, + "step": 26575 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26576 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.8774, + "step": 26577 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 4.028, + "step": 26578 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26579 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26580 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26581 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.8492, + "step": 26582 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 5.219, + "step": 26583 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.9892, + "step": 26584 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.9775, + "step": 26585 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.9435, + "step": 26586 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26587 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26588 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 2.0528, + "step": 26589 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 1.234, + "step": 26590 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 4.0162, + "step": 26591 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.7857, + "step": 26592 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26593 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26594 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26595 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.5782, + "step": 26596 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26597 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26598 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26599 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26600 + }, + { + "epoch": 2.22, + "eval_loss": NaN, + "eval_runtime": 45.8876, + "eval_samples_per_second": 54.481, + "eval_steps_per_second": 6.821, + "step": 26600 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26601 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26602 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.4612, + "step": 26603 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.644, + "step": 26604 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 3.576, + "step": 26605 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26606 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26607 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26608 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.6581, + "step": 26609 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26610 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 4.4976, + "step": 26611 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26612 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.8612, + "step": 26613 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 7.2464, + "step": 26614 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26615 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 9.5335, + "step": 26616 + }, + { + "epoch": 2.22, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26617 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.1927, + "step": 26618 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.0642, + "step": 26619 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26620 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26621 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26622 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.1625, + "step": 26623 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 4.4543, + "step": 26624 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.0094, + "step": 26625 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26626 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26627 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.8722, + "step": 26628 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26629 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26630 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.7632, + "step": 26631 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.7476, + "step": 26632 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 6.8134, + "step": 26633 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26634 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.0694, + "step": 26635 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26636 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 10.0083, + "step": 26637 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26638 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26639 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26640 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26641 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26642 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26643 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.8586, + "step": 26644 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26645 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26646 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26647 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.0695, + "step": 26648 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26649 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 5.8168, + "step": 26650 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26651 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.8106, + "step": 26652 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.0446, + "step": 26653 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.1918, + "step": 26654 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26655 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26656 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26657 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.929, + "step": 26658 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.3417, + "step": 26659 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.8379, + "step": 26660 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26661 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26662 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26663 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26664 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26665 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26666 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.1726, + "step": 26667 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26668 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 5.8123, + "step": 26669 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26670 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.9694, + "step": 26671 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26672 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.8384, + "step": 26673 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26674 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26675 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 4.0207, + "step": 26676 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.5567, + "step": 26677 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26678 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.4142, + "step": 26679 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26680 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26681 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.0276, + "step": 26682 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26683 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26684 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26685 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26686 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.7692, + "step": 26687 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.5205, + "step": 26688 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26689 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.4787, + "step": 26690 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.1722, + "step": 26691 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26692 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26693 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 4.1654, + "step": 26694 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26695 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26696 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 4.0242, + "step": 26697 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.9661, + "step": 26698 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26699 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26700 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26701 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.0867, + "step": 26702 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26703 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.6992, + "step": 26704 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26705 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.2866, + "step": 26706 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.8852, + "step": 26707 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.8172, + "step": 26708 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 4.1735, + "step": 26709 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 6.1897, + "step": 26710 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26711 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 9.7854, + "step": 26712 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.9043, + "step": 26713 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26714 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.7301, + "step": 26715 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26716 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.1724, + "step": 26717 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26718 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26719 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 4.7307, + "step": 26720 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26721 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.0945, + "step": 26722 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26723 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26724 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 1.8292, + "step": 26725 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26726 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.3472, + "step": 26727 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26728 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26729 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26730 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26731 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26732 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 3.245, + "step": 26733 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26734 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 8.9095, + "step": 26735 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 2.1261, + "step": 26736 + }, + { + "epoch": 2.23, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26737 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 7.4972, + "step": 26738 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 4.3176, + "step": 26739 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26740 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26741 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26742 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26743 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 8.6258, + "step": 26744 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 6.3076, + "step": 26745 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 7.9697, + "step": 26746 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26747 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.9558, + "step": 26748 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26749 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.6759, + "step": 26750 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 7.366, + "step": 26751 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.1741, + "step": 26752 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 11.1615, + "step": 26753 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26754 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26755 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.6542, + "step": 26756 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26757 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26758 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.9407, + "step": 26759 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26760 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26761 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26762 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.737, + "step": 26763 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.9583, + "step": 26764 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26765 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.8294, + "step": 26766 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.7777, + "step": 26767 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 8.6468, + "step": 26768 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26769 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.9612, + "step": 26770 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 6.9888, + "step": 26771 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.584, + "step": 26772 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.7265, + "step": 26773 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.8023, + "step": 26774 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.1422, + "step": 26775 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.5488, + "step": 26776 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.915, + "step": 26777 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26778 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26779 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 9.2227, + "step": 26780 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.2386, + "step": 26781 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 8.056, + "step": 26782 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.7416, + "step": 26783 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26784 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.9233, + "step": 26785 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 9.8181, + "step": 26786 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.6111, + "step": 26787 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 5.792, + "step": 26788 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26789 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26790 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.7595, + "step": 26791 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.0621, + "step": 26792 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 8.9951, + "step": 26793 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 4.2413, + "step": 26794 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 6.9382, + "step": 26795 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 7.6297, + "step": 26796 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26797 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26798 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.6327, + "step": 26799 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26800 + }, + { + "epoch": 2.24, + "eval_loss": NaN, + "eval_runtime": 46.2488, + "eval_samples_per_second": 54.055, + "eval_steps_per_second": 6.768, + "step": 26800 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.0324, + "step": 26801 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26802 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.1694, + "step": 26803 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.2648, + "step": 26804 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26805 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26806 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.8833, + "step": 26807 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26808 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 9.8526, + "step": 26809 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26810 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.2571, + "step": 26811 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.0657, + "step": 26812 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.4547, + "step": 26813 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 4.6038, + "step": 26814 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.1716, + "step": 26815 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26816 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26817 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 6.3642, + "step": 26818 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26819 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26820 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26821 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.6633, + "step": 26822 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 9.1653, + "step": 26823 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26824 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.9689, + "step": 26825 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26826 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26827 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 8.8081, + "step": 26828 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26829 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.0162, + "step": 26830 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26831 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26832 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.8137, + "step": 26833 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 8.4927, + "step": 26834 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26835 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 4.6201, + "step": 26836 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26837 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26838 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.2897, + "step": 26839 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26840 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26841 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26842 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26843 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26844 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.3126, + "step": 26845 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 2.384, + "step": 26846 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.4581, + "step": 26847 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.841, + "step": 26848 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 5.478, + "step": 26849 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26850 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 1.1376, + "step": 26851 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26852 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 3.2243, + "step": 26853 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 8.5047, + "step": 26854 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26855 + }, + { + "epoch": 2.24, + "learning_rate": 0.00019986624683032854, + "loss": 13.8459, + "step": 26856 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.6373, + "step": 26857 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26858 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.0731, + "step": 26859 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 5.6998, + "step": 26860 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.1954, + "step": 26861 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.8302, + "step": 26862 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26863 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 9.2955, + "step": 26864 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26865 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.9993, + "step": 26866 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.8706, + "step": 26867 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26868 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26869 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 8.2586, + "step": 26870 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26871 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.24, + "step": 26872 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 5.7161, + "step": 26873 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.0003, + "step": 26874 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26875 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.0481, + "step": 26876 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26877 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.0572, + "step": 26878 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 10.4221, + "step": 26879 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.5508, + "step": 26880 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 4.0248, + "step": 26881 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.1471, + "step": 26882 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 9.0593, + "step": 26883 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.8627, + "step": 26884 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.5412, + "step": 26885 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26886 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26887 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 4.4043, + "step": 26888 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26889 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26890 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.9525, + "step": 26891 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 4.1905, + "step": 26892 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.5317, + "step": 26893 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.8247, + "step": 26894 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26895 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26896 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.9437, + "step": 26897 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26898 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 9.4549, + "step": 26899 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26900 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.8807, + "step": 26901 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.3039, + "step": 26902 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 4.3083, + "step": 26903 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 9.0204, + "step": 26904 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26905 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26906 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26907 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.53, + "step": 26908 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.2869, + "step": 26909 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26910 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26911 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.1793, + "step": 26912 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.1614, + "step": 26913 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.1978, + "step": 26914 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26915 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26916 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26917 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26918 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 11.5385, + "step": 26919 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.9796, + "step": 26920 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.6393, + "step": 26921 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26922 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.4729, + "step": 26923 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26924 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.9346, + "step": 26925 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.0202, + "step": 26926 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26927 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26928 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26929 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26930 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26931 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 4.0714, + "step": 26932 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.6148, + "step": 26933 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 8.2766, + "step": 26934 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.9849, + "step": 26935 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.9992, + "step": 26936 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 5.5926, + "step": 26937 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.9353, + "step": 26938 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 5.7474, + "step": 26939 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.8029, + "step": 26940 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26941 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.3572, + "step": 26942 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 8.4231, + "step": 26943 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26944 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26945 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.1419, + "step": 26946 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.9667, + "step": 26947 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.0398, + "step": 26948 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.0804, + "step": 26949 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.0436, + "step": 26950 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.016, + "step": 26951 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26952 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26953 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26954 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26955 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 3.5104, + "step": 26956 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.0, + "step": 26957 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26958 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26959 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26960 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 6.4902, + "step": 26961 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.8982, + "step": 26962 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 7.5146, + "step": 26963 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26964 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.0555, + "step": 26965 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.9458, + "step": 26966 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26967 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26968 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 8.9877, + "step": 26969 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26970 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 11.1796, + "step": 26971 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 2.4685, + "step": 26972 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26973 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26974 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26975 + }, + { + "epoch": 2.25, + "learning_rate": 0.00019986624683032854, + "loss": 1.1046, + "step": 26976 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26977 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.9374, + "step": 26978 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.001, + "step": 26979 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 8.356, + "step": 26980 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26981 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 9.2228, + "step": 26982 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 6.2409, + "step": 26983 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26984 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26985 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26986 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 6.7886, + "step": 26987 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26988 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.9448, + "step": 26989 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.0765, + "step": 26990 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.8593, + "step": 26991 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 3.8977, + "step": 26992 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26993 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 8.0234, + "step": 26994 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 26995 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.9109, + "step": 26996 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.9343, + "step": 26997 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.2241, + "step": 26998 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 8.3621, + "step": 26999 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27000 + }, + { + "epoch": 2.26, + "eval_loss": NaN, + "eval_runtime": 47.203, + "eval_samples_per_second": 52.963, + "eval_steps_per_second": 6.631, + "step": 27000 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.3469, + "step": 27001 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.0551, + "step": 27002 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27003 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 7.1545, + "step": 27004 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27005 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27006 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27007 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.8191, + "step": 27008 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.9767, + "step": 27009 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27010 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.0249, + "step": 27011 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27012 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27013 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27014 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27015 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27016 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.8947, + "step": 27017 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.3892, + "step": 27018 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.8745, + "step": 27019 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27020 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.1832, + "step": 27021 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27022 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 3.9337, + "step": 27023 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.553, + "step": 27024 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.5506, + "step": 27025 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 7.9076, + "step": 27026 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27027 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.7975, + "step": 27028 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27029 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.8712, + "step": 27030 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 3.3304, + "step": 27031 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27032 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27033 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.0931, + "step": 27034 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27035 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27036 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 7.8264, + "step": 27037 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.0152, + "step": 27038 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27039 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.8948, + "step": 27040 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27041 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 5.8655, + "step": 27042 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 3.8267, + "step": 27043 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.8641, + "step": 27044 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.9256, + "step": 27045 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27046 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 9.4441, + "step": 27047 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27048 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27049 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27050 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.8739, + "step": 27051 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 3.792, + "step": 27052 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27053 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 8.4318, + "step": 27054 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.6277, + "step": 27055 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27056 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27057 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.962, + "step": 27058 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.192, + "step": 27059 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27060 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 7.8961, + "step": 27061 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.804, + "step": 27062 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 4.4456, + "step": 27063 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 3.4476, + "step": 27064 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 5.4411, + "step": 27065 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 3.1868, + "step": 27066 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 8.6438, + "step": 27067 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27068 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 7.064, + "step": 27069 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.0244, + "step": 27070 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27071 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27072 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.7878, + "step": 27073 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 5.1009, + "step": 27074 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 9.3389, + "step": 27075 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27076 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27077 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27078 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 8.1065, + "step": 27079 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27080 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27081 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.7038, + "step": 27082 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 7.133, + "step": 27083 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27084 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 2.2529, + "step": 27085 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27086 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27087 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 8.7612, + "step": 27088 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 5.3059, + "step": 27089 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27090 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27091 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27092 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27093 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27094 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 1.061, + "step": 27095 + }, + { + "epoch": 2.26, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27096 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27097 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27098 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27099 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.8947, + "step": 27100 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.182, + "step": 27101 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27102 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.0112, + "step": 27103 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27104 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27105 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27106 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.9123, + "step": 27107 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27108 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27109 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 7.0362, + "step": 27110 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27111 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.4233, + "step": 27112 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.8577, + "step": 27113 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.7508, + "step": 27114 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.1275, + "step": 27115 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27116 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27117 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.4939, + "step": 27118 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.2153, + "step": 27119 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 7.4353, + "step": 27120 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 6.6189, + "step": 27121 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 7.9775, + "step": 27122 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27123 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.5865, + "step": 27124 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27125 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27126 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27127 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.5413, + "step": 27128 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27129 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27130 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27131 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.9043, + "step": 27132 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27133 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.7515, + "step": 27134 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27135 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.0193, + "step": 27136 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.3565, + "step": 27137 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.6073, + "step": 27138 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27139 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.3458, + "step": 27140 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.9993, + "step": 27141 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.6052, + "step": 27142 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27143 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 6.4696, + "step": 27144 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 8.4753, + "step": 27145 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.2854, + "step": 27146 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27147 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 5.5696, + "step": 27148 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27149 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.6962, + "step": 27150 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.9267, + "step": 27151 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.6804, + "step": 27152 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.9293, + "step": 27153 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27154 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.3542, + "step": 27155 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27156 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27157 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.6242, + "step": 27158 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.2968, + "step": 27159 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.9362, + "step": 27160 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.5732, + "step": 27161 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.3779, + "step": 27162 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.7233, + "step": 27163 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.8913, + "step": 27164 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.7763, + "step": 27165 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.8751, + "step": 27166 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 8.7506, + "step": 27167 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.6338, + "step": 27168 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.7239, + "step": 27169 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27170 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27171 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.0876, + "step": 27172 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27173 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.972, + "step": 27174 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27175 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.1285, + "step": 27176 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27177 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 1.0296, + "step": 27178 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 8.6998, + "step": 27179 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27180 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.4901, + "step": 27181 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.868, + "step": 27182 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.5194, + "step": 27183 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 8.8038, + "step": 27184 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27185 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27186 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 2.9675, + "step": 27187 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 8.351, + "step": 27188 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.5196, + "step": 27189 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 5.5101, + "step": 27190 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27191 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27192 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27193 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.8972, + "step": 27194 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27195 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27196 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.6292, + "step": 27197 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 5.0658, + "step": 27198 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27199 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 4.479, + "step": 27200 + }, + { + "epoch": 2.27, + "eval_loss": NaN, + "eval_runtime": 48.7832, + "eval_samples_per_second": 51.247, + "eval_steps_per_second": 6.416, + "step": 27200 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 3.8488, + "step": 27201 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 6.4491, + "step": 27202 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27203 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27204 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 10.1044, + "step": 27205 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 7.4827, + "step": 27206 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27207 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27208 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27209 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27210 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27211 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27212 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27213 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27214 + }, + { + "epoch": 2.27, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27215 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 8.8416, + "step": 27216 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27217 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27218 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.5994, + "step": 27219 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.3088, + "step": 27220 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 8.7961, + "step": 27221 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27222 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27223 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.5448, + "step": 27224 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.4382, + "step": 27225 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.7229, + "step": 27226 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.1461, + "step": 27227 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27228 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27229 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 1.9293, + "step": 27230 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27231 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27232 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27233 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27234 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27235 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.5856, + "step": 27236 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 5.0141, + "step": 27237 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 4.2299, + "step": 27238 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27239 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.9538, + "step": 27240 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.3662, + "step": 27241 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27242 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27243 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.3383, + "step": 27244 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27245 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 1.0318, + "step": 27246 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27247 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27248 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.3144, + "step": 27249 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27250 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27251 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27252 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.073, + "step": 27253 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27254 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.8888, + "step": 27255 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27256 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27257 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.5418, + "step": 27258 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.0845, + "step": 27259 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.1811, + "step": 27260 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 4.3353, + "step": 27261 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.7348, + "step": 27262 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 11.5645, + "step": 27263 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27264 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.9472, + "step": 27265 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 1.8139, + "step": 27266 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.4719, + "step": 27267 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27268 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 4.2387, + "step": 27269 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 5.0365, + "step": 27270 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27271 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 4.2233, + "step": 27272 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27273 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 5.7917, + "step": 27274 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27275 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 4.2253, + "step": 27276 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27277 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27278 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.1306, + "step": 27279 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 1.2896, + "step": 27280 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27281 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27282 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27283 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27284 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.8945, + "step": 27285 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27286 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 6.7331, + "step": 27287 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.6439, + "step": 27288 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 10.071, + "step": 27289 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.2106, + "step": 27290 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27291 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 9.2023, + "step": 27292 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27293 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.9932, + "step": 27294 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 5.6609, + "step": 27295 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27296 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 10.751, + "step": 27297 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.0691, + "step": 27298 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27299 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27300 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 5.2596, + "step": 27301 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 9.3227, + "step": 27302 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.956, + "step": 27303 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27304 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.9118, + "step": 27305 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27306 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 4.6136, + "step": 27307 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.86, + "step": 27308 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27309 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.1832, + "step": 27310 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.6462, + "step": 27311 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27312 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27313 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27314 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27315 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 6.614, + "step": 27316 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27317 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27318 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.8054, + "step": 27319 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 4.4231, + "step": 27320 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27321 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27322 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 7.869, + "step": 27323 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 1.4693, + "step": 27324 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27325 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 8.2449, + "step": 27326 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27327 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 2.0728, + "step": 27328 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27329 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.4458, + "step": 27330 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27331 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27332 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27333 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27334 + }, + { + "epoch": 2.28, + "learning_rate": 0.00019986624683032854, + "loss": 3.0867, + "step": 27335 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27336 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 3.1082, + "step": 27337 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27338 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.9044, + "step": 27339 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27340 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.798, + "step": 27341 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.8845, + "step": 27342 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27343 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 6.2233, + "step": 27344 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.9556, + "step": 27345 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.4604, + "step": 27346 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 7.3842, + "step": 27347 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.8926, + "step": 27348 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.2272, + "step": 27349 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27350 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.8919, + "step": 27351 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27352 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.9882, + "step": 27353 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.2487, + "step": 27354 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.9419, + "step": 27355 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.1163, + "step": 27356 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.5535, + "step": 27357 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.2155, + "step": 27358 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 5.0085, + "step": 27359 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.6826, + "step": 27360 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27361 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27362 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.1968, + "step": 27363 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27364 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27365 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.0168, + "step": 27366 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27367 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.903, + "step": 27368 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.542, + "step": 27369 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27370 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27371 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 10.9158, + "step": 27372 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27373 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27374 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27375 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.2316, + "step": 27376 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27377 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 3.8604, + "step": 27378 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 9.2908, + "step": 27379 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27380 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 8.2676, + "step": 27381 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27382 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.0824, + "step": 27383 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 13.1796, + "step": 27384 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 9.8585, + "step": 27385 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27386 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.1757, + "step": 27387 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.0953, + "step": 27388 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27389 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.5016, + "step": 27390 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27391 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27392 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.1369, + "step": 27393 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27394 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.9414, + "step": 27395 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 7.4836, + "step": 27396 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27397 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 6.2404, + "step": 27398 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27399 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27400 + }, + { + "epoch": 2.29, + "eval_loss": NaN, + "eval_runtime": 47.0235, + "eval_samples_per_second": 53.165, + "eval_steps_per_second": 6.656, + "step": 27400 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27401 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27402 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 7.6503, + "step": 27403 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 5.9295, + "step": 27404 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27405 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.9887, + "step": 27406 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.0411, + "step": 27407 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 3.9064, + "step": 27408 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27409 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27410 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 3.6146, + "step": 27411 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27412 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 8.9569, + "step": 27413 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27414 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.1624, + "step": 27415 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.115, + "step": 27416 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.158, + "step": 27417 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 10.8368, + "step": 27418 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27419 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27420 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.1716, + "step": 27421 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27422 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27423 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.154, + "step": 27424 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27425 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.134, + "step": 27426 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27427 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.7817, + "step": 27428 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 3.3605, + "step": 27429 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.2686, + "step": 27430 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 3.8119, + "step": 27431 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.9269, + "step": 27432 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27433 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.2981, + "step": 27434 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27435 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27436 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 3.6934, + "step": 27437 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27438 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.9484, + "step": 27439 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 8.2662, + "step": 27440 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27441 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27442 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.8584, + "step": 27443 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27444 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 10.714, + "step": 27445 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 7.9982, + "step": 27446 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27447 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27448 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 4.6196, + "step": 27449 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27450 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27451 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27452 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 2.749, + "step": 27453 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27454 + }, + { + "epoch": 2.29, + "learning_rate": 0.00019986624683032854, + "loss": 1.959, + "step": 27455 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27456 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27457 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 9.118, + "step": 27458 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27459 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27460 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27461 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27462 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27463 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.834, + "step": 27464 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 8.5822, + "step": 27465 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.5428, + "step": 27466 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27467 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.7298, + "step": 27468 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27469 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27470 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27471 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27472 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.8674, + "step": 27473 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.3457, + "step": 27474 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27475 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.1098, + "step": 27476 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.3123, + "step": 27477 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27478 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27479 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.8502, + "step": 27480 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27481 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27482 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.835, + "step": 27483 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.7196, + "step": 27484 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27485 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27486 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27487 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.7069, + "step": 27488 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.0554, + "step": 27489 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 2.0864, + "step": 27490 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.0497, + "step": 27491 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.8048, + "step": 27492 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27493 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.8566, + "step": 27494 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27495 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 2.2494, + "step": 27496 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27497 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27498 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.9782, + "step": 27499 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.967, + "step": 27500 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27501 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27502 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.1965, + "step": 27503 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.9169, + "step": 27504 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27505 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.5061, + "step": 27506 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27507 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 2.0092, + "step": 27508 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 8.8451, + "step": 27509 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 4.891, + "step": 27510 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27511 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27512 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27513 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.8414, + "step": 27514 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 2.1624, + "step": 27515 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27516 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27517 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27518 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.563, + "step": 27519 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 6.5731, + "step": 27520 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.924, + "step": 27521 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 2.3456, + "step": 27522 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.9494, + "step": 27523 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27524 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.9403, + "step": 27525 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.7297, + "step": 27526 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27527 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27528 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.3959, + "step": 27529 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.9357, + "step": 27530 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27531 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.928, + "step": 27532 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27533 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 5.0348, + "step": 27534 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.8696, + "step": 27535 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27536 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.8683, + "step": 27537 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 10.5979, + "step": 27538 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.6801, + "step": 27539 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 4.5843, + "step": 27540 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 8.1665, + "step": 27541 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 9.1935, + "step": 27542 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.8977, + "step": 27543 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 6.9318, + "step": 27544 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 4.1039, + "step": 27545 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 2.6802, + "step": 27546 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.6276, + "step": 27547 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27548 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.5422, + "step": 27549 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.4516, + "step": 27550 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 4.3484, + "step": 27551 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.7541, + "step": 27552 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27553 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.386, + "step": 27554 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 7.6991, + "step": 27555 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27556 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27557 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 2.9024, + "step": 27558 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.9679, + "step": 27559 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 9.7619, + "step": 27560 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 1.7754, + "step": 27561 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27562 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27563 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 4.827, + "step": 27564 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27565 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 4.5056, + "step": 27566 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27567 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27568 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 4.4395, + "step": 27569 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27570 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.7959, + "step": 27571 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27572 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27573 + }, + { + "epoch": 2.3, + "learning_rate": 0.00019986624683032854, + "loss": 3.9949, + "step": 27574 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27575 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.0456, + "step": 27576 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27577 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 6.1455, + "step": 27578 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27579 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.9554, + "step": 27580 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 10.9012, + "step": 27581 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.9493, + "step": 27582 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27583 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27584 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 8.7525, + "step": 27585 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.8159, + "step": 27586 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27587 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27588 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 3.0934, + "step": 27589 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 2.1644, + "step": 27590 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27591 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 2.2881, + "step": 27592 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.8522, + "step": 27593 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.7127, + "step": 27594 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 3.6026, + "step": 27595 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27596 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27597 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27598 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27599 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27600 + }, + { + "epoch": 2.31, + "eval_loss": NaN, + "eval_runtime": 49.61, + "eval_samples_per_second": 50.393, + "eval_steps_per_second": 6.309, + "step": 27600 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 6.2183, + "step": 27601 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.958, + "step": 27602 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.0175, + "step": 27603 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.1165, + "step": 27604 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27605 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 3.7351, + "step": 27606 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 5.3714, + "step": 27607 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 5.3854, + "step": 27608 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.0482, + "step": 27609 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.7198, + "step": 27610 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27611 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27612 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27613 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.8891, + "step": 27614 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27615 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27616 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27617 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.2694, + "step": 27618 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 2.1612, + "step": 27619 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27620 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27621 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27622 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.9875, + "step": 27623 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 8.0664, + "step": 27624 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 2.1448, + "step": 27625 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27626 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.7553, + "step": 27627 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 6.6266, + "step": 27628 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27629 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27630 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27631 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 8.6341, + "step": 27632 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27633 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27634 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.0195, + "step": 27635 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27636 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.6217, + "step": 27637 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 12.7903, + "step": 27638 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 2.1921, + "step": 27639 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 6.6645, + "step": 27640 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27641 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 5.2565, + "step": 27642 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27643 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27644 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.8482, + "step": 27645 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27646 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.9129, + "step": 27647 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27648 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.5613, + "step": 27649 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.0676, + "step": 27650 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27651 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.8535, + "step": 27652 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 3.9758, + "step": 27653 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27654 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27655 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 3.5629, + "step": 27656 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27657 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27658 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27659 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 2.9983, + "step": 27660 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 9.3598, + "step": 27661 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 7.6738, + "step": 27662 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27663 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27664 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.9462, + "step": 27665 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.9242, + "step": 27666 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27667 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27668 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 8.3993, + "step": 27669 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.3279, + "step": 27670 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.2357, + "step": 27671 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 5.6747, + "step": 27672 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.7497, + "step": 27673 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27674 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27675 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27676 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27677 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.8019, + "step": 27678 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 4.1364, + "step": 27679 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.8513, + "step": 27680 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27681 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 3.0599, + "step": 27682 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 9.404, + "step": 27683 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27684 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27685 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27686 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27687 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27688 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27689 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 1.9472, + "step": 27690 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27691 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 3.2009, + "step": 27692 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27693 + }, + { + "epoch": 2.31, + "learning_rate": 0.00019986624683032854, + "loss": 2.9738, + "step": 27694 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.9963, + "step": 27695 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 9.7674, + "step": 27696 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 5.8426, + "step": 27697 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.3428, + "step": 27698 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 4.0235, + "step": 27699 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27700 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 5.8196, + "step": 27701 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27702 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.9479, + "step": 27703 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 8.407, + "step": 27704 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.9735, + "step": 27705 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27706 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27707 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27708 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.0885, + "step": 27709 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27710 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27711 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.716, + "step": 27712 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27713 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.7858, + "step": 27714 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.0163, + "step": 27715 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.7891, + "step": 27716 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.2133, + "step": 27717 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.0511, + "step": 27718 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.1484, + "step": 27719 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.8433, + "step": 27720 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.765, + "step": 27721 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27722 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27723 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 6.6876, + "step": 27724 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 5.8507, + "step": 27725 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.8208, + "step": 27726 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27727 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.0335, + "step": 27728 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.6617, + "step": 27729 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 5.1966, + "step": 27730 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27731 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 5.7818, + "step": 27732 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27733 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 7.2042, + "step": 27734 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.9635, + "step": 27735 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27736 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27737 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27738 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27739 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.0398, + "step": 27740 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27741 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 4.0533, + "step": 27742 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 8.3106, + "step": 27743 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27744 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 4.6403, + "step": 27745 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 6.6482, + "step": 27746 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27747 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.9813, + "step": 27748 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 4.7911, + "step": 27749 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27750 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27751 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 8.1557, + "step": 27752 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27753 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.6391, + "step": 27754 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.8746, + "step": 27755 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27756 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 6.5618, + "step": 27757 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.2105, + "step": 27758 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27759 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.7684, + "step": 27760 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27761 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.4117, + "step": 27762 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27763 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27764 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27765 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 6.5614, + "step": 27766 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27767 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 4.1247, + "step": 27768 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27769 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27770 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 4.9194, + "step": 27771 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27772 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 8.8264, + "step": 27773 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27774 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27775 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.9862, + "step": 27776 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.9189, + "step": 27777 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 8.4835, + "step": 27778 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.8896, + "step": 27779 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.9898, + "step": 27780 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27781 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27782 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.91, + "step": 27783 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27784 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27785 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27786 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.7307, + "step": 27787 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27788 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27789 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.714, + "step": 27790 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27791 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27792 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 1.247, + "step": 27793 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27794 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27795 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27796 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.7137, + "step": 27797 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27798 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27799 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.5676, + "step": 27800 + }, + { + "epoch": 2.32, + "eval_loss": NaN, + "eval_runtime": 47.6435, + "eval_samples_per_second": 52.473, + "eval_steps_per_second": 6.57, + "step": 27800 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 7.7744, + "step": 27801 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27802 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27803 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27804 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27805 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27806 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27807 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27808 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 4.2069, + "step": 27809 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.9527, + "step": 27810 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 3.2275, + "step": 27811 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27812 + }, + { + "epoch": 2.32, + "learning_rate": 0.00019986624683032854, + "loss": 2.1355, + "step": 27813 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.025, + "step": 27814 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27815 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.642, + "step": 27816 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.1685, + "step": 27817 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 4.1023, + "step": 27818 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.2453, + "step": 27819 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27820 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.5887, + "step": 27821 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.8855, + "step": 27822 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27823 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27824 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27825 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.4747, + "step": 27826 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.7482, + "step": 27827 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.8239, + "step": 27828 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 4.582, + "step": 27829 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.0105, + "step": 27830 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27831 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.5312, + "step": 27832 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.2284, + "step": 27833 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 9.8577, + "step": 27834 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 4.1672, + "step": 27835 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27836 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.9657, + "step": 27837 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.85, + "step": 27838 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.2408, + "step": 27839 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27840 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27841 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27842 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.9192, + "step": 27843 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27844 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.7199, + "step": 27845 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27846 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27847 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27848 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.944, + "step": 27849 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.0574, + "step": 27850 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.8717, + "step": 27851 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27852 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.763, + "step": 27853 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.6669, + "step": 27854 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.4828, + "step": 27855 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.8574, + "step": 27856 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27857 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.9109, + "step": 27858 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 6.8304, + "step": 27859 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 4.2534, + "step": 27860 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27861 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 8.6917, + "step": 27862 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 8.0089, + "step": 27863 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27864 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 4.1694, + "step": 27865 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.0088, + "step": 27866 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.6389, + "step": 27867 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27868 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27869 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27870 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27871 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27872 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27873 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27874 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.8679, + "step": 27875 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27876 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.6606, + "step": 27877 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.8421, + "step": 27878 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.6909, + "step": 27879 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27880 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27881 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27882 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 5.4899, + "step": 27883 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27884 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27885 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27886 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27887 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27888 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27889 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27890 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 8.3067, + "step": 27891 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27892 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.8949, + "step": 27893 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27894 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.3067, + "step": 27895 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.9714, + "step": 27896 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27897 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27898 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27899 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.5786, + "step": 27900 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.6152, + "step": 27901 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.7806, + "step": 27902 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.0925, + "step": 27903 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27904 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27905 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 1.8451, + "step": 27906 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27907 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 7.2948, + "step": 27908 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.0661, + "step": 27909 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 2.7348, + "step": 27910 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 4.0642, + "step": 27911 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27912 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 5.734, + "step": 27913 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 6.0216, + "step": 27914 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.9868, + "step": 27915 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27916 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.8254, + "step": 27917 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 9.6904, + "step": 27918 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27919 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 5.1799, + "step": 27920 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 8.8244, + "step": 27921 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 4.6002, + "step": 27922 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27923 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 6.2177, + "step": 27924 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 8.2483, + "step": 27925 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27926 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27927 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.8413, + "step": 27928 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 3.9149, + "step": 27929 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27930 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27931 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27932 + }, + { + "epoch": 2.33, + "learning_rate": 0.00019986624683032854, + "loss": 0.9903, + "step": 27933 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27934 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27935 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27936 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27937 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.8556, + "step": 27938 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.5595, + "step": 27939 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.7938, + "step": 27940 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.8259, + "step": 27941 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 5.5618, + "step": 27942 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.0535, + "step": 27943 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 6.435, + "step": 27944 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27945 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.9872, + "step": 27946 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.3147, + "step": 27947 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.8265, + "step": 27948 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.0481, + "step": 27949 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.0772, + "step": 27950 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27951 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 6.5428, + "step": 27952 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27953 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.8565, + "step": 27954 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27955 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27956 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27957 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27958 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27959 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27960 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.2185, + "step": 27961 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27962 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27963 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27964 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27965 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.1786, + "step": 27966 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.0417, + "step": 27967 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27968 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27969 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27970 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.9194, + "step": 27971 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.9829, + "step": 27972 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.953, + "step": 27973 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 5.3635, + "step": 27974 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 4.9656, + "step": 27975 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27976 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27977 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.8372, + "step": 27978 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.064, + "step": 27979 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27980 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.8163, + "step": 27981 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27982 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 9.4076, + "step": 27983 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27984 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27985 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27986 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.9441, + "step": 27987 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 4.9011, + "step": 27988 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.9249, + "step": 27989 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27990 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27991 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.1976, + "step": 27992 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 4.3213, + "step": 27993 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27994 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.5925, + "step": 27995 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.9858, + "step": 27996 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27997 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 27998 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.0836, + "step": 27999 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28000 + }, + { + "epoch": 2.34, + "eval_loss": NaN, + "eval_runtime": 48.4249, + "eval_samples_per_second": 51.626, + "eval_steps_per_second": 6.464, + "step": 28000 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.9428, + "step": 28001 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28002 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28003 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 8.509, + "step": 28004 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28005 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 10.122, + "step": 28006 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 4.1141, + "step": 28007 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28008 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28009 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28010 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.9172, + "step": 28011 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28012 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.8975, + "step": 28013 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.6957, + "step": 28014 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28015 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28016 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28017 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28018 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28019 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28020 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.8992, + "step": 28021 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 13.6535, + "step": 28022 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28023 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28024 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28025 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 7.0163, + "step": 28026 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28027 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 7.704, + "step": 28028 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28029 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 8.5736, + "step": 28030 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28031 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.899, + "step": 28032 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.9865, + "step": 28033 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 8.2006, + "step": 28034 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.7982, + "step": 28035 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.0949, + "step": 28036 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.6302, + "step": 28037 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.7632, + "step": 28038 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.9433, + "step": 28039 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28040 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28041 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.9133, + "step": 28042 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28043 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28044 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 5.4392, + "step": 28045 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 6.2632, + "step": 28046 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 2.7578, + "step": 28047 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28048 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28049 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 1.8064, + "step": 28050 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28051 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 3.9945, + "step": 28052 + }, + { + "epoch": 2.34, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28053 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.1391, + "step": 28054 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.0119, + "step": 28055 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.088, + "step": 28056 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.1374, + "step": 28057 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.9987, + "step": 28058 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.9679, + "step": 28059 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28060 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28061 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28062 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28063 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28064 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.1205, + "step": 28065 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.0191, + "step": 28066 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 6.6719, + "step": 28067 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28068 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.9501, + "step": 28069 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28070 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.5119, + "step": 28071 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 6.1418, + "step": 28072 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.9455, + "step": 28073 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28074 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.055, + "step": 28075 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28076 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.977, + "step": 28077 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.1793, + "step": 28078 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.8117, + "step": 28079 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.4728, + "step": 28080 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 5.8777, + "step": 28081 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28082 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 10.1304, + "step": 28083 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.9125, + "step": 28084 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 8.7138, + "step": 28085 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28086 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 3.4006, + "step": 28087 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28088 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28089 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.0788, + "step": 28090 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28091 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28092 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.6283, + "step": 28093 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28094 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 6.6529, + "step": 28095 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28096 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28097 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 8.6052, + "step": 28098 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28099 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 5.2019, + "step": 28100 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.9785, + "step": 28101 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28102 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 6.8994, + "step": 28103 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.3775, + "step": 28104 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28105 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.1398, + "step": 28106 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.1655, + "step": 28107 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28108 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 3.9671, + "step": 28109 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.7492, + "step": 28110 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.2903, + "step": 28111 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28112 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.0165, + "step": 28113 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.0247, + "step": 28114 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28115 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28116 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28117 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.3308, + "step": 28118 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28119 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 3.6536, + "step": 28120 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28121 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 5.8739, + "step": 28122 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.9009, + "step": 28123 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 5.7639, + "step": 28124 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 9.9025, + "step": 28125 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.8762, + "step": 28126 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28127 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.6939, + "step": 28128 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 9.9391, + "step": 28129 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 3.1799, + "step": 28130 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.8972, + "step": 28131 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28132 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.1655, + "step": 28133 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28134 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28135 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 8.0228, + "step": 28136 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.1359, + "step": 28137 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 1.9384, + "step": 28138 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 5.6806, + "step": 28139 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.0064, + "step": 28140 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.7472, + "step": 28141 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.3228, + "step": 28142 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 8.8256, + "step": 28143 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.0125, + "step": 28144 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28145 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.0729, + "step": 28146 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 8.1177, + "step": 28147 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28148 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.8175, + "step": 28149 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.9041, + "step": 28150 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.5643, + "step": 28151 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 2.0905, + "step": 28152 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.7259, + "step": 28153 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.785, + "step": 28154 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28155 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28156 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28157 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28158 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.8447, + "step": 28159 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 3.3281, + "step": 28160 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28161 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.3305, + "step": 28162 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 3.746, + "step": 28163 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28164 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.2912, + "step": 28165 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28166 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 4.6883, + "step": 28167 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 7.6487, + "step": 28168 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28169 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28170 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 9.1879, + "step": 28171 + }, + { + "epoch": 2.35, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28172 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28173 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28174 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 1.184, + "step": 28175 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.4425, + "step": 28176 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 1.0943, + "step": 28177 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28178 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28179 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 4.0213, + "step": 28180 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28181 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 7.2523, + "step": 28182 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 2.1, + "step": 28183 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 3.2112, + "step": 28184 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 2.4725, + "step": 28185 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 4.0859, + "step": 28186 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 1.8855, + "step": 28187 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28188 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 3.1855, + "step": 28189 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28190 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 1.9155, + "step": 28191 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28192 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28193 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 1.9073, + "step": 28194 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 3.4602, + "step": 28195 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28196 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 3.579, + "step": 28197 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28198 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28199 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 6.3073, + "step": 28200 + }, + { + "epoch": 2.36, + "eval_loss": NaN, + "eval_runtime": 42.1074, + "eval_samples_per_second": 59.372, + "eval_steps_per_second": 7.433, + "step": 28200 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 3.9493, + "step": 28201 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28202 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28203 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28204 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28205 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28206 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28207 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 4.323, + "step": 28208 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28209 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 4.9368, + "step": 28210 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28211 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 8.0377, + "step": 28212 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 1.8754, + "step": 28213 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 1.6693, + "step": 28214 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 3.9753, + "step": 28215 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28216 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28217 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28218 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28219 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 0.0, + "step": 28220 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 4.9515, + "step": 28221 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986624683032854, + "loss": 3.7085, + "step": 28222 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 3.5041, + "step": 28223 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 9.3619, + "step": 28224 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.9635, + "step": 28225 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28226 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28227 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 2.0069, + "step": 28228 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28229 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28230 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 4.1225, + "step": 28231 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 4.0217, + "step": 28232 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 2.0328, + "step": 28233 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 8.0295, + "step": 28234 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28235 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.0776, + "step": 28236 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.2512, + "step": 28237 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 2.1259, + "step": 28238 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28239 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 5.8952, + "step": 28240 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28241 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.5584, + "step": 28242 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28243 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28244 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28245 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28246 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.9595, + "step": 28247 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28248 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28249 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 1.976, + "step": 28250 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.9428, + "step": 28251 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 4.0202, + "step": 28252 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28253 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 1.9489, + "step": 28254 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28255 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 3.3902, + "step": 28256 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 1.6188, + "step": 28257 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28258 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 9.3229, + "step": 28259 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 6.6129, + "step": 28260 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.6772, + "step": 28261 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28262 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28263 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.9547, + "step": 28264 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28265 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 9.6179, + "step": 28266 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28267 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.7179, + "step": 28268 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 1.7677, + "step": 28269 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 3.4797, + "step": 28270 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.7904, + "step": 28271 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28272 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.8427, + "step": 28273 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 1.0863, + "step": 28274 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.6164, + "step": 28275 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28276 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28277 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28278 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28279 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.984, + "step": 28280 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28281 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 1.1942, + "step": 28282 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28283 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 7.7185, + "step": 28284 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 9.7151, + "step": 28285 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28286 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 4.102, + "step": 28287 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.9588, + "step": 28288 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28289 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28290 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 4.3842, + "step": 28291 + }, + { + "epoch": 2.36, + "learning_rate": 0.00019986067378159223, + "loss": 1.1555, + "step": 28292 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28293 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.1929, + "step": 28294 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.2472, + "step": 28295 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.1341, + "step": 28296 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.2552, + "step": 28297 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 6.5459, + "step": 28298 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28299 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 8.1586, + "step": 28300 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 8.7945, + "step": 28301 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.9709, + "step": 28302 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 6.1444, + "step": 28303 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28304 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28305 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 7.8471, + "step": 28306 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28307 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28308 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28309 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.9246, + "step": 28310 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28311 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 8.5814, + "step": 28312 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28313 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 6.9159, + "step": 28314 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28315 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 9.4039, + "step": 28316 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.5228, + "step": 28317 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 6.1965, + "step": 28318 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28319 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.7129, + "step": 28320 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.5585, + "step": 28321 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28322 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.9866, + "step": 28323 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28324 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28325 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 7.1678, + "step": 28326 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28327 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28328 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28329 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28330 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28331 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.4193, + "step": 28332 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28333 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 6.821, + "step": 28334 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 7.6783, + "step": 28335 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.303, + "step": 28336 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28337 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.9752, + "step": 28338 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.3383, + "step": 28339 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 7.6204, + "step": 28340 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28341 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28342 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.056, + "step": 28343 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.9711, + "step": 28344 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28345 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.7396, + "step": 28346 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28347 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.9637, + "step": 28348 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28349 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28350 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.6099, + "step": 28351 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28352 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 6.5579, + "step": 28353 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28354 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.9662, + "step": 28355 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 6.3793, + "step": 28356 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.971, + "step": 28357 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 8.7214, + "step": 28358 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.5924, + "step": 28359 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.1151, + "step": 28360 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28361 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.1155, + "step": 28362 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28363 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.6032, + "step": 28364 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.0339, + "step": 28365 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28366 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28367 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28368 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.077, + "step": 28369 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 8.3908, + "step": 28370 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.1955, + "step": 28371 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28372 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.1089, + "step": 28373 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 7.9393, + "step": 28374 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.8042, + "step": 28375 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.5679, + "step": 28376 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.7085, + "step": 28377 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.5958, + "step": 28378 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28379 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 4.0378, + "step": 28380 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28381 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.6722, + "step": 28382 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28383 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28384 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28385 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.1593, + "step": 28386 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28387 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.6545, + "step": 28388 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 8.481, + "step": 28389 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28390 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.055, + "step": 28391 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.0104, + "step": 28392 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 5.2907, + "step": 28393 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.0817, + "step": 28394 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.8928, + "step": 28395 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28396 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28397 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.7971, + "step": 28398 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28399 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28400 + }, + { + "epoch": 2.37, + "eval_loss": NaN, + "eval_runtime": 41.582, + "eval_samples_per_second": 60.122, + "eval_steps_per_second": 7.527, + "step": 28400 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.1659, + "step": 28401 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28402 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28403 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28404 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 2.5789, + "step": 28405 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28406 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28407 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 1.0734, + "step": 28408 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 7.7442, + "step": 28409 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.5836, + "step": 28410 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 7.6442, + "step": 28411 + }, + { + "epoch": 2.37, + "learning_rate": 0.00019986067378159223, + "loss": 3.1071, + "step": 28412 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 10.6428, + "step": 28413 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 8.1971, + "step": 28414 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.0384, + "step": 28415 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28416 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28417 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28418 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28419 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28420 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 7.7417, + "step": 28421 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28422 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 4.0982, + "step": 28423 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28424 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 2.7644, + "step": 28425 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28426 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.0222, + "step": 28427 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28428 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28429 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28430 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 3.3764, + "step": 28431 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28432 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 6.7731, + "step": 28433 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28434 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28435 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 3.2932, + "step": 28436 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28437 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.6982, + "step": 28438 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 4.6842, + "step": 28439 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 2.9938, + "step": 28440 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 8.05, + "step": 28441 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 4.0145, + "step": 28442 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 4.9731, + "step": 28443 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28444 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.4355, + "step": 28445 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.7826, + "step": 28446 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.5844, + "step": 28447 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28448 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 9.1347, + "step": 28449 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28450 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.908, + "step": 28451 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28452 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 7.9405, + "step": 28453 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28454 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.6702, + "step": 28455 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 3.345, + "step": 28456 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28457 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 9.0048, + "step": 28458 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28459 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.0714, + "step": 28460 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28461 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28462 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28463 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28464 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 6.1316, + "step": 28465 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28466 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28467 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.7499, + "step": 28468 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28469 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 7.9274, + "step": 28470 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28471 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 9.2409, + "step": 28472 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.2865, + "step": 28473 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28474 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.5219, + "step": 28475 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.886, + "step": 28476 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28477 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.0725, + "step": 28478 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 8.7345, + "step": 28479 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.1431, + "step": 28480 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28481 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 8.6179, + "step": 28482 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 2.0657, + "step": 28483 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28484 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28485 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 4.1961, + "step": 28486 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 7.3529, + "step": 28487 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28488 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 8.3245, + "step": 28489 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 6.1422, + "step": 28490 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28491 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28492 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28493 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.9731, + "step": 28494 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 3.7591, + "step": 28495 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28496 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28497 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28498 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 3.984, + "step": 28499 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.0701, + "step": 28500 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 8.4851, + "step": 28501 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28502 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.0301, + "step": 28503 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28504 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 4.7144, + "step": 28505 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.8776, + "step": 28506 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28507 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28508 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 2.2016, + "step": 28509 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28510 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 7.3432, + "step": 28511 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28512 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28513 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 3.2496, + "step": 28514 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 3.0568, + "step": 28515 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28516 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28517 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 7.7895, + "step": 28518 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28519 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28520 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28521 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 2.1189, + "step": 28522 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28523 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 1.0724, + "step": 28524 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 6.0221, + "step": 28525 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.6446, + "step": 28526 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28527 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28528 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 5.6811, + "step": 28529 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 2.0205, + "step": 28530 + }, + { + "epoch": 2.38, + "learning_rate": 0.00019986067378159223, + "loss": 7.8163, + "step": 28531 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.8652, + "step": 28532 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.0907, + "step": 28533 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.8273, + "step": 28534 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 6.0901, + "step": 28535 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.5057, + "step": 28536 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.2101, + "step": 28537 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 6.6464, + "step": 28538 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28539 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 5.4335, + "step": 28540 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.5824, + "step": 28541 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.6742, + "step": 28542 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28543 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28544 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 6.1079, + "step": 28545 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28546 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 10.7342, + "step": 28547 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28548 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 3.5292, + "step": 28549 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28550 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.1193, + "step": 28551 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 5.3966, + "step": 28552 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.592, + "step": 28553 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 5.5146, + "step": 28554 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.1049, + "step": 28555 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.9631, + "step": 28556 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.7117, + "step": 28557 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.9467, + "step": 28558 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28559 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 3.1638, + "step": 28560 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28561 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 5.8202, + "step": 28562 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.3299, + "step": 28563 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.7642, + "step": 28564 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28565 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28566 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28567 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28568 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.9457, + "step": 28569 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28570 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.6405, + "step": 28571 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28572 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28573 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.0805, + "step": 28574 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28575 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 5.035, + "step": 28576 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28577 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.8049, + "step": 28578 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.9988, + "step": 28579 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 5.2289, + "step": 28580 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.7561, + "step": 28581 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.8297, + "step": 28582 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28583 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.8791, + "step": 28584 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28585 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.4939, + "step": 28586 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 3.1691, + "step": 28587 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28588 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28589 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28590 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28591 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28592 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.4147, + "step": 28593 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.1368, + "step": 28594 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28595 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28596 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.7388, + "step": 28597 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28598 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28599 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 3.8632, + "step": 28600 + }, + { + "epoch": 2.39, + "eval_loss": NaN, + "eval_runtime": 40.2964, + "eval_samples_per_second": 62.04, + "eval_steps_per_second": 7.767, + "step": 28600 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.0013, + "step": 28601 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 2.8541, + "step": 28602 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28603 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.0314, + "step": 28604 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28605 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28606 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28607 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28608 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 11.5227, + "step": 28609 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28610 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28611 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28612 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28613 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.2204, + "step": 28614 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 3.4307, + "step": 28615 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28616 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 3.6729, + "step": 28617 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28618 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28619 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.4027, + "step": 28620 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.7943, + "step": 28621 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28622 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.8339, + "step": 28623 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28624 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28625 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.5926, + "step": 28626 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.8041, + "step": 28627 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28628 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.5014, + "step": 28629 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28630 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28631 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.4533, + "step": 28632 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28633 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28634 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.8457, + "step": 28635 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28636 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 9.794, + "step": 28637 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 2.2331, + "step": 28638 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28639 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 5.0981, + "step": 28640 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 4.7058, + "step": 28641 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.5194, + "step": 28642 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28643 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28644 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28645 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.9878, + "step": 28646 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28647 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 1.0169, + "step": 28648 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.9611, + "step": 28649 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28650 + }, + { + "epoch": 2.39, + "learning_rate": 0.00019986067378159223, + "loss": 7.3735, + "step": 28651 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 9.3624, + "step": 28652 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.9949, + "step": 28653 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.89, + "step": 28654 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.2544, + "step": 28655 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28656 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.625, + "step": 28657 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.775, + "step": 28658 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28659 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.6723, + "step": 28660 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 8.4263, + "step": 28661 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.5371, + "step": 28662 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28663 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.0536, + "step": 28664 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28665 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28666 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28667 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.5742, + "step": 28668 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.4395, + "step": 28669 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28670 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28671 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.4136, + "step": 28672 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28673 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.5976, + "step": 28674 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28675 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28676 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28677 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 5.7901, + "step": 28678 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.9682, + "step": 28679 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.6874, + "step": 28680 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.6192, + "step": 28681 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.1653, + "step": 28682 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.3384, + "step": 28683 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28684 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28685 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28686 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.0761, + "step": 28687 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.5597, + "step": 28688 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.9583, + "step": 28689 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28690 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.844, + "step": 28691 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28692 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28693 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.8992, + "step": 28694 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28695 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.4943, + "step": 28696 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.7267, + "step": 28697 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.6393, + "step": 28698 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28699 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.1284, + "step": 28700 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 5.0142, + "step": 28701 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.6508, + "step": 28702 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 8.5618, + "step": 28703 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28704 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28705 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 10.4872, + "step": 28706 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.9944, + "step": 28707 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.1489, + "step": 28708 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.6383, + "step": 28709 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28710 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.8893, + "step": 28711 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 9.5919, + "step": 28712 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28713 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.7066, + "step": 28714 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28715 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 6.3173, + "step": 28716 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28717 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.5175, + "step": 28718 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28719 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.7221, + "step": 28720 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28721 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28722 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28723 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28724 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.6823, + "step": 28725 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 9.2724, + "step": 28726 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28727 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 8.304, + "step": 28728 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28729 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 8.2182, + "step": 28730 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.8503, + "step": 28731 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.4994, + "step": 28732 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28733 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.7662, + "step": 28734 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 8.027, + "step": 28735 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.9569, + "step": 28736 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.9366, + "step": 28737 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.9703, + "step": 28738 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.9135, + "step": 28739 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.8375, + "step": 28740 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.8571, + "step": 28741 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28742 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 7.0426, + "step": 28743 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28744 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.7694, + "step": 28745 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 6.8055, + "step": 28746 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 8.414, + "step": 28747 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.7442, + "step": 28748 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28749 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28750 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28751 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28752 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.8728, + "step": 28753 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28754 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28755 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28756 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28757 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 5.9572, + "step": 28758 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.814, + "step": 28759 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28760 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28761 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.8834, + "step": 28762 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 4.1031, + "step": 28763 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28764 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.7993, + "step": 28765 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 1.7789, + "step": 28766 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 6.3594, + "step": 28767 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 5.656, + "step": 28768 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 3.182, + "step": 28769 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28770 + }, + { + "epoch": 2.4, + "learning_rate": 0.00019986067378159223, + "loss": 2.1464, + "step": 28771 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 10.1676, + "step": 28772 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28773 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.071, + "step": 28774 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28775 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28776 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28777 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28778 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28779 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28780 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 8.6128, + "step": 28781 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.8296, + "step": 28782 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.2213, + "step": 28783 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28784 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28785 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.2405, + "step": 28786 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28787 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.8746, + "step": 28788 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28789 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28790 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.8621, + "step": 28791 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28792 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28793 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28794 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 1.4962, + "step": 28795 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.3364, + "step": 28796 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 4.5643, + "step": 28797 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28798 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 8.5192, + "step": 28799 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28800 + }, + { + "epoch": 2.41, + "eval_loss": NaN, + "eval_runtime": 40.2123, + "eval_samples_per_second": 62.17, + "eval_steps_per_second": 7.784, + "step": 28800 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28801 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28802 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.625, + "step": 28803 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.6579, + "step": 28804 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28805 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.3629, + "step": 28806 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.6334, + "step": 28807 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28808 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28809 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28810 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 4.5104, + "step": 28811 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28812 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.0751, + "step": 28813 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28814 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.9764, + "step": 28815 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 5.0481, + "step": 28816 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 9.4498, + "step": 28817 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 3.4728, + "step": 28818 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28819 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 5.265, + "step": 28820 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28821 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28822 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.6671, + "step": 28823 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 4.2021, + "step": 28824 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28825 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28826 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28827 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.1449, + "step": 28828 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 9.2411, + "step": 28829 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28830 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 6.991, + "step": 28831 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28832 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28833 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28834 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 1.7712, + "step": 28835 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 5.8753, + "step": 28836 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 8.4948, + "step": 28837 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28838 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 4.5438, + "step": 28839 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28840 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 6.1062, + "step": 28841 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28842 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.8786, + "step": 28843 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28844 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.6, + "step": 28845 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 5.024, + "step": 28846 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28847 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.7974, + "step": 28848 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28849 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28850 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28851 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.0461, + "step": 28852 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.0622, + "step": 28853 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.0523, + "step": 28854 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 1.9516, + "step": 28855 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28856 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 5.3335, + "step": 28857 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28858 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.9047, + "step": 28859 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28860 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 2.5495, + "step": 28861 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 1.0065, + "step": 28862 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28863 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 4.7024, + "step": 28864 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.039, + "step": 28865 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28866 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 1.1532, + "step": 28867 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28868 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28869 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28870 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.145, + "step": 28871 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28872 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28873 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.8049, + "step": 28874 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28875 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.9083, + "step": 28876 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28877 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28878 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.9745, + "step": 28879 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28880 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 4.1681, + "step": 28881 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28882 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 5.5086, + "step": 28883 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28884 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 1.8575, + "step": 28885 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 1.0415, + "step": 28886 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28887 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28888 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 4.2202, + "step": 28889 + }, + { + "epoch": 2.41, + "learning_rate": 0.00019986067378159223, + "loss": 7.7175, + "step": 28890 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28891 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 7.5498, + "step": 28892 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.7506, + "step": 28893 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 5.1533, + "step": 28894 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.4552, + "step": 28895 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.2279, + "step": 28896 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28897 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 6.873, + "step": 28898 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 3.5384, + "step": 28899 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.5687, + "step": 28900 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28901 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.4684, + "step": 28902 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28903 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 9.969, + "step": 28904 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.1396, + "step": 28905 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.0946, + "step": 28906 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28907 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.8942, + "step": 28908 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28909 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 5.8925, + "step": 28910 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28911 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.4676, + "step": 28912 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.9773, + "step": 28913 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.19, + "step": 28914 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.8498, + "step": 28915 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28916 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.15, + "step": 28917 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 8.3889, + "step": 28918 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28919 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.4038, + "step": 28920 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.3047, + "step": 28921 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28922 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.589, + "step": 28923 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 6.5971, + "step": 28924 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28925 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.448, + "step": 28926 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28927 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28928 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28929 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.9452, + "step": 28930 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.7046, + "step": 28931 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28932 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 6.7636, + "step": 28933 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.7301, + "step": 28934 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28935 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28936 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28937 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28938 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28939 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28940 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28941 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28942 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.7818, + "step": 28943 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.3799, + "step": 28944 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28945 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28946 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.8862, + "step": 28947 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.0313, + "step": 28948 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 7.7965, + "step": 28949 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.7719, + "step": 28950 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28951 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.4717, + "step": 28952 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.0769, + "step": 28953 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28954 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.882, + "step": 28955 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 3.4072, + "step": 28956 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28957 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28958 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28959 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28960 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28961 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28962 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28963 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28964 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.8944, + "step": 28965 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.2659, + "step": 28966 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28967 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.9469, + "step": 28968 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28969 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28970 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28971 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28972 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.3343, + "step": 28973 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28974 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28975 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28976 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28977 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28978 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28979 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 3.1518, + "step": 28980 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.2604, + "step": 28981 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.956, + "step": 28982 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.7054, + "step": 28983 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 5.0203, + "step": 28984 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.2117, + "step": 28985 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 1.8167, + "step": 28986 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28987 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28988 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.2041, + "step": 28989 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28990 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.1431, + "step": 28991 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 2.0654, + "step": 28992 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28993 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28994 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 3.7561, + "step": 28995 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.8555, + "step": 28996 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.5727, + "step": 28997 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.3636, + "step": 28998 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 28999 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 6.5729, + "step": 29000 + }, + { + "epoch": 2.42, + "eval_loss": NaN, + "eval_runtime": 40.173, + "eval_samples_per_second": 62.231, + "eval_steps_per_second": 7.791, + "step": 29000 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 11.0236, + "step": 29001 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 5.0541, + "step": 29002 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29003 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29004 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 5.1633, + "step": 29005 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 4.8497, + "step": 29006 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29007 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 3.7278, + "step": 29008 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29009 + }, + { + "epoch": 2.42, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29010 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29011 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.0951, + "step": 29012 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 4.2334, + "step": 29013 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.4207, + "step": 29014 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29015 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29016 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 4.2475, + "step": 29017 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29018 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 5.1178, + "step": 29019 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29020 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.1945, + "step": 29021 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29022 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.0532, + "step": 29023 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29024 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 2.6585, + "step": 29025 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29026 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29027 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 10.7156, + "step": 29028 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29029 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 4.7734, + "step": 29030 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29031 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 5.8895, + "step": 29032 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 7.2498, + "step": 29033 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.1231, + "step": 29034 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29035 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29036 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 7.6487, + "step": 29037 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29038 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 7.6223, + "step": 29039 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.5669, + "step": 29040 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 2.2339, + "step": 29041 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.2358, + "step": 29042 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29043 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29044 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29045 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29046 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29047 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29048 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.1161, + "step": 29049 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.7598, + "step": 29050 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.3643, + "step": 29051 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.4408, + "step": 29052 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29053 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29054 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 2.6516, + "step": 29055 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.821, + "step": 29056 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.8386, + "step": 29057 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.1422, + "step": 29058 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29059 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 4.58, + "step": 29060 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29061 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.9636, + "step": 29062 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29063 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29064 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.929, + "step": 29065 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 6.0797, + "step": 29066 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29067 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29068 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 10.7778, + "step": 29069 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29070 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.4513, + "step": 29071 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29072 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29073 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.8041, + "step": 29074 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 2.0177, + "step": 29075 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29076 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 4.3118, + "step": 29077 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 2.8739, + "step": 29078 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29079 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29080 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29081 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.9044, + "step": 29082 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 7.849, + "step": 29083 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 6.7501, + "step": 29084 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29085 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.8235, + "step": 29086 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 2.2032, + "step": 29087 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.1027, + "step": 29088 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29089 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29090 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29091 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.8305, + "step": 29092 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29093 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29094 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29095 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29096 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.3898, + "step": 29097 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29098 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29099 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 4.1179, + "step": 29100 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 6.561, + "step": 29101 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.1073, + "step": 29102 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29103 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.8215, + "step": 29104 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29105 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29106 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.9678, + "step": 29107 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.7428, + "step": 29108 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.8515, + "step": 29109 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 7.1195, + "step": 29110 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.0215, + "step": 29111 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 10.3258, + "step": 29112 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 7.6631, + "step": 29113 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29114 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 5.1187, + "step": 29115 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.9506, + "step": 29116 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.8162, + "step": 29117 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29118 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29119 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.9961, + "step": 29120 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29121 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29122 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29123 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.8205, + "step": 29124 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.1244, + "step": 29125 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 8.4323, + "step": 29126 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 1.05, + "step": 29127 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29128 + }, + { + "epoch": 2.43, + "learning_rate": 0.00019986067378159223, + "loss": 3.8482, + "step": 29129 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.9492, + "step": 29130 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29131 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29132 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29133 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.9813, + "step": 29134 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.2897, + "step": 29135 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29136 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29137 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29138 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 7.4281, + "step": 29139 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29140 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.9973, + "step": 29141 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29142 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29143 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29144 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.9005, + "step": 29145 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.8442, + "step": 29146 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29147 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29148 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.732, + "step": 29149 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 6.6885, + "step": 29150 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29151 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 2.2879, + "step": 29152 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29153 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.9847, + "step": 29154 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29155 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29156 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 7.2515, + "step": 29157 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.2711, + "step": 29158 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29159 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29160 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 9.2922, + "step": 29161 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29162 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.7343, + "step": 29163 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 2.313, + "step": 29164 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29165 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29166 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29167 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.9188, + "step": 29168 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 5.7464, + "step": 29169 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.9778, + "step": 29170 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29171 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29172 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.9394, + "step": 29173 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.344, + "step": 29174 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29175 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29176 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29177 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29178 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.6862, + "step": 29179 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.2617, + "step": 29180 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.005, + "step": 29181 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 2.1162, + "step": 29182 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29183 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29184 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29185 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.5263, + "step": 29186 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29187 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.0227, + "step": 29188 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29189 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 5.4958, + "step": 29190 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.3436, + "step": 29191 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.3569, + "step": 29192 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29193 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29194 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29195 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29196 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 5.4342, + "step": 29197 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 2.0449, + "step": 29198 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.7885, + "step": 29199 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 2.2243, + "step": 29200 + }, + { + "epoch": 2.44, + "eval_loss": NaN, + "eval_runtime": 40.1189, + "eval_samples_per_second": 62.315, + "eval_steps_per_second": 7.802, + "step": 29200 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 5.5111, + "step": 29201 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29202 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29203 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.5545, + "step": 29204 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29205 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29206 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 13.9708, + "step": 29207 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 9.1881, + "step": 29208 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29209 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 8.4563, + "step": 29210 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29211 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.1818, + "step": 29212 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29213 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29214 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.6233, + "step": 29215 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 2.1954, + "step": 29216 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 9.5061, + "step": 29217 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.5516, + "step": 29218 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.1637, + "step": 29219 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 7.743, + "step": 29220 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 7.4482, + "step": 29221 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 7.34, + "step": 29222 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29223 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.8461, + "step": 29224 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29225 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29226 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.3367, + "step": 29227 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 5.3815, + "step": 29228 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.0696, + "step": 29229 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29230 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 4.2418, + "step": 29231 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 3.7301, + "step": 29232 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29233 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.8829, + "step": 29234 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29235 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.9655, + "step": 29236 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 8.3307, + "step": 29237 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.1988, + "step": 29238 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29239 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29240 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 1.1346, + "step": 29241 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 5.1872, + "step": 29242 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29243 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29244 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 2.7714, + "step": 29245 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29246 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29247 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29248 + }, + { + "epoch": 2.44, + "learning_rate": 0.00019986067378159223, + "loss": 7.5357, + "step": 29249 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29250 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29251 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019986067378159223, + "loss": 0.0, + "step": 29252 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.6553, + "step": 29253 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 5.3917, + "step": 29254 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29255 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 6.9367, + "step": 29256 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 6.7252, + "step": 29257 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.9135, + "step": 29258 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29259 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 4.0095, + "step": 29260 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29261 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29262 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29263 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.8207, + "step": 29264 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 7.4688, + "step": 29265 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29266 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.0525, + "step": 29267 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29268 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.2993, + "step": 29269 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29270 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 8.2408, + "step": 29271 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29272 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29273 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.0458, + "step": 29274 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 14.2342, + "step": 29275 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.9257, + "step": 29276 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 5.2459, + "step": 29277 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.9188, + "step": 29278 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 8.8095, + "step": 29279 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 4.6772, + "step": 29280 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29281 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29282 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29283 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 4.0821, + "step": 29284 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29285 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 7.133, + "step": 29286 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.9629, + "step": 29287 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29288 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29289 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29290 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 6.7928, + "step": 29291 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 9.2787, + "step": 29292 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.1942, + "step": 29293 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.884, + "step": 29294 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.9935, + "step": 29295 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.4749, + "step": 29296 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.3771, + "step": 29297 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29298 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 5.0291, + "step": 29299 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29300 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29301 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 6.9428, + "step": 29302 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 5.0664, + "step": 29303 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.8141, + "step": 29304 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.7827, + "step": 29305 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29306 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29307 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29308 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.4382, + "step": 29309 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.9661, + "step": 29310 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.0412, + "step": 29311 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29312 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29313 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.0753, + "step": 29314 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29315 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29316 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.0062, + "step": 29317 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29318 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.4804, + "step": 29319 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29320 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29321 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.1344, + "step": 29322 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.4134, + "step": 29323 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29324 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29325 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.9249, + "step": 29326 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 4.7933, + "step": 29327 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.334, + "step": 29328 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.0707, + "step": 29329 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29330 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29331 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.8452, + "step": 29332 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.9854, + "step": 29333 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.519, + "step": 29334 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 7.6544, + "step": 29335 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 15.5478, + "step": 29336 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29337 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.8944, + "step": 29338 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.9427, + "step": 29339 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 7.7845, + "step": 29340 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29341 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29342 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29343 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.9768, + "step": 29344 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 12.6154, + "step": 29345 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29346 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.9207, + "step": 29347 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29348 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29349 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29350 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.2166, + "step": 29351 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.0344, + "step": 29352 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29353 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29354 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29355 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.7719, + "step": 29356 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29357 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.233, + "step": 29358 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 8.6323, + "step": 29359 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29360 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29361 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 5.0365, + "step": 29362 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 1.8772, + "step": 29363 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 2.0865, + "step": 29364 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29365 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 3.4938, + "step": 29366 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.9853, + "step": 29367 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29368 + }, + { + "epoch": 2.45, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29369 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.9075, + "step": 29370 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.0057, + "step": 29371 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29372 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29373 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.6397, + "step": 29374 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29375 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29376 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29377 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 9.8999, + "step": 29378 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29379 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.4075, + "step": 29380 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29381 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29382 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29383 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29384 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29385 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29386 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29387 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 7.7485, + "step": 29388 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.9825, + "step": 29389 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.4296, + "step": 29390 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29391 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29392 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29393 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29394 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29395 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 6.4955, + "step": 29396 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29397 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29398 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.4174, + "step": 29399 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.6708, + "step": 29400 + }, + { + "epoch": 2.46, + "eval_loss": NaN, + "eval_runtime": 40.1565, + "eval_samples_per_second": 62.256, + "eval_steps_per_second": 7.795, + "step": 29400 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.2918, + "step": 29401 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29402 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29403 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29404 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29405 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.8248, + "step": 29406 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.0396, + "step": 29407 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 2.1304, + "step": 29408 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.2453, + "step": 29409 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29410 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29411 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 2.0463, + "step": 29412 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.0214, + "step": 29413 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29414 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.6605, + "step": 29415 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.2548, + "step": 29416 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29417 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.2345, + "step": 29418 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 13.3451, + "step": 29419 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.1614, + "step": 29420 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 10.2439, + "step": 29421 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 7.1416, + "step": 29422 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29423 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.4647, + "step": 29424 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29425 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 2.0212, + "step": 29426 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29427 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.6789, + "step": 29428 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29429 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.9633, + "step": 29430 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29431 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.0555, + "step": 29432 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29433 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 7.3782, + "step": 29434 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29435 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 9.0157, + "step": 29436 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29437 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29438 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29439 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29440 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.0631, + "step": 29441 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 7.1459, + "step": 29442 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29443 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29444 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29445 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 12.1726, + "step": 29446 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.3421, + "step": 29447 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29448 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29449 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.0883, + "step": 29450 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.9854, + "step": 29451 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29452 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.3384, + "step": 29453 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29454 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29455 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 2.8972, + "step": 29456 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29457 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29458 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29459 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.079, + "step": 29460 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.1297, + "step": 29461 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.7505, + "step": 29462 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29463 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 8.5316, + "step": 29464 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.9011, + "step": 29465 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29466 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.2324, + "step": 29467 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 2.0402, + "step": 29468 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29469 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29470 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29471 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29472 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29473 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 9.0899, + "step": 29474 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.3267, + "step": 29475 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.7483, + "step": 29476 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29477 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 6.2135, + "step": 29478 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 12.1701, + "step": 29479 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29480 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 1.9068, + "step": 29481 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29482 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 4.9492, + "step": 29483 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29484 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29485 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 6.4899, + "step": 29486 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 0.9621, + "step": 29487 + }, + { + "epoch": 2.46, + "learning_rate": 0.00019985510073285592, + "loss": 3.165, + "step": 29488 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.0015, + "step": 29489 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.097, + "step": 29490 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 3.6798, + "step": 29491 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29492 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 5.9367, + "step": 29493 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29494 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29495 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29496 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29497 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29498 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 10.6969, + "step": 29499 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 8.4132, + "step": 29500 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29501 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.8716, + "step": 29502 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29503 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.4883, + "step": 29504 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29505 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29506 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.6648, + "step": 29507 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29508 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 8.1055, + "step": 29509 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.2594, + "step": 29510 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 4.4673, + "step": 29511 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 2.0238, + "step": 29512 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.9016, + "step": 29513 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 2.1047, + "step": 29514 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29515 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29516 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29517 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 6.5199, + "step": 29518 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 3.1527, + "step": 29519 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29520 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29521 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 8.0659, + "step": 29522 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29523 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 5.0841, + "step": 29524 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29525 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29526 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29527 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29528 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29529 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29530 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29531 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.023, + "step": 29532 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.9695, + "step": 29533 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 2.0884, + "step": 29534 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.9106, + "step": 29535 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29536 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.0265, + "step": 29537 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29538 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.2303, + "step": 29539 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 5.4628, + "step": 29540 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29541 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 8.749, + "step": 29542 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29543 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29544 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29545 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29546 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.3924, + "step": 29547 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29548 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.7607, + "step": 29549 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 6.3035, + "step": 29550 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 6.9052, + "step": 29551 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 8.5475, + "step": 29552 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29553 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29554 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.9898, + "step": 29555 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 6.4863, + "step": 29556 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 4.2479, + "step": 29557 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29558 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.962, + "step": 29559 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29560 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.0479, + "step": 29561 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 2.7001, + "step": 29562 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29563 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29564 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29565 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29566 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29567 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 5.8166, + "step": 29568 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29569 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29570 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29571 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.6134, + "step": 29572 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29573 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29574 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29575 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.7757, + "step": 29576 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.7054, + "step": 29577 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29578 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 8.5197, + "step": 29579 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29580 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 2.6687, + "step": 29581 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29582 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29583 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29584 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 1.9766, + "step": 29585 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29586 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 6.499, + "step": 29587 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.2887, + "step": 29588 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 3.7837, + "step": 29589 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29590 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 3.1354, + "step": 29591 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29592 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29593 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29594 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 8.9036, + "step": 29595 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 3.0863, + "step": 29596 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29597 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.0466, + "step": 29598 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29599 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29600 + }, + { + "epoch": 2.47, + "eval_loss": NaN, + "eval_runtime": 40.1245, + "eval_samples_per_second": 62.306, + "eval_steps_per_second": 7.801, + "step": 29600 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29601 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.7671, + "step": 29602 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 6.5055, + "step": 29603 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29604 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29605 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 7.0832, + "step": 29606 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 12.5936, + "step": 29607 + }, + { + "epoch": 2.47, + "learning_rate": 0.00019985510073285592, + "loss": 2.7733, + "step": 29608 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.9851, + "step": 29609 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29610 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 2.1087, + "step": 29611 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 2.1686, + "step": 29612 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.8663, + "step": 29613 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29614 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29615 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 4.0877, + "step": 29616 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.6187, + "step": 29617 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.6847, + "step": 29618 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 2.3972, + "step": 29619 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29620 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29621 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29622 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29623 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 8.1489, + "step": 29624 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.8779, + "step": 29625 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.1792, + "step": 29626 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29627 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29628 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.305, + "step": 29629 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 4.1282, + "step": 29630 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 4.3579, + "step": 29631 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29632 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29633 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29634 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.9374, + "step": 29635 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29636 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29637 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29638 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 7.8886, + "step": 29639 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.5435, + "step": 29640 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29641 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29642 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29643 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 3.1493, + "step": 29644 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29645 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29646 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29647 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29648 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29649 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29650 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29651 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29652 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29653 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29654 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 3.848, + "step": 29655 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29656 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.5825, + "step": 29657 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29658 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.2341, + "step": 29659 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29660 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.0311, + "step": 29661 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29662 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 3.0167, + "step": 29663 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.9825, + "step": 29664 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29665 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.8472, + "step": 29666 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 3.3218, + "step": 29667 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29668 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29669 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29670 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.7843, + "step": 29671 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.932, + "step": 29672 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.4285, + "step": 29673 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29674 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29675 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.9709, + "step": 29676 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.9587, + "step": 29677 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.9654, + "step": 29678 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.7956, + "step": 29679 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29680 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 2.045, + "step": 29681 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29682 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29683 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29684 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 2.3158, + "step": 29685 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.8139, + "step": 29686 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.6736, + "step": 29687 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 5.9807, + "step": 29688 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29689 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.9413, + "step": 29690 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.7748, + "step": 29691 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29692 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 6.2868, + "step": 29693 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29694 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29695 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29696 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29697 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 4.5499, + "step": 29698 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29699 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 5.5981, + "step": 29700 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 9.2865, + "step": 29701 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29702 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29703 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 3.9953, + "step": 29704 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29705 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29706 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29707 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.945, + "step": 29708 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29709 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 6.0055, + "step": 29710 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 2.1569, + "step": 29711 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.8877, + "step": 29712 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.8198, + "step": 29713 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29714 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29715 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29716 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 8.362, + "step": 29717 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29718 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29719 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29720 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29721 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29722 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29723 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 4.1199, + "step": 29724 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 5.7183, + "step": 29725 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29726 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 1.778, + "step": 29727 + }, + { + "epoch": 2.48, + "learning_rate": 0.00019985510073285592, + "loss": 0.7897, + "step": 29728 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29729 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29730 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29731 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.6962, + "step": 29732 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.6242, + "step": 29733 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29734 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.8839, + "step": 29735 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29736 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.5175, + "step": 29737 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 6.1334, + "step": 29738 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.0221, + "step": 29739 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29740 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 4.3294, + "step": 29741 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 6.9114, + "step": 29742 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29743 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29744 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29745 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29746 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 10.5549, + "step": 29747 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.8653, + "step": 29748 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 9.6946, + "step": 29749 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29750 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29751 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.9028, + "step": 29752 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29753 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29754 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29755 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.2456, + "step": 29756 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29757 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29758 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29759 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29760 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29761 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.3696, + "step": 29762 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.5194, + "step": 29763 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.6385, + "step": 29764 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.2893, + "step": 29765 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 5.1602, + "step": 29766 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 6.1065, + "step": 29767 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29768 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29769 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29770 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.87, + "step": 29771 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.8599, + "step": 29772 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.1004, + "step": 29773 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.9702, + "step": 29774 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.032, + "step": 29775 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29776 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29777 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 4.4648, + "step": 29778 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 8.0242, + "step": 29779 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.7359, + "step": 29780 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 9.5886, + "step": 29781 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 4.6619, + "step": 29782 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.2738, + "step": 29783 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29784 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29785 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29786 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.9302, + "step": 29787 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.7034, + "step": 29788 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 7.9384, + "step": 29789 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.6667, + "step": 29790 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29791 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29792 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29793 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29794 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29795 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29796 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.1042, + "step": 29797 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.8363, + "step": 29798 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 6.1977, + "step": 29799 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.7415, + "step": 29800 + }, + { + "epoch": 2.49, + "eval_loss": NaN, + "eval_runtime": 40.1555, + "eval_samples_per_second": 62.258, + "eval_steps_per_second": 7.795, + "step": 29800 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29801 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.9911, + "step": 29802 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.0235, + "step": 29803 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29804 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29805 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 4.0817, + "step": 29806 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.142, + "step": 29807 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29808 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 7.7175, + "step": 29809 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.8059, + "step": 29810 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 6.6458, + "step": 29811 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 8.4701, + "step": 29812 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29813 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.0257, + "step": 29814 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 4.8373, + "step": 29815 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29816 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29817 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29818 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29819 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29820 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29821 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 8.1738, + "step": 29822 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29823 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.0072, + "step": 29824 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 7.9195, + "step": 29825 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.5799, + "step": 29826 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29827 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 8.1494, + "step": 29828 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.1613, + "step": 29829 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29830 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29831 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.912, + "step": 29832 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29833 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29834 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.8909, + "step": 29835 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29836 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 7.3639, + "step": 29837 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29838 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.1112, + "step": 29839 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29840 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29841 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.8563, + "step": 29842 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29843 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.321, + "step": 29844 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 3.8956, + "step": 29845 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 1.6167, + "step": 29846 + }, + { + "epoch": 2.49, + "learning_rate": 0.00019985510073285592, + "loss": 2.5663, + "step": 29847 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29848 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.2037, + "step": 29849 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.0484, + "step": 29850 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.0422, + "step": 29851 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29852 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.466, + "step": 29853 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.3334, + "step": 29854 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29855 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 3.9172, + "step": 29856 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29857 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29858 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.0764, + "step": 29859 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.4775, + "step": 29860 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 6.1863, + "step": 29861 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29862 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29863 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29864 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.7531, + "step": 29865 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29866 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29867 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.8545, + "step": 29868 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29869 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.5691, + "step": 29870 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 5.8674, + "step": 29871 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29872 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29873 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29874 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29875 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29876 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.0468, + "step": 29877 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.5133, + "step": 29878 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.2139, + "step": 29879 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29880 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.1086, + "step": 29881 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29882 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.8689, + "step": 29883 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29884 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.5629, + "step": 29885 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.5018, + "step": 29886 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 3.5538, + "step": 29887 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29888 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 10.3316, + "step": 29889 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.784, + "step": 29890 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.7824, + "step": 29891 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.6017, + "step": 29892 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29893 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.9626, + "step": 29894 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29895 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 3.4737, + "step": 29896 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29897 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29898 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29899 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.0716, + "step": 29900 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29901 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 7.311, + "step": 29902 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29903 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.9263, + "step": 29904 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29905 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 3.2488, + "step": 29906 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 7.3106, + "step": 29907 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29908 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 8.2763, + "step": 29909 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29910 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.842, + "step": 29911 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.0906, + "step": 29912 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29913 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29914 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29915 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29916 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29917 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29918 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.0985, + "step": 29919 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.9444, + "step": 29920 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.152, + "step": 29921 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29922 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.0994, + "step": 29923 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.6327, + "step": 29924 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 6.8136, + "step": 29925 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29926 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29927 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29928 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.0905, + "step": 29929 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.7227, + "step": 29930 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.9093, + "step": 29931 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.9396, + "step": 29932 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29933 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29934 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29935 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29936 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29937 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.0241, + "step": 29938 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 7.1363, + "step": 29939 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.8416, + "step": 29940 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.4746, + "step": 29941 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29942 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29943 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29944 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29945 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29946 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.1124, + "step": 29947 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 1.3057, + "step": 29948 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 3.3468, + "step": 29949 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29950 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29951 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 3.8341, + "step": 29952 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29953 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 7.6792, + "step": 29954 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.0121, + "step": 29955 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 6.3196, + "step": 29956 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29957 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29958 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29959 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 3.9049, + "step": 29960 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29961 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29962 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 2.0726, + "step": 29963 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29964 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.9329, + "step": 29965 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29966 + }, + { + "epoch": 2.5, + "learning_rate": 0.00019985510073285592, + "loss": 4.0681, + "step": 29967 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 4.8964, + "step": 29968 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29969 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29970 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29971 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29972 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.4143, + "step": 29973 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29974 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29975 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 7.8862, + "step": 29976 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29977 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29978 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29979 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.7835, + "step": 29980 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 4.304, + "step": 29981 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 1.7045, + "step": 29982 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29983 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 6.2775, + "step": 29984 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.8936, + "step": 29985 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29986 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 7.9746, + "step": 29987 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29988 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29989 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 10.347, + "step": 29990 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29991 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 8.3418, + "step": 29992 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29993 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29994 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29995 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 1.0989, + "step": 29996 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 7.9941, + "step": 29997 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29998 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 29999 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30000 + }, + { + "epoch": 2.51, + "eval_loss": NaN, + "eval_runtime": 40.1319, + "eval_samples_per_second": 62.295, + "eval_steps_per_second": 7.799, + "step": 30000 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 1.1579, + "step": 30001 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.2451, + "step": 30002 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30003 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.7951, + "step": 30004 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30005 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.7662, + "step": 30006 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 5.0288, + "step": 30007 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 6.4901, + "step": 30008 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30009 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30010 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.7891, + "step": 30011 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30012 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30013 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.3561, + "step": 30014 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30015 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30016 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30017 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.2305, + "step": 30018 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30019 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 8.0163, + "step": 30020 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30021 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 4.3549, + "step": 30022 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.0169, + "step": 30023 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 5.7064, + "step": 30024 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.9286, + "step": 30025 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30026 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.4959, + "step": 30027 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 7.278, + "step": 30028 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 10.2974, + "step": 30029 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30030 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.7232, + "step": 30031 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30032 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 4.9405, + "step": 30033 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.8735, + "step": 30034 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30035 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.2824, + "step": 30036 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 11.1577, + "step": 30037 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.6745, + "step": 30038 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 7.7511, + "step": 30039 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 1.6207, + "step": 30040 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.9221, + "step": 30041 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.2438, + "step": 30042 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30043 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.1016, + "step": 30044 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 8.2369, + "step": 30045 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30046 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.0812, + "step": 30047 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 8.6334, + "step": 30048 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30049 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30050 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30051 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 8.7232, + "step": 30052 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30053 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30054 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 8.9167, + "step": 30055 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 7.6997, + "step": 30056 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30057 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 4.4627, + "step": 30058 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 1.0484, + "step": 30059 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.6636, + "step": 30060 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 1.7104, + "step": 30061 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 9.3296, + "step": 30062 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 6.7596, + "step": 30063 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 8.0362, + "step": 30064 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 2.04, + "step": 30065 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30066 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 1.2406, + "step": 30067 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30068 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30069 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30070 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30071 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30072 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30073 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 4.0062, + "step": 30074 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 6.8136, + "step": 30075 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 4.1149, + "step": 30076 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.628, + "step": 30077 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30078 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30079 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30080 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.8777, + "step": 30081 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 3.4573, + "step": 30082 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.9026, + "step": 30083 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30084 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30085 + }, + { + "epoch": 2.51, + "learning_rate": 0.00019985510073285592, + "loss": 7.9917, + "step": 30086 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.993, + "step": 30087 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30088 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.7147, + "step": 30089 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30090 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 4.3493, + "step": 30091 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 9.0249, + "step": 30092 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 6.6308, + "step": 30093 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.1081, + "step": 30094 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 3.3604, + "step": 30095 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30096 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30097 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 2.242, + "step": 30098 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30099 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.8667, + "step": 30100 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.6944, + "step": 30101 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.8076, + "step": 30102 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30103 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.1791, + "step": 30104 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30105 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.9379, + "step": 30106 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30107 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.8977, + "step": 30108 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 7.4968, + "step": 30109 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30110 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30111 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 7.3125, + "step": 30112 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 6.3523, + "step": 30113 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.0063, + "step": 30114 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30115 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30116 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30117 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.2911, + "step": 30118 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 2.2471, + "step": 30119 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30120 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.6813, + "step": 30121 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30122 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.1855, + "step": 30123 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.9618, + "step": 30124 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30125 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30126 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 3.5028, + "step": 30127 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30128 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 2.4217, + "step": 30129 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.9778, + "step": 30130 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30131 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30132 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30133 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.9319, + "step": 30134 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 3.8014, + "step": 30135 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30136 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30137 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30138 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.7336, + "step": 30139 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30140 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 2.1047, + "step": 30141 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30142 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30143 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.9375, + "step": 30144 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30145 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 2.9982, + "step": 30146 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30147 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30148 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.9093, + "step": 30149 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30150 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30151 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30152 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 6.1423, + "step": 30153 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30154 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30155 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.9902, + "step": 30156 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 2.0376, + "step": 30157 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30158 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30159 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30160 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30161 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.9399, + "step": 30162 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30163 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30164 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30165 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.8411, + "step": 30166 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30167 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30168 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.9451, + "step": 30169 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30170 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30171 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.0208, + "step": 30172 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.0961, + "step": 30173 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.8261, + "step": 30174 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30175 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 7.6166, + "step": 30176 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30177 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30178 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 2.2614, + "step": 30179 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30180 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30181 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30182 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.536, + "step": 30183 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.0727, + "step": 30184 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.742, + "step": 30185 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30186 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30187 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30188 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.9317, + "step": 30189 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30190 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 9.819, + "step": 30191 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30192 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 9.1703, + "step": 30193 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.3663, + "step": 30194 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.8254, + "step": 30195 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.7797, + "step": 30196 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 4.0909, + "step": 30197 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.2565, + "step": 30198 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30199 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.2768, + "step": 30200 + }, + { + "epoch": 2.52, + "eval_loss": NaN, + "eval_runtime": 40.1186, + "eval_samples_per_second": 62.315, + "eval_steps_per_second": 7.802, + "step": 30200 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30201 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 7.5326, + "step": 30202 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 5.8096, + "step": 30203 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 8.5786, + "step": 30204 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 1.5445, + "step": 30205 + }, + { + "epoch": 2.52, + "learning_rate": 0.00019985510073285592, + "loss": 4.6497, + "step": 30206 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.0295, + "step": 30207 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30208 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 7.1311, + "step": 30209 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30210 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.4685, + "step": 30211 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30212 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30213 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.3411, + "step": 30214 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30215 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 2.7555, + "step": 30216 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.3497, + "step": 30217 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30218 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.7874, + "step": 30219 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.52, + "step": 30220 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30221 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 8.1959, + "step": 30222 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30223 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 2.3699, + "step": 30224 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30225 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30226 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.067, + "step": 30227 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30228 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.9397, + "step": 30229 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30230 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30231 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30232 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.0832, + "step": 30233 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30234 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30235 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.0846, + "step": 30236 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.2885, + "step": 30237 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30238 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.9476, + "step": 30239 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30240 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30241 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.5863, + "step": 30242 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.4045, + "step": 30243 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.3415, + "step": 30244 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.9514, + "step": 30245 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30246 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30247 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30248 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30249 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30250 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30251 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.7801, + "step": 30252 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30253 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.7302, + "step": 30254 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.2741, + "step": 30255 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30256 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 2.0786, + "step": 30257 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.1857, + "step": 30258 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.8488, + "step": 30259 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30260 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.4958, + "step": 30261 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30262 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.6597, + "step": 30263 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.0989, + "step": 30264 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.7902, + "step": 30265 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 7.7478, + "step": 30266 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30267 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.5215, + "step": 30268 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30269 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 5.4063, + "step": 30270 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.5751, + "step": 30271 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.0551, + "step": 30272 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.8416, + "step": 30273 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.8627, + "step": 30274 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30275 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.792, + "step": 30276 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.9908, + "step": 30277 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.9335, + "step": 30278 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30279 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.1345, + "step": 30280 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.1335, + "step": 30281 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 9.8375, + "step": 30282 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 2.0183, + "step": 30283 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 8.9884, + "step": 30284 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 7.4584, + "step": 30285 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 9.7313, + "step": 30286 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.9495, + "step": 30287 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.0058, + "step": 30288 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30289 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.0524, + "step": 30290 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30291 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.7708, + "step": 30292 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30293 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30294 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 5.5294, + "step": 30295 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 8.6722, + "step": 30296 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30297 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30298 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.1208, + "step": 30299 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.2468, + "step": 30300 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.7738, + "step": 30301 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30302 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30303 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30304 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30305 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.153, + "step": 30306 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30307 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.8377, + "step": 30308 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 5.8504, + "step": 30309 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.1612, + "step": 30310 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 3.9, + "step": 30311 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 4.2914, + "step": 30312 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30313 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30314 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30315 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30316 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30317 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 2.3859, + "step": 30318 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30319 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30320 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30321 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30322 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 1.8742, + "step": 30323 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30324 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.8534, + "step": 30325 + }, + { + "epoch": 2.53, + "learning_rate": 0.00019985510073285592, + "loss": 0.9676, + "step": 30326 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30327 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30328 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30329 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30330 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30331 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30332 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 6.1342, + "step": 30333 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30334 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.1149, + "step": 30335 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30336 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30337 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30338 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.417, + "step": 30339 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.0628, + "step": 30340 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.2578, + "step": 30341 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30342 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.6254, + "step": 30343 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30344 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.0559, + "step": 30345 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30346 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30347 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30348 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.897, + "step": 30349 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.1523, + "step": 30350 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30351 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30352 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30353 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 10.6666, + "step": 30354 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.3725, + "step": 30355 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 8.181, + "step": 30356 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30357 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30358 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.8866, + "step": 30359 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.7503, + "step": 30360 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30361 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30362 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.3781, + "step": 30363 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.1594, + "step": 30364 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.1642, + "step": 30365 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.9817, + "step": 30366 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30367 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.9862, + "step": 30368 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.5917, + "step": 30369 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.2275, + "step": 30370 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.1748, + "step": 30371 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30372 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.9935, + "step": 30373 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.5606, + "step": 30374 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30375 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.111, + "step": 30376 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.9315, + "step": 30377 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 6.7679, + "step": 30378 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30379 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.3103, + "step": 30380 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.2859, + "step": 30381 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30382 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.1242, + "step": 30383 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.5938, + "step": 30384 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 9.3303, + "step": 30385 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.1259, + "step": 30386 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30387 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.76, + "step": 30388 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30389 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.9494, + "step": 30390 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.289, + "step": 30391 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.2281, + "step": 30392 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30393 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30394 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.896, + "step": 30395 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.0863, + "step": 30396 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 6.1785, + "step": 30397 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 5.7855, + "step": 30398 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30399 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.045, + "step": 30400 + }, + { + "epoch": 2.54, + "eval_loss": NaN, + "eval_runtime": 40.1306, + "eval_samples_per_second": 62.297, + "eval_steps_per_second": 7.8, + "step": 30400 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.9893, + "step": 30401 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30402 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.5201, + "step": 30403 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30404 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30405 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30406 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30407 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.6363, + "step": 30408 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30409 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.8527, + "step": 30410 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.864, + "step": 30411 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.1841, + "step": 30412 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30413 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30414 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.8907, + "step": 30415 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30416 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.0713, + "step": 30417 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30418 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30419 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.4453, + "step": 30420 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.8325, + "step": 30421 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.224, + "step": 30422 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30423 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30424 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30425 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.9297, + "step": 30426 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.2134, + "step": 30427 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 7.4244, + "step": 30428 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.5099, + "step": 30429 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 4.5296, + "step": 30430 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.9605, + "step": 30431 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.7215, + "step": 30432 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 6.311, + "step": 30433 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 3.0738, + "step": 30434 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30435 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.7706, + "step": 30436 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 2.1605, + "step": 30437 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30438 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30439 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.7965, + "step": 30440 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30441 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 1.9858, + "step": 30442 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30443 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30444 + }, + { + "epoch": 2.54, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30445 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30446 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.3352, + "step": 30447 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30448 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30449 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.3003, + "step": 30450 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 3.5321, + "step": 30451 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30452 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30453 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 9.0663, + "step": 30454 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 3.5314, + "step": 30455 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30456 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.1387, + "step": 30457 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 7.9134, + "step": 30458 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.0775, + "step": 30459 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 11.2853, + "step": 30460 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 3.7548, + "step": 30461 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30462 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30463 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30464 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.0135, + "step": 30465 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30466 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30467 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 6.3449, + "step": 30468 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.9054, + "step": 30469 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30470 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.8389, + "step": 30471 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 5.6201, + "step": 30472 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.1747, + "step": 30473 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30474 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 5.3945, + "step": 30475 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30476 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30477 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.0394, + "step": 30478 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30479 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30480 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.9649, + "step": 30481 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.2951, + "step": 30482 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.0988, + "step": 30483 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.4738, + "step": 30484 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.3724, + "step": 30485 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 7.9256, + "step": 30486 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.9691, + "step": 30487 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30488 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.5204, + "step": 30489 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30490 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30491 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 7.8156, + "step": 30492 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30493 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30494 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.7485, + "step": 30495 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30496 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30497 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30498 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30499 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.0516, + "step": 30500 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30501 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30502 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 7.803, + "step": 30503 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.5325, + "step": 30504 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 5.1537, + "step": 30505 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30506 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 6.2851, + "step": 30507 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.3665, + "step": 30508 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30509 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.2501, + "step": 30510 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30511 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30512 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.8911, + "step": 30513 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.041, + "step": 30514 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30515 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30516 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30517 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 3.3941, + "step": 30518 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30519 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30520 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.305, + "step": 30521 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30522 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.6491, + "step": 30523 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.2091, + "step": 30524 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.8088, + "step": 30525 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 6.4482, + "step": 30526 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30527 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30528 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 12.7626, + "step": 30529 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30530 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30531 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30532 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.1362, + "step": 30533 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30534 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.1976, + "step": 30535 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30536 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.9447, + "step": 30537 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.8069, + "step": 30538 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 4.8128, + "step": 30539 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30540 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30541 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30542 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30543 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30544 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30545 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30546 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 8.1833, + "step": 30547 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 3.7141, + "step": 30548 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.9875, + "step": 30549 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30550 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.2583, + "step": 30551 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.0983, + "step": 30552 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30553 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.0181, + "step": 30554 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.0322, + "step": 30555 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30556 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.8319, + "step": 30557 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30558 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 1.8082, + "step": 30559 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 2.9236, + "step": 30560 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 3.658, + "step": 30561 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30562 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.7727, + "step": 30563 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30564 + }, + { + "epoch": 2.55, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30565 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30566 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30567 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.5334, + "step": 30568 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 6.1986, + "step": 30569 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30570 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30571 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 3.8526, + "step": 30572 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30573 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30574 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.8632, + "step": 30575 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30576 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 3.4387, + "step": 30577 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30578 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 6.8052, + "step": 30579 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30580 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 8.5719, + "step": 30581 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.1998, + "step": 30582 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30583 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 4.2404, + "step": 30584 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.0045, + "step": 30585 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.081, + "step": 30586 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 3.9881, + "step": 30587 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30588 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 5.5164, + "step": 30589 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.0511, + "step": 30590 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30591 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.0887, + "step": 30592 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.4098, + "step": 30593 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30594 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.126, + "step": 30595 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30596 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 9.097, + "step": 30597 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 3.6338, + "step": 30598 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30599 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 9.6403, + "step": 30600 + }, + { + "epoch": 2.56, + "eval_loss": NaN, + "eval_runtime": 40.0551, + "eval_samples_per_second": 62.414, + "eval_steps_per_second": 7.814, + "step": 30600 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.6306, + "step": 30601 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30602 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.9915, + "step": 30603 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.6675, + "step": 30604 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30605 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30606 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30607 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30608 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30609 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 4.2654, + "step": 30610 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.7566, + "step": 30611 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30612 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30613 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30614 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30615 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 10.8047, + "step": 30616 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.9689, + "step": 30617 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 7.561, + "step": 30618 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30619 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30620 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30621 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 3.8279, + "step": 30622 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30623 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30624 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 7.3324, + "step": 30625 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 4.0683, + "step": 30626 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 4.5025, + "step": 30627 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30628 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30629 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30630 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 6.0836, + "step": 30631 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30632 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.7834, + "step": 30633 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30634 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.2236, + "step": 30635 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 7.0858, + "step": 30636 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 7.9139, + "step": 30637 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 4.0536, + "step": 30638 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30639 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 7.9606, + "step": 30640 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.3956, + "step": 30641 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30642 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.8958, + "step": 30643 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30644 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 8.4912, + "step": 30645 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30646 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30647 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 4.0461, + "step": 30648 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 6.98, + "step": 30649 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 8.7128, + "step": 30650 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.1705, + "step": 30651 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30652 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30653 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.1826, + "step": 30654 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 8.8405, + "step": 30655 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30656 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30657 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30658 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 5.6369, + "step": 30659 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30660 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.8695, + "step": 30661 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 6.411, + "step": 30662 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 3.3166, + "step": 30663 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30664 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 3.895, + "step": 30665 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30666 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30667 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30668 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30669 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30670 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 4.7676, + "step": 30671 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30672 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 7.7352, + "step": 30673 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 2.3232, + "step": 30674 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 10.8151, + "step": 30675 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 5.3945, + "step": 30676 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30677 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30678 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30679 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 7.964, + "step": 30680 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 6.1166, + "step": 30681 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30682 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.6937, + "step": 30683 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30684 + }, + { + "epoch": 2.56, + "learning_rate": 0.00019985510073285592, + "loss": 1.6936, + "step": 30685 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.6584, + "step": 30686 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.0803, + "step": 30687 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.7887, + "step": 30688 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.8891, + "step": 30689 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.247, + "step": 30690 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30691 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 8.0147, + "step": 30692 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30693 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30694 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30695 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30696 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30697 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30698 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.9576, + "step": 30699 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.8013, + "step": 30700 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.0343, + "step": 30701 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 4.037, + "step": 30702 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 8.4227, + "step": 30703 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.0797, + "step": 30704 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30705 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.7191, + "step": 30706 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.1172, + "step": 30707 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30708 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.9634, + "step": 30709 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30710 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.5855, + "step": 30711 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 4.6028, + "step": 30712 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30713 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.7553, + "step": 30714 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30715 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.0836, + "step": 30716 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30717 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30718 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30719 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30720 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30721 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30722 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30723 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30724 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.3915, + "step": 30725 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 7.6767, + "step": 30726 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.7196, + "step": 30727 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.9785, + "step": 30728 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.2459, + "step": 30729 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 7.503, + "step": 30730 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.0215, + "step": 30731 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.0529, + "step": 30732 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 6.0597, + "step": 30733 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30734 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30735 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 6.2396, + "step": 30736 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30737 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 6.3159, + "step": 30738 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 6.2497, + "step": 30739 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.0313, + "step": 30740 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.7116, + "step": 30741 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 4.303, + "step": 30742 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.2978, + "step": 30743 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30744 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.1558, + "step": 30745 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30746 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30747 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.5602, + "step": 30748 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.7978, + "step": 30749 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30750 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.9816, + "step": 30751 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.0752, + "step": 30752 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30753 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.8592, + "step": 30754 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 10.7533, + "step": 30755 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 7.0607, + "step": 30756 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 7.3012, + "step": 30757 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30758 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.0375, + "step": 30759 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 8.5196, + "step": 30760 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30761 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30762 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.1834, + "step": 30763 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 4.6767, + "step": 30764 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30765 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 7.1603, + "step": 30766 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.9708, + "step": 30767 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 9.906, + "step": 30768 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.9317, + "step": 30769 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.8216, + "step": 30770 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 5.6489, + "step": 30771 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30772 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.3761, + "step": 30773 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30774 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 5.8419, + "step": 30775 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.8157, + "step": 30776 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.1741, + "step": 30777 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.1172, + "step": 30778 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.5575, + "step": 30779 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30780 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 8.074, + "step": 30781 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30782 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.0445, + "step": 30783 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 10.0982, + "step": 30784 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.219, + "step": 30785 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.1024, + "step": 30786 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 5.1664, + "step": 30787 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.0139, + "step": 30788 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30789 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30790 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30791 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30792 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 9.0141, + "step": 30793 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.2192, + "step": 30794 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30795 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30796 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.4457, + "step": 30797 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 2.1527, + "step": 30798 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30799 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 8.6218, + "step": 30800 + }, + { + "epoch": 2.57, + "eval_loss": NaN, + "eval_runtime": 40.0781, + "eval_samples_per_second": 62.378, + "eval_steps_per_second": 7.81, + "step": 30800 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 1.5076, + "step": 30801 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 0.956, + "step": 30802 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 3.1627, + "step": 30803 + }, + { + "epoch": 2.57, + "learning_rate": 0.00019985510073285592, + "loss": 5.9105, + "step": 30804 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30805 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 7.6395, + "step": 30806 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 7.0987, + "step": 30807 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.0152, + "step": 30808 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 8.9599, + "step": 30809 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.887, + "step": 30810 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 9.6745, + "step": 30811 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30812 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30813 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30814 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 9.7365, + "step": 30815 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30816 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 5.024, + "step": 30817 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.9701, + "step": 30818 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30819 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.4336, + "step": 30820 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30821 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.1569, + "step": 30822 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.0013, + "step": 30823 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30824 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30825 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.7844, + "step": 30826 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.2882, + "step": 30827 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30828 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30829 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.0644, + "step": 30830 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30831 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30832 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30833 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.4444, + "step": 30834 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 8.1839, + "step": 30835 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30836 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.1712, + "step": 30837 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 3.8163, + "step": 30838 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.1013, + "step": 30839 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.06, + "step": 30840 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30841 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30842 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 9.24, + "step": 30843 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30844 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.9308, + "step": 30845 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.284, + "step": 30846 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 8.8777, + "step": 30847 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30848 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.5124, + "step": 30849 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.0794, + "step": 30850 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30851 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30852 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30853 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 3.3814, + "step": 30854 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 3.8529, + "step": 30855 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30856 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 8.5381, + "step": 30857 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 8.647, + "step": 30858 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 3.3724, + "step": 30859 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30860 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30861 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30862 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.9416, + "step": 30863 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30864 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.6945, + "step": 30865 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.4152, + "step": 30866 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.7135, + "step": 30867 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.962, + "step": 30868 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30869 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.0525, + "step": 30870 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.9997, + "step": 30871 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30872 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30873 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30874 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30875 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 3.0477, + "step": 30876 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30877 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30878 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30879 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30880 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.1677, + "step": 30881 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 7.9435, + "step": 30882 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30883 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30884 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30885 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.9173, + "step": 30886 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30887 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 5.4561, + "step": 30888 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30889 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30890 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30891 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30892 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.1842, + "step": 30893 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 8.5962, + "step": 30894 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.0828, + "step": 30895 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 7.9634, + "step": 30896 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30897 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 9.3358, + "step": 30898 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30899 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 7.3209, + "step": 30900 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30901 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30902 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.0483, + "step": 30903 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30904 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.9476, + "step": 30905 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30906 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30907 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.1127, + "step": 30908 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30909 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.8886, + "step": 30910 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30911 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.8335, + "step": 30912 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 1.9216, + "step": 30913 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.3207, + "step": 30914 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30915 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30916 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30917 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.9893, + "step": 30918 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 3.5237, + "step": 30919 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 6.5907, + "step": 30920 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 5.2217, + "step": 30921 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30922 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 2.8293, + "step": 30923 + }, + { + "epoch": 2.58, + "learning_rate": 0.00019985510073285592, + "loss": 4.5642, + "step": 30924 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.3719, + "step": 30925 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.5777, + "step": 30926 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30927 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.0181, + "step": 30928 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.0385, + "step": 30929 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 6.7914, + "step": 30930 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.059, + "step": 30931 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.6414, + "step": 30932 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.8554, + "step": 30933 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.1545, + "step": 30934 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.6717, + "step": 30935 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.8718, + "step": 30936 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.4073, + "step": 30937 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30938 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.0897, + "step": 30939 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 8.3866, + "step": 30940 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30941 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.8685, + "step": 30942 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30943 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.4064, + "step": 30944 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 5.7526, + "step": 30945 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.1813, + "step": 30946 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30947 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 9.229, + "step": 30948 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30949 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.2752, + "step": 30950 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.2417, + "step": 30951 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 7.4413, + "step": 30952 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30953 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30954 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.0611, + "step": 30955 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.9893, + "step": 30956 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.9169, + "step": 30957 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 5.2656, + "step": 30958 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.7707, + "step": 30959 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30960 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30961 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30962 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 5.342, + "step": 30963 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30964 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30965 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30966 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.1124, + "step": 30967 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30968 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.0153, + "step": 30969 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.0678, + "step": 30970 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.0558, + "step": 30971 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30972 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30973 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30974 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30975 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.0123, + "step": 30976 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30977 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.8551, + "step": 30978 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 5.0672, + "step": 30979 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30980 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.807, + "step": 30981 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.977, + "step": 30982 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30983 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30984 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.1681, + "step": 30985 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30986 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30987 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 5.6876, + "step": 30988 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.6244, + "step": 30989 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30990 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 9.5808, + "step": 30991 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.5775, + "step": 30992 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.2085, + "step": 30993 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30994 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.1385, + "step": 30995 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.9967, + "step": 30996 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30997 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 30998 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.724, + "step": 30999 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31000 + }, + { + "epoch": 2.59, + "eval_loss": NaN, + "eval_runtime": 40.0672, + "eval_samples_per_second": 62.395, + "eval_steps_per_second": 7.812, + "step": 31000 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.0727, + "step": 31001 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31002 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31003 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31004 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.1522, + "step": 31005 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.2706, + "step": 31006 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31007 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.1845, + "step": 31008 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31009 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.8492, + "step": 31010 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31011 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.5253, + "step": 31012 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31013 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31014 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.8643, + "step": 31015 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31016 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.7135, + "step": 31017 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31018 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 7.5265, + "step": 31019 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 1.9213, + "step": 31020 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31021 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31022 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31023 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.0038, + "step": 31024 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31025 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.9041, + "step": 31026 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31027 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 6.6544, + "step": 31028 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31029 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.7604, + "step": 31030 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.8282, + "step": 31031 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31032 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31033 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31034 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 2.0244, + "step": 31035 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31036 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 6.7704, + "step": 31037 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 8.4245, + "step": 31038 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.3258, + "step": 31039 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 3.7618, + "step": 31040 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31041 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 0.9224, + "step": 31042 + }, + { + "epoch": 2.59, + "learning_rate": 0.00019985510073285592, + "loss": 4.6108, + "step": 31043 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31044 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.5861, + "step": 31045 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31046 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 2.5292, + "step": 31047 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 2.717, + "step": 31048 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.8129, + "step": 31049 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31050 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31051 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31052 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31053 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31054 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.8212, + "step": 31055 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 8.9326, + "step": 31056 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31057 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31058 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.9888, + "step": 31059 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 2.1536, + "step": 31060 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 9.2301, + "step": 31061 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31062 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.4141, + "step": 31063 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31064 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.0028, + "step": 31065 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31066 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.972, + "step": 31067 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.0008, + "step": 31068 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31069 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31070 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 2.4494, + "step": 31071 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.776, + "step": 31072 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.5379, + "step": 31073 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.8367, + "step": 31074 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31075 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.9785, + "step": 31076 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.778, + "step": 31077 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31078 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.9202, + "step": 31079 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31080 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.0914, + "step": 31081 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.899, + "step": 31082 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.0399, + "step": 31083 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31084 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 9.3983, + "step": 31085 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31086 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31087 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.9168, + "step": 31088 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 4.8565, + "step": 31089 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.8801, + "step": 31090 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31091 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 4.4654, + "step": 31092 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31093 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.9887, + "step": 31094 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 9.908, + "step": 31095 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31096 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.1619, + "step": 31097 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31098 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 8.6917, + "step": 31099 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31100 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31101 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 5.7281, + "step": 31102 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.954, + "step": 31103 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.0983, + "step": 31104 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.3201, + "step": 31105 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 8.0561, + "step": 31106 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31107 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31108 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 10.7632, + "step": 31109 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31110 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31111 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 10.0495, + "step": 31112 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.1149, + "step": 31113 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 8.0511, + "step": 31114 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31115 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.8513, + "step": 31116 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31117 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.9838, + "step": 31118 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.7794, + "step": 31119 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.1768, + "step": 31120 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31121 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 2.127, + "step": 31122 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31123 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.7779, + "step": 31124 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.1246, + "step": 31125 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.4035, + "step": 31126 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.3334, + "step": 31127 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 9.3164, + "step": 31128 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.6023, + "step": 31129 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31130 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31131 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.7485, + "step": 31132 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.9823, + "step": 31133 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31134 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 7.3738, + "step": 31135 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.1336, + "step": 31136 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31137 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.8379, + "step": 31138 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31139 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 8.5834, + "step": 31140 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 9.0313, + "step": 31141 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.9997, + "step": 31142 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31143 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31144 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 10.6267, + "step": 31145 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31146 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.0179, + "step": 31147 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.701, + "step": 31148 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31149 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31150 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31151 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 6.0471, + "step": 31152 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.6481, + "step": 31153 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31154 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.9011, + "step": 31155 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 5.3799, + "step": 31156 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 3.4481, + "step": 31157 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 1.992, + "step": 31158 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.8748, + "step": 31159 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 5.8023, + "step": 31160 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31161 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31162 + }, + { + "epoch": 2.6, + "learning_rate": 0.00019985510073285592, + "loss": 4.2582, + "step": 31163 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.0202, + "step": 31164 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.5841, + "step": 31165 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 5.4461, + "step": 31166 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 6.7789, + "step": 31167 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 5.8136, + "step": 31168 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31169 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31170 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.7608, + "step": 31171 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.3805, + "step": 31172 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31173 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.3257, + "step": 31174 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.8558, + "step": 31175 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31176 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31177 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.6144, + "step": 31178 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 10.0501, + "step": 31179 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 8.8083, + "step": 31180 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.4709, + "step": 31181 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31182 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31183 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 5.4092, + "step": 31184 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.133, + "step": 31185 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31186 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.7225, + "step": 31187 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31188 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.838, + "step": 31189 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.2704, + "step": 31190 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.1889, + "step": 31191 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.0138, + "step": 31192 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31193 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31194 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.0702, + "step": 31195 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31196 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31197 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.153, + "step": 31198 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 10.3833, + "step": 31199 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.8158, + "step": 31200 + }, + { + "epoch": 2.61, + "eval_loss": NaN, + "eval_runtime": 40.0588, + "eval_samples_per_second": 62.408, + "eval_steps_per_second": 7.814, + "step": 31200 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31201 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 8.2681, + "step": 31202 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.9033, + "step": 31203 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.942, + "step": 31204 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31205 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.5521, + "step": 31206 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.3574, + "step": 31207 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 8.6709, + "step": 31208 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31209 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.6525, + "step": 31210 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.0476, + "step": 31211 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31212 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.9514, + "step": 31213 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31214 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31215 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.2778, + "step": 31216 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.6369, + "step": 31217 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.7095, + "step": 31218 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 5.8964, + "step": 31219 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.0054, + "step": 31220 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.6174, + "step": 31221 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 5.3798, + "step": 31222 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.7004, + "step": 31223 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.0823, + "step": 31224 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.8072, + "step": 31225 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31226 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 6.4254, + "step": 31227 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.7304, + "step": 31228 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31229 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.718, + "step": 31230 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31231 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.0343, + "step": 31232 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 6.1088, + "step": 31233 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 8.1093, + "step": 31234 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31235 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 9.7477, + "step": 31236 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31237 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.1054, + "step": 31238 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.9091, + "step": 31239 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31240 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.468, + "step": 31241 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31242 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.8751, + "step": 31243 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.9565, + "step": 31244 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31245 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31246 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31247 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31248 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.035, + "step": 31249 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31250 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31251 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31252 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31253 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31254 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 5.9984, + "step": 31255 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31256 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.7272, + "step": 31257 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 10.4144, + "step": 31258 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.0285, + "step": 31259 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31260 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.0269, + "step": 31261 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.1375, + "step": 31262 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.3732, + "step": 31263 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31264 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.8707, + "step": 31265 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 6.7851, + "step": 31266 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.9591, + "step": 31267 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.915, + "step": 31268 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 8.3058, + "step": 31269 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.1211, + "step": 31270 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 2.1616, + "step": 31271 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 5.3828, + "step": 31272 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 9.1276, + "step": 31273 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 9.5581, + "step": 31274 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31275 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31276 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.7726, + "step": 31277 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 4.3946, + "step": 31278 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.1033, + "step": 31279 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 7.9384, + "step": 31280 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.7987, + "step": 31281 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 1.1573, + "step": 31282 + }, + { + "epoch": 2.61, + "learning_rate": 0.00019985510073285592, + "loss": 3.3481, + "step": 31283 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31284 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31285 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31286 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31287 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 9.7192, + "step": 31288 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31289 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31290 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 6.6331, + "step": 31291 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31292 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31293 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31294 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31295 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31296 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 9.2543, + "step": 31297 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 3.0896, + "step": 31298 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31299 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31300 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31301 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31302 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.4464, + "step": 31303 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31304 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31305 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31306 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31307 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31308 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.2429, + "step": 31309 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.7997, + "step": 31310 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 5.2076, + "step": 31311 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.0981, + "step": 31312 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.4613, + "step": 31313 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 7.2937, + "step": 31314 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31315 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31316 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31317 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 7.0469, + "step": 31318 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31319 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.8718, + "step": 31320 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 3.795, + "step": 31321 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31322 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 4.1794, + "step": 31323 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.9658, + "step": 31324 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31325 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.7211, + "step": 31326 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.0141, + "step": 31327 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 3.4259, + "step": 31328 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31329 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.8888, + "step": 31330 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.3678, + "step": 31331 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.8501, + "step": 31332 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31333 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 8.519, + "step": 31334 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 4.3481, + "step": 31335 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31336 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.0983, + "step": 31337 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 4.1034, + "step": 31338 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31339 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31340 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31341 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31342 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.0022, + "step": 31343 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31344 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31345 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31346 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.0902, + "step": 31347 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31348 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 3.8795, + "step": 31349 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31350 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.8838, + "step": 31351 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 9.2193, + "step": 31352 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31353 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31354 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31355 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31356 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31357 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31358 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31359 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 8.6639, + "step": 31360 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 3.685, + "step": 31361 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31362 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.9869, + "step": 31363 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 4.7939, + "step": 31364 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 6.1771, + "step": 31365 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31366 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 8.5103, + "step": 31367 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.3804, + "step": 31368 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.8312, + "step": 31369 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31370 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.9002, + "step": 31371 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 6.3078, + "step": 31372 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31373 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.7501, + "step": 31374 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 8.5256, + "step": 31375 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31376 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31377 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31378 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.9945, + "step": 31379 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31380 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.9403, + "step": 31381 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31382 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 9.5698, + "step": 31383 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.8028, + "step": 31384 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 9.0693, + "step": 31385 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 4.3847, + "step": 31386 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31387 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 8.962, + "step": 31388 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31389 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.0121, + "step": 31390 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31391 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31392 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.7292, + "step": 31393 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.6949, + "step": 31394 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 3.0324, + "step": 31395 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 5.1599, + "step": 31396 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 3.2434, + "step": 31397 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.4675, + "step": 31398 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 7.089, + "step": 31399 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 2.1601, + "step": 31400 + }, + { + "epoch": 2.62, + "eval_loss": NaN, + "eval_runtime": 39.9859, + "eval_samples_per_second": 62.522, + "eval_steps_per_second": 7.828, + "step": 31400 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 1.4709, + "step": 31401 + }, + { + "epoch": 2.62, + "learning_rate": 0.00019985510073285592, + "loss": 9.6757, + "step": 31402 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 3.0594, + "step": 31403 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.0983, + "step": 31404 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31405 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31406 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31407 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 6.4033, + "step": 31408 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 4.3033, + "step": 31409 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 4.8203, + "step": 31410 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 8.1479, + "step": 31411 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 4.1799, + "step": 31412 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.0331, + "step": 31413 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31414 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31415 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.2888, + "step": 31416 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.8596, + "step": 31417 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31418 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.2097, + "step": 31419 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31420 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31421 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31422 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31423 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31424 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 5.6604, + "step": 31425 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31426 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31427 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.8537, + "step": 31428 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31429 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31430 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.9248, + "step": 31431 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31432 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 4.949, + "step": 31433 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.8005, + "step": 31434 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.9892, + "step": 31435 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.031, + "step": 31436 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 9.0869, + "step": 31437 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.0835, + "step": 31438 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 3.1034, + "step": 31439 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.5543, + "step": 31440 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.0675, + "step": 31441 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31442 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31443 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.0946, + "step": 31444 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31445 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.376, + "step": 31446 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.1646, + "step": 31447 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 4.0778, + "step": 31448 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 6.2471, + "step": 31449 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31450 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31451 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.9613, + "step": 31452 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.0444, + "step": 31453 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31454 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31455 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 3.4728, + "step": 31456 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.5494, + "step": 31457 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 4.0176, + "step": 31458 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.9978, + "step": 31459 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31460 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31461 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31462 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31463 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31464 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31465 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31466 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31467 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.2115, + "step": 31468 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.876, + "step": 31469 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31470 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31471 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.3237, + "step": 31472 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.0607, + "step": 31473 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.9481, + "step": 31474 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 8.7648, + "step": 31475 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.1211, + "step": 31476 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.8947, + "step": 31477 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 8.1859, + "step": 31478 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.1186, + "step": 31479 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 3.8615, + "step": 31480 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31481 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 6.1616, + "step": 31482 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31483 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 5.7829, + "step": 31484 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 8.4324, + "step": 31485 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31486 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 3.435, + "step": 31487 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.3675, + "step": 31488 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.3415, + "step": 31489 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.826, + "step": 31490 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 3.9953, + "step": 31491 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31492 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 5.9249, + "step": 31493 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 5.4187, + "step": 31494 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 7.1738, + "step": 31495 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31496 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31497 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.8365, + "step": 31498 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31499 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.9624, + "step": 31500 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31501 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.8296, + "step": 31502 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31503 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31504 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.1403, + "step": 31505 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 3.2609, + "step": 31506 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 8.8422, + "step": 31507 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31508 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 4.2761, + "step": 31509 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31510 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.0901, + "step": 31511 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31512 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31513 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 5.8231, + "step": 31514 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 1.959, + "step": 31515 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31516 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31517 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 2.5683, + "step": 31518 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31519 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31520 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 5.7671, + "step": 31521 + }, + { + "epoch": 2.63, + "learning_rate": 0.00019985510073285592, + "loss": 5.0582, + "step": 31522 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 6.9191, + "step": 31523 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.5148, + "step": 31524 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 5.9939, + "step": 31525 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31526 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.9659, + "step": 31527 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31528 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 2.1117, + "step": 31529 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 9.5048, + "step": 31530 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31531 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.7167, + "step": 31532 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.7301, + "step": 31533 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.9728, + "step": 31534 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31535 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.1187, + "step": 31536 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31537 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.9927, + "step": 31538 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 8.5555, + "step": 31539 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31540 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.707, + "step": 31541 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.9467, + "step": 31542 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 11.6081, + "step": 31543 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31544 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 5.4046, + "step": 31545 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.8993, + "step": 31546 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.9763, + "step": 31547 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31548 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.3966, + "step": 31549 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.4873, + "step": 31550 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31551 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 2.9845, + "step": 31552 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.2259, + "step": 31553 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31554 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31555 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.2085, + "step": 31556 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.6768, + "step": 31557 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31558 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 2.2084, + "step": 31559 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31560 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31561 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.868, + "step": 31562 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.7425, + "step": 31563 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31564 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31565 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31566 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 2.1445, + "step": 31567 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 7.3857, + "step": 31568 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 6.6914, + "step": 31569 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31570 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.0184, + "step": 31571 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.8724, + "step": 31572 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 2.0955, + "step": 31573 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31574 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31575 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.0144, + "step": 31576 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31577 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31578 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 6.6873, + "step": 31579 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.4069, + "step": 31580 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31581 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.8884, + "step": 31582 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.8724, + "step": 31583 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.9337, + "step": 31584 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.848, + "step": 31585 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.0705, + "step": 31586 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31587 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.4622, + "step": 31588 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 6.2792, + "step": 31589 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.9086, + "step": 31590 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31591 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 7.7328, + "step": 31592 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.9524, + "step": 31593 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31594 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31595 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 7.7515, + "step": 31596 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.4527, + "step": 31597 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31598 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31599 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 6.3865, + "step": 31600 + }, + { + "epoch": 2.64, + "eval_loss": NaN, + "eval_runtime": 39.9947, + "eval_samples_per_second": 62.508, + "eval_steps_per_second": 7.826, + "step": 31600 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31601 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31602 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.7506, + "step": 31603 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 2.5746, + "step": 31604 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.9608, + "step": 31605 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.0816, + "step": 31606 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31607 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 4.9623, + "step": 31608 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.9401, + "step": 31609 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31610 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31611 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31612 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31613 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.8395, + "step": 31614 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31615 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 6.9843, + "step": 31616 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.7512, + "step": 31617 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31618 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31619 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.8998, + "step": 31620 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.1948, + "step": 31621 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.0352, + "step": 31622 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31623 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.2182, + "step": 31624 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.587, + "step": 31625 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31626 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.2597, + "step": 31627 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.0173, + "step": 31628 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31629 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31630 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 7.0702, + "step": 31631 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 3.7052, + "step": 31632 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.9244, + "step": 31633 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31634 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 7.1125, + "step": 31635 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 2.8894, + "step": 31636 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31637 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31638 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31639 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31640 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31641 + }, + { + "epoch": 2.64, + "learning_rate": 0.00019985510073285592, + "loss": 1.7052, + "step": 31642 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 3.7844, + "step": 31643 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31644 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 10.4513, + "step": 31645 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.9703, + "step": 31646 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31647 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.8524, + "step": 31648 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 5.963, + "step": 31649 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.8661, + "step": 31650 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.8831, + "step": 31651 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31652 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 2.6157, + "step": 31653 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31654 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 3.7941, + "step": 31655 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.0016, + "step": 31656 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.1815, + "step": 31657 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31658 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31659 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31660 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 7.5643, + "step": 31661 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31662 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31663 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 9.2821, + "step": 31664 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31665 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 13.0876, + "step": 31666 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.3405, + "step": 31667 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 8.1423, + "step": 31668 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31669 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31670 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31671 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 8.7258, + "step": 31672 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 7.8198, + "step": 31673 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 5.6943, + "step": 31674 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.7824, + "step": 31675 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31676 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31677 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 7.6174, + "step": 31678 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.8225, + "step": 31679 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.9573, + "step": 31680 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31681 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 3.7228, + "step": 31682 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31683 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 2.2769, + "step": 31684 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 6.4278, + "step": 31685 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 6.7073, + "step": 31686 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31687 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.9477, + "step": 31688 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31689 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 10.4034, + "step": 31690 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31691 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31692 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31693 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31694 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.5846, + "step": 31695 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31696 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.7762, + "step": 31697 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 9.3609, + "step": 31698 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 7.9087, + "step": 31699 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31700 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.2502, + "step": 31701 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31702 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31703 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 6.7082, + "step": 31704 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31705 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31706 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31707 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31708 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 6.5, + "step": 31709 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.0915, + "step": 31710 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 8.3589, + "step": 31711 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.9165, + "step": 31712 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31713 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.3722, + "step": 31714 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31715 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 5.7804, + "step": 31716 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 8.0594, + "step": 31717 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 5.243, + "step": 31718 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31719 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 2.2915, + "step": 31720 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.0572, + "step": 31721 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31722 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 6.1022, + "step": 31723 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31724 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31725 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.0283, + "step": 31726 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.9073, + "step": 31727 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31728 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31729 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 2.2619, + "step": 31730 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.5675, + "step": 31731 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.9795, + "step": 31732 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31733 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 3.9967, + "step": 31734 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31735 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31736 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31737 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 3.931, + "step": 31738 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 3.1014, + "step": 31739 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31740 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31741 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31742 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 2.042, + "step": 31743 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.118, + "step": 31744 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.0965, + "step": 31745 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31746 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 2.0624, + "step": 31747 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 8.0903, + "step": 31748 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 9.1863, + "step": 31749 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 2.644, + "step": 31750 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 6.0258, + "step": 31751 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31752 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31753 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 4.6013, + "step": 31754 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31755 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.7047, + "step": 31756 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 1.415, + "step": 31757 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31758 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31759 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31760 + }, + { + "epoch": 2.65, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31761 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31762 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31763 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 9.7601, + "step": 31764 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.8564, + "step": 31765 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31766 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.9099, + "step": 31767 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 10.9292, + "step": 31768 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.1776, + "step": 31769 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31770 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.9003, + "step": 31771 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.0272, + "step": 31772 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 9.9398, + "step": 31773 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.1714, + "step": 31774 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 6.675, + "step": 31775 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.4359, + "step": 31776 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.6478, + "step": 31777 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31778 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.4405, + "step": 31779 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.6349, + "step": 31780 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.8851, + "step": 31781 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.3162, + "step": 31782 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.1485, + "step": 31783 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.8714, + "step": 31784 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 11.3716, + "step": 31785 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31786 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.5733, + "step": 31787 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31788 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.021, + "step": 31789 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 5.6775, + "step": 31790 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31791 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.951, + "step": 31792 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.1252, + "step": 31793 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 5.2926, + "step": 31794 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31795 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.4363, + "step": 31796 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 3.4772, + "step": 31797 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31798 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31799 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31800 + }, + { + "epoch": 2.66, + "eval_loss": NaN, + "eval_runtime": 39.9785, + "eval_samples_per_second": 62.534, + "eval_steps_per_second": 7.829, + "step": 31800 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.6555, + "step": 31801 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.2401, + "step": 31802 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.0625, + "step": 31803 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.3038, + "step": 31804 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31805 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.526, + "step": 31806 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 3.8981, + "step": 31807 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.6639, + "step": 31808 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31809 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.3141, + "step": 31810 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.0675, + "step": 31811 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.5781, + "step": 31812 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.9052, + "step": 31813 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31814 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.5106, + "step": 31815 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31816 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.0568, + "step": 31817 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 6.4236, + "step": 31818 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31819 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31820 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31821 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.2919, + "step": 31822 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.3194, + "step": 31823 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 3.7745, + "step": 31824 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.271, + "step": 31825 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.0009, + "step": 31826 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.1332, + "step": 31827 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 8.095, + "step": 31828 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.325, + "step": 31829 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 8.0986, + "step": 31830 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 11.5063, + "step": 31831 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31832 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.7483, + "step": 31833 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31834 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 8.5896, + "step": 31835 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31836 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31837 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.1204, + "step": 31838 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31839 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.1569, + "step": 31840 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31841 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31842 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 8.3236, + "step": 31843 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31844 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31845 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31846 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.7544, + "step": 31847 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.4589, + "step": 31848 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 8.2917, + "step": 31849 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31850 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31851 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.8849, + "step": 31852 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.3183, + "step": 31853 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31854 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.8638, + "step": 31855 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31856 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31857 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.8753, + "step": 31858 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.6502, + "step": 31859 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31860 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31861 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31862 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31863 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 2.0351, + "step": 31864 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 7.1788, + "step": 31865 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.3738, + "step": 31866 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31867 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31868 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.8482, + "step": 31869 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 11.9772, + "step": 31870 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31871 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 5.6529, + "step": 31872 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31873 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31874 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 4.8051, + "step": 31875 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 1.6393, + "step": 31876 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31877 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31878 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31879 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31880 + }, + { + "epoch": 2.66, + "learning_rate": 0.00019985510073285592, + "loss": 0.7085, + "step": 31881 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31882 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 7.7521, + "step": 31883 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 5.1369, + "step": 31884 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31885 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.4068, + "step": 31886 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31887 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.9555, + "step": 31888 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31889 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.2014, + "step": 31890 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.0222, + "step": 31891 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 2.3074, + "step": 31892 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 7.3482, + "step": 31893 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31894 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.9149, + "step": 31895 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 5.2543, + "step": 31896 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.9045, + "step": 31897 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.7447, + "step": 31898 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31899 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.9031, + "step": 31900 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31901 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31902 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31903 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.6189, + "step": 31904 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31905 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 11.9304, + "step": 31906 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 8.3447, + "step": 31907 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.2097, + "step": 31908 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 2.3118, + "step": 31909 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.9645, + "step": 31910 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.0547, + "step": 31911 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31912 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.9264, + "step": 31913 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31914 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31915 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31916 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 2.0146, + "step": 31917 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.6995, + "step": 31918 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.2535, + "step": 31919 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31920 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.3849, + "step": 31921 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 6.6968, + "step": 31922 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31923 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31924 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 8.268, + "step": 31925 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31926 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 5.5805, + "step": 31927 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 5.9774, + "step": 31928 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.7949, + "step": 31929 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 2.002, + "step": 31930 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31931 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 6.7887, + "step": 31932 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31933 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31934 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31935 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.9947, + "step": 31936 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.0842, + "step": 31937 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 7.9192, + "step": 31938 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31939 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31940 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31941 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.8508, + "step": 31942 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31943 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 9.6911, + "step": 31944 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31945 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 5.449, + "step": 31946 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31947 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31948 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.774, + "step": 31949 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 7.9876, + "step": 31950 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.5766, + "step": 31951 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 5.4483, + "step": 31952 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31953 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31954 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.9137, + "step": 31955 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31956 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31957 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 6.44, + "step": 31958 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 5.1313, + "step": 31959 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31960 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31961 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.8827, + "step": 31962 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31963 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.2244, + "step": 31964 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31965 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31966 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.2005, + "step": 31967 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31968 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.6607, + "step": 31969 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31970 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 6.5679, + "step": 31971 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31972 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31973 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31974 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.975, + "step": 31975 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31976 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31977 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.7079, + "step": 31978 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31979 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.02, + "step": 31980 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31981 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31982 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31983 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31984 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.6163, + "step": 31985 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 3.1204, + "step": 31986 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.8103, + "step": 31987 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.0252, + "step": 31988 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31989 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 9.1049, + "step": 31990 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31991 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 4.3807, + "step": 31992 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 6.2882, + "step": 31993 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31994 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.6345, + "step": 31995 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31996 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31997 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 31998 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 1.9588, + "step": 31999 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32000 + }, + { + "epoch": 2.67, + "eval_loss": NaN, + "eval_runtime": 39.9583, + "eval_samples_per_second": 62.565, + "eval_steps_per_second": 7.833, + "step": 32000 + }, + { + "epoch": 2.67, + "learning_rate": 0.00019985510073285592, + "loss": 7.4847, + "step": 32001 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32002 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32003 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32004 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32005 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32006 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 8.2718, + "step": 32007 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32008 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.4133, + "step": 32009 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32010 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32011 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32012 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.7902, + "step": 32013 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32014 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32015 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.5306, + "step": 32016 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32017 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.0573, + "step": 32018 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.6107, + "step": 32019 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.4993, + "step": 32020 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32021 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 4.4453, + "step": 32022 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.9961, + "step": 32023 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 4.3226, + "step": 32024 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.9956, + "step": 32025 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32026 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32027 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.8958, + "step": 32028 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32029 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32030 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32031 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.1637, + "step": 32032 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32033 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32034 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32035 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32036 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32037 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.8232, + "step": 32038 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 5.6442, + "step": 32039 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32040 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 4.2509, + "step": 32041 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 5.8438, + "step": 32042 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.6334, + "step": 32043 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32044 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32045 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32046 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32047 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.8477, + "step": 32048 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.2481, + "step": 32049 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 8.735, + "step": 32050 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32051 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.008, + "step": 32052 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.1842, + "step": 32053 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32054 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 5.9631, + "step": 32055 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 9.4207, + "step": 32056 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32057 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.9181, + "step": 32058 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.8113, + "step": 32059 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 10.451, + "step": 32060 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32061 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.7216, + "step": 32062 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 7.1203, + "step": 32063 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.6296, + "step": 32064 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32065 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.1548, + "step": 32066 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 6.8732, + "step": 32067 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32068 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32069 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 5.3584, + "step": 32070 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32071 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32072 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.5741, + "step": 32073 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.1068, + "step": 32074 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.2541, + "step": 32075 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 5.6555, + "step": 32076 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.103, + "step": 32077 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32078 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32079 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32080 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32081 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 4.6527, + "step": 32082 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32083 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32084 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 2.075, + "step": 32085 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.6596, + "step": 32086 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 4.4899, + "step": 32087 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 7.9769, + "step": 32088 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 6.0755, + "step": 32089 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32090 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32091 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.0238, + "step": 32092 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32093 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.9094, + "step": 32094 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32095 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 7.4991, + "step": 32096 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32097 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32098 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32099 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32100 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.6713, + "step": 32101 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.3003, + "step": 32102 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 6.3967, + "step": 32103 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32104 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32105 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 8.565, + "step": 32106 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32107 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32108 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32109 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32110 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32111 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32112 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.9731, + "step": 32113 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32114 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32115 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32116 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 1.8443, + "step": 32117 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32118 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32119 + }, + { + "epoch": 2.68, + "learning_rate": 0.00019985510073285592, + "loss": 3.987, + "step": 32120 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32121 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 3.8599, + "step": 32122 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.8281, + "step": 32123 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 6.2407, + "step": 32124 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.6826, + "step": 32125 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9819, + "step": 32126 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9928, + "step": 32127 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32128 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.9928, + "step": 32129 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.2236, + "step": 32130 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.8502, + "step": 32131 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 7.193, + "step": 32132 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.2753, + "step": 32133 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 8.3279, + "step": 32134 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 3.5661, + "step": 32135 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32136 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32137 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32138 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32139 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 9.7414, + "step": 32140 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32141 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 6.7536, + "step": 32142 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32143 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 7.4352, + "step": 32144 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 5.9078, + "step": 32145 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 6.7455, + "step": 32146 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9763, + "step": 32147 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 3.0831, + "step": 32148 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.9492, + "step": 32149 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 4.7799, + "step": 32150 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.688, + "step": 32151 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 6.3467, + "step": 32152 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32153 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.8784, + "step": 32154 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 8.3178, + "step": 32155 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.8322, + "step": 32156 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32157 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9529, + "step": 32158 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32159 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32160 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 5.2965, + "step": 32161 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 4.1336, + "step": 32162 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 4.0521, + "step": 32163 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.2783, + "step": 32164 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32165 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 4.4643, + "step": 32166 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.156, + "step": 32167 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 3.6663, + "step": 32168 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9867, + "step": 32169 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 5.8601, + "step": 32170 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.4403, + "step": 32171 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 3.7121, + "step": 32172 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.7104, + "step": 32173 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32174 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 5.9651, + "step": 32175 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32176 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.95, + "step": 32177 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 4.428, + "step": 32178 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32179 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32180 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 3.5594, + "step": 32181 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32182 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.8489, + "step": 32183 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32184 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32185 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9891, + "step": 32186 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.889, + "step": 32187 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 7.0899, + "step": 32188 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 10.4785, + "step": 32189 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32190 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32191 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 3.865, + "step": 32192 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.8844, + "step": 32193 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32194 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.0372, + "step": 32195 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32196 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32197 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 8.1492, + "step": 32198 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32199 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.9399, + "step": 32200 + }, + { + "epoch": 2.69, + "eval_loss": NaN, + "eval_runtime": 39.9252, + "eval_samples_per_second": 62.617, + "eval_steps_per_second": 7.84, + "step": 32200 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.2506, + "step": 32201 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.0808, + "step": 32202 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 1.1593, + "step": 32203 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32204 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32205 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32206 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32207 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.0292, + "step": 32208 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.416, + "step": 32209 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 6.4012, + "step": 32210 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32211 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32212 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9948, + "step": 32213 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 2.1672, + "step": 32214 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32215 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 5.0697, + "step": 32216 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 4.0703, + "step": 32217 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32218 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32219 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32220 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32221 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.0, + "step": 32222 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 0.9259, + "step": 32223 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 8.8598, + "step": 32224 + }, + { + "epoch": 2.69, + "learning_rate": 0.00019985510073285592, + "loss": 4.1952, + "step": 32225 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 4.0897, + "step": 32226 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 3.8264, + "step": 32227 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 2.408, + "step": 32228 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 3.7503, + "step": 32229 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32230 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 4.5581, + "step": 32231 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 7.5102, + "step": 32232 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 10.7298, + "step": 32233 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 6.2779, + "step": 32234 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 4.6165, + "step": 32235 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32236 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32237 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 3.8168, + "step": 32238 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32239 + }, + { + "epoch": 2.69, + "learning_rate": 0.0001998495276841196, + "loss": 2.3058, + "step": 32240 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.8144, + "step": 32241 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.552, + "step": 32242 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32243 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32244 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32245 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32246 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32247 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 6.7423, + "step": 32248 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.6302, + "step": 32249 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.903, + "step": 32250 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.9459, + "step": 32251 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32252 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32253 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.9018, + "step": 32254 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32255 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32256 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 6.8029, + "step": 32257 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32258 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32259 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32260 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.0108, + "step": 32261 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32262 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.9154, + "step": 32263 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32264 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 4.1211, + "step": 32265 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 5.9018, + "step": 32266 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32267 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32268 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 4.1753, + "step": 32269 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 6.7335, + "step": 32270 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.8172, + "step": 32271 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32272 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.3193, + "step": 32273 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.0368, + "step": 32274 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.4915, + "step": 32275 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32276 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.0797, + "step": 32277 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.961, + "step": 32278 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.2141, + "step": 32279 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.0838, + "step": 32280 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 5.7856, + "step": 32281 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32282 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.3271, + "step": 32283 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.6536, + "step": 32284 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32285 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.1731, + "step": 32286 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.5502, + "step": 32287 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32288 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.7429, + "step": 32289 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32290 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32291 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.6473, + "step": 32292 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32293 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32294 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32295 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.3999, + "step": 32296 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 4.9464, + "step": 32297 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.9062, + "step": 32298 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.7987, + "step": 32299 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.1933, + "step": 32300 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 4.9471, + "step": 32301 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.1894, + "step": 32302 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.0106, + "step": 32303 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.6862, + "step": 32304 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 9.1254, + "step": 32305 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 5.0091, + "step": 32306 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32307 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32308 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32309 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32310 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32311 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.9362, + "step": 32312 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32313 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.9334, + "step": 32314 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.2267, + "step": 32315 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32316 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32317 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32318 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.5243, + "step": 32319 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.3885, + "step": 32320 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32321 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32322 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32323 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.9485, + "step": 32324 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.6683, + "step": 32325 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 6.4419, + "step": 32326 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.4706, + "step": 32327 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32328 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.0883, + "step": 32329 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.2385, + "step": 32330 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32331 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32332 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.2433, + "step": 32333 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32334 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32335 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32336 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.5221, + "step": 32337 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32338 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.6533, + "step": 32339 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32340 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.87, + "step": 32341 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32342 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 3.9997, + "step": 32343 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32344 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32345 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 4.2675, + "step": 32346 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.8191, + "step": 32347 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.8557, + "step": 32348 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32349 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 7.3925, + "step": 32350 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 1.7063, + "step": 32351 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.0161, + "step": 32352 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32353 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 4.4817, + "step": 32354 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.6778, + "step": 32355 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 2.4141, + "step": 32356 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.988, + "step": 32357 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32358 + }, + { + "epoch": 2.7, + "learning_rate": 0.0001998495276841196, + "loss": 8.3474, + "step": 32359 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.2413, + "step": 32360 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32361 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.0726, + "step": 32362 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 10.469, + "step": 32363 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32364 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32365 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.1344, + "step": 32366 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 10.8439, + "step": 32367 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32368 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.1192, + "step": 32369 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.2186, + "step": 32370 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.8838, + "step": 32371 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.1865, + "step": 32372 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.9748, + "step": 32373 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.2043, + "step": 32374 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.9984, + "step": 32375 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.7623, + "step": 32376 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.1345, + "step": 32377 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.9124, + "step": 32378 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 4.7608, + "step": 32379 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32380 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.723, + "step": 32381 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.8909, + "step": 32382 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.0461, + "step": 32383 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32384 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32385 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32386 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32387 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.891, + "step": 32388 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 4.0892, + "step": 32389 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.9879, + "step": 32390 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.5317, + "step": 32391 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32392 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32393 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32394 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.1778, + "step": 32395 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.982, + "step": 32396 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32397 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32398 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.7181, + "step": 32399 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 5.5056, + "step": 32400 + }, + { + "epoch": 2.71, + "eval_loss": NaN, + "eval_runtime": 39.9421, + "eval_samples_per_second": 62.591, + "eval_steps_per_second": 7.836, + "step": 32400 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32401 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32402 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.3657, + "step": 32403 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.1816, + "step": 32404 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.0584, + "step": 32405 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32406 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32407 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.0959, + "step": 32408 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 4.7671, + "step": 32409 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.3384, + "step": 32410 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 9.5064, + "step": 32411 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.2648, + "step": 32412 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32413 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.874, + "step": 32414 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 4.5465, + "step": 32415 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.1077, + "step": 32416 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32417 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 5.8543, + "step": 32418 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.6042, + "step": 32419 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32420 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.6596, + "step": 32421 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32422 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32423 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.4668, + "step": 32424 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32425 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32426 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.9782, + "step": 32427 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32428 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.7541, + "step": 32429 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.3496, + "step": 32430 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.9716, + "step": 32431 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32432 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.0555, + "step": 32433 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.861, + "step": 32434 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.7145, + "step": 32435 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.7599, + "step": 32436 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32437 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32438 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.1783, + "step": 32439 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32440 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32441 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32442 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.7848, + "step": 32443 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32444 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.9292, + "step": 32445 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32446 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.2729, + "step": 32447 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32448 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.843, + "step": 32449 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32450 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32451 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32452 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.7431, + "step": 32453 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 5.3121, + "step": 32454 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32455 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32456 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32457 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32458 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.4168, + "step": 32459 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32460 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32461 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.0734, + "step": 32462 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32463 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.5545, + "step": 32464 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.0137, + "step": 32465 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32466 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 4.5734, + "step": 32467 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.8335, + "step": 32468 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32469 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32470 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 1.1056, + "step": 32471 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 8.3452, + "step": 32472 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 4.4803, + "step": 32473 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 3.8201, + "step": 32474 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 6.4014, + "step": 32475 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 7.8756, + "step": 32476 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 4.5331, + "step": 32477 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 2.2078, + "step": 32478 + }, + { + "epoch": 2.71, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32479 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.7112, + "step": 32480 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.7327, + "step": 32481 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32482 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32483 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 5.6642, + "step": 32484 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32485 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32486 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 7.3658, + "step": 32487 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.9309, + "step": 32488 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 4.0783, + "step": 32489 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32490 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.8632, + "step": 32491 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32492 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32493 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.9356, + "step": 32494 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32495 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32496 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 11.0517, + "step": 32497 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32498 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32499 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32500 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32501 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32502 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.1579, + "step": 32503 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.0409, + "step": 32504 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32505 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.2793, + "step": 32506 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32507 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32508 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32509 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32510 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.7087, + "step": 32511 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.8669, + "step": 32512 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32513 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.7948, + "step": 32514 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32515 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.8394, + "step": 32516 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32517 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32518 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.9648, + "step": 32519 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 4.0926, + "step": 32520 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.5912, + "step": 32521 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 7.1428, + "step": 32522 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 7.1957, + "step": 32523 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32524 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32525 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 16.1446, + "step": 32526 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.2081, + "step": 32527 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32528 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 5.5739, + "step": 32529 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.8759, + "step": 32530 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32531 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32532 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32533 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 5.3244, + "step": 32534 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 4.2051, + "step": 32535 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.6294, + "step": 32536 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32537 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.9537, + "step": 32538 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 7.466, + "step": 32539 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.5839, + "step": 32540 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.7396, + "step": 32541 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.7154, + "step": 32542 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 4.2415, + "step": 32543 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32544 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32545 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.7393, + "step": 32546 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32547 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.1034, + "step": 32548 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.0651, + "step": 32549 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32550 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.34, + "step": 32551 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 5.3299, + "step": 32552 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32553 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32554 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.4875, + "step": 32555 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32556 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32557 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32558 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32559 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 7.4712, + "step": 32560 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32561 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.6688, + "step": 32562 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32563 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.8315, + "step": 32564 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.0515, + "step": 32565 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.0943, + "step": 32566 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 9.2393, + "step": 32567 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.8363, + "step": 32568 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.5532, + "step": 32569 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 5.5983, + "step": 32570 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32571 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32572 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32573 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.9898, + "step": 32574 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.8967, + "step": 32575 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32576 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32577 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32578 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32579 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.0158, + "step": 32580 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32581 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 7.8708, + "step": 32582 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32583 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32584 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32585 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 1.7012, + "step": 32586 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 5.6761, + "step": 32587 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.451, + "step": 32588 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32589 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32590 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 3.9281, + "step": 32591 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 11.0396, + "step": 32592 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32593 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32594 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32595 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 5.355, + "step": 32596 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 2.0257, + "step": 32597 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 13.8269, + "step": 32598 + }, + { + "epoch": 2.72, + "learning_rate": 0.0001998495276841196, + "loss": 8.5837, + "step": 32599 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.3077, + "step": 32600 + }, + { + "epoch": 2.73, + "eval_loss": NaN, + "eval_runtime": 39.936, + "eval_samples_per_second": 62.6, + "eval_steps_per_second": 7.838, + "step": 32600 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32601 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.4603, + "step": 32602 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.7913, + "step": 32603 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32604 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 6.8762, + "step": 32605 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32606 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32607 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.2214, + "step": 32608 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.8298, + "step": 32609 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.2119, + "step": 32610 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32611 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32612 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32613 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.6159, + "step": 32614 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 8.6826, + "step": 32615 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 8.41, + "step": 32616 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32617 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 5.7369, + "step": 32618 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32619 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32620 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.7099, + "step": 32621 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.3209, + "step": 32622 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.8503, + "step": 32623 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.7842, + "step": 32624 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 8.9478, + "step": 32625 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.2461, + "step": 32626 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.4253, + "step": 32627 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.9896, + "step": 32628 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.4045, + "step": 32629 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.3755, + "step": 32630 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 11.9396, + "step": 32631 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.3642, + "step": 32632 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32633 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.2449, + "step": 32634 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32635 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32636 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.2188, + "step": 32637 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32638 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32639 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 9.3811, + "step": 32640 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32641 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.2905, + "step": 32642 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32643 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 6.272, + "step": 32644 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 7.8466, + "step": 32645 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32646 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.833, + "step": 32647 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 5.7518, + "step": 32648 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32649 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.8975, + "step": 32650 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32651 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.5248, + "step": 32652 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.9435, + "step": 32653 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32654 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32655 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32656 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32657 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32658 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 9.2239, + "step": 32659 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.9991, + "step": 32660 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 6.1661, + "step": 32661 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.1725, + "step": 32662 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 10.8159, + "step": 32663 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32664 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32665 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.1849, + "step": 32666 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.2937, + "step": 32667 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 7.7643, + "step": 32668 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.9541, + "step": 32669 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.1589, + "step": 32670 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.0536, + "step": 32671 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.8364, + "step": 32672 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32673 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.476, + "step": 32674 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.0433, + "step": 32675 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.7892, + "step": 32676 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32677 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.8977, + "step": 32678 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.9401, + "step": 32679 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 7.2527, + "step": 32680 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32681 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.3219, + "step": 32682 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.0195, + "step": 32683 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32684 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32685 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 7.6698, + "step": 32686 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32687 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.03, + "step": 32688 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.8009, + "step": 32689 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32690 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32691 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.9108, + "step": 32692 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32693 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.5187, + "step": 32694 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32695 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32696 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 2.036, + "step": 32697 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 5.0427, + "step": 32698 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 4.3272, + "step": 32699 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.9387, + "step": 32700 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.9184, + "step": 32701 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32702 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32703 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 5.0855, + "step": 32704 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.0667, + "step": 32705 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32706 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32707 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 7.568, + "step": 32708 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32709 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.0501, + "step": 32710 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 1.9817, + "step": 32711 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.1614, + "step": 32712 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32713 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32714 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 8.34, + "step": 32715 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 3.75, + "step": 32716 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32717 + }, + { + "epoch": 2.73, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32718 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32719 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.2381, + "step": 32720 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 7.196, + "step": 32721 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.9917, + "step": 32722 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32723 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.4463, + "step": 32724 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.9073, + "step": 32725 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.5125, + "step": 32726 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.5914, + "step": 32727 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 13.4394, + "step": 32728 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32729 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 7.8852, + "step": 32730 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.8977, + "step": 32731 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.9704, + "step": 32732 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32733 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.5231, + "step": 32734 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32735 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32736 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.1422, + "step": 32737 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.1249, + "step": 32738 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.1141, + "step": 32739 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 11.0289, + "step": 32740 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.4725, + "step": 32741 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.4181, + "step": 32742 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 8.1079, + "step": 32743 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.0102, + "step": 32744 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 9.7897, + "step": 32745 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32746 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.1593, + "step": 32747 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.121, + "step": 32748 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32749 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.2394, + "step": 32750 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.9637, + "step": 32751 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.9953, + "step": 32752 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32753 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.5696, + "step": 32754 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32755 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.9632, + "step": 32756 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32757 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32758 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.8616, + "step": 32759 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32760 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.8585, + "step": 32761 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32762 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32763 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.2316, + "step": 32764 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32765 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32766 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.8854, + "step": 32767 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32768 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.4907, + "step": 32769 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 6.9536, + "step": 32770 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32771 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32772 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.1867, + "step": 32773 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32774 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.0854, + "step": 32775 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.3434, + "step": 32776 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32777 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32778 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 8.6377, + "step": 32779 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32780 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.9762, + "step": 32781 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.7054, + "step": 32782 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 10.2118, + "step": 32783 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32784 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.6246, + "step": 32785 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32786 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.9925, + "step": 32787 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32788 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32789 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.0801, + "step": 32790 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32791 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.9248, + "step": 32792 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32793 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.3964, + "step": 32794 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.9212, + "step": 32795 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 7.5169, + "step": 32796 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32797 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.3259, + "step": 32798 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.5799, + "step": 32799 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32800 + }, + { + "epoch": 2.74, + "eval_loss": NaN, + "eval_runtime": 39.899, + "eval_samples_per_second": 62.658, + "eval_steps_per_second": 7.845, + "step": 32800 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32801 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32802 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.0214, + "step": 32803 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32804 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.2917, + "step": 32805 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.3725, + "step": 32806 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.8864, + "step": 32807 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.4643, + "step": 32808 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 7.1324, + "step": 32809 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32810 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.9962, + "step": 32811 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 7.8662, + "step": 32812 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.044, + "step": 32813 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32814 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 2.1212, + "step": 32815 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32816 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.4878, + "step": 32817 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32818 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.7569, + "step": 32819 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 7.1323, + "step": 32820 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32821 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32822 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32823 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 3.0687, + "step": 32824 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 6.8177, + "step": 32825 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.3301, + "step": 32826 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32827 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 8.4885, + "step": 32828 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.0006, + "step": 32829 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32830 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 4.4314, + "step": 32831 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32832 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32833 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 1.9017, + "step": 32834 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 9.7886, + "step": 32835 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 6.8286, + "step": 32836 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 7.1508, + "step": 32837 + }, + { + "epoch": 2.74, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32838 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32839 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 8.1034, + "step": 32840 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 3.4543, + "step": 32841 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32842 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32843 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 3.7081, + "step": 32844 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32845 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32846 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.3164, + "step": 32847 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32848 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.1382, + "step": 32849 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.3288, + "step": 32850 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32851 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.902, + "step": 32852 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32853 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.7589, + "step": 32854 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.4213, + "step": 32855 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.3108, + "step": 32856 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32857 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.77, + "step": 32858 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32859 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32860 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 9.1206, + "step": 32861 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32862 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32863 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32864 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32865 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.0783, + "step": 32866 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 5.1484, + "step": 32867 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32868 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32869 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32870 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.9611, + "step": 32871 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32872 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.9364, + "step": 32873 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.0431, + "step": 32874 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.6573, + "step": 32875 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.5007, + "step": 32876 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.1622, + "step": 32877 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.912, + "step": 32878 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32879 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 4.3311, + "step": 32880 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 4.9814, + "step": 32881 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32882 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32883 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.002, + "step": 32884 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32885 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32886 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.032, + "step": 32887 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.9409, + "step": 32888 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.0309, + "step": 32889 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32890 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32891 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32892 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 9.2296, + "step": 32893 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 3.5999, + "step": 32894 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32895 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.7779, + "step": 32896 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32897 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32898 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.355, + "step": 32899 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32900 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32901 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.8312, + "step": 32902 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 8.4508, + "step": 32903 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 3.4204, + "step": 32904 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.6667, + "step": 32905 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.5356, + "step": 32906 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32907 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 4.5991, + "step": 32908 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32909 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32910 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32911 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 4.6051, + "step": 32912 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32913 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.3346, + "step": 32914 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32915 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 5.8238, + "step": 32916 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32917 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32918 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 5.9629, + "step": 32919 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.0268, + "step": 32920 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.9829, + "step": 32921 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32922 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.9408, + "step": 32923 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.2073, + "step": 32924 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32925 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32926 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.9983, + "step": 32927 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 5.8076, + "step": 32928 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32929 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32930 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.8951, + "step": 32931 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32932 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32933 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 4.1995, + "step": 32934 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32935 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 3.2997, + "step": 32936 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 6.5984, + "step": 32937 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.5964, + "step": 32938 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 4.1629, + "step": 32939 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32940 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32941 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32942 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32943 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32944 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.9833, + "step": 32945 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.3466, + "step": 32946 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.8478, + "step": 32947 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 3.7626, + "step": 32948 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32949 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 2.0286, + "step": 32950 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 7.7101, + "step": 32951 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32952 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 1.4588, + "step": 32953 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32954 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32955 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32956 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 4.6136, + "step": 32957 + }, + { + "epoch": 2.75, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32958 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.1172, + "step": 32959 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32960 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.7641, + "step": 32961 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32962 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.998, + "step": 32963 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32964 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.154, + "step": 32965 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32966 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32967 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.3037, + "step": 32968 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.322, + "step": 32969 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.9744, + "step": 32970 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 4.0476, + "step": 32971 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32972 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.7506, + "step": 32973 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32974 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32975 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32976 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32977 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.0579, + "step": 32978 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32979 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32980 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32981 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32982 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32983 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.0448, + "step": 32984 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.8948, + "step": 32985 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32986 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.0895, + "step": 32987 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 8.3004, + "step": 32988 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32989 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32990 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32991 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.1713, + "step": 32992 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.8115, + "step": 32993 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.0884, + "step": 32994 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 5.8769, + "step": 32995 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.0665, + "step": 32996 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 4.2246, + "step": 32997 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32998 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 32999 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33000 + }, + { + "epoch": 2.76, + "eval_loss": NaN, + "eval_runtime": 39.8767, + "eval_samples_per_second": 62.693, + "eval_steps_per_second": 7.849, + "step": 33000 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 9.2614, + "step": 33001 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 4.8521, + "step": 33002 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.0689, + "step": 33003 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.4217, + "step": 33004 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33005 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33006 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.7537, + "step": 33007 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.2652, + "step": 33008 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.6325, + "step": 33009 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.2917, + "step": 33010 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 7.3094, + "step": 33011 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33012 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33013 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.5233, + "step": 33014 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.4104, + "step": 33015 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33016 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.625, + "step": 33017 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.992, + "step": 33018 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33019 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 6.73, + "step": 33020 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33021 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.4238, + "step": 33022 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 5.2233, + "step": 33023 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 8.0648, + "step": 33024 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.6171, + "step": 33025 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 6.0468, + "step": 33026 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.8305, + "step": 33027 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33028 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33029 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33030 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 7.0254, + "step": 33031 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 4.367, + "step": 33032 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.4123, + "step": 33033 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.93, + "step": 33034 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33035 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33036 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.637, + "step": 33037 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33038 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33039 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 2.1595, + "step": 33040 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.0838, + "step": 33041 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.0595, + "step": 33042 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.3508, + "step": 33043 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.3812, + "step": 33044 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33045 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33046 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33047 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33048 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33049 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33050 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 4.2249, + "step": 33051 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33052 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33053 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 7.5286, + "step": 33054 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 5.5182, + "step": 33055 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.699, + "step": 33056 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.7212, + "step": 33057 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33058 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 4.8859, + "step": 33059 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.6152, + "step": 33060 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33061 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.9118, + "step": 33062 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.9281, + "step": 33063 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 9.2211, + "step": 33064 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33065 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33066 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.2817, + "step": 33067 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33068 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.5775, + "step": 33069 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 10.0437, + "step": 33070 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.871, + "step": 33071 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.4987, + "step": 33072 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.8465, + "step": 33073 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 1.1176, + "step": 33074 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33075 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33076 + }, + { + "epoch": 2.76, + "learning_rate": 0.0001998495276841196, + "loss": 3.7237, + "step": 33077 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 2.3785, + "step": 33078 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 9.3411, + "step": 33079 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33080 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 4.0559, + "step": 33081 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33082 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 6.7288, + "step": 33083 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.4503, + "step": 33084 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 4.456, + "step": 33085 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33086 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33087 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.8715, + "step": 33088 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.7687, + "step": 33089 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33090 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33091 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.9517, + "step": 33092 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33093 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 6.9768, + "step": 33094 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.9074, + "step": 33095 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 4.032, + "step": 33096 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.8976, + "step": 33097 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.6059, + "step": 33098 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.9446, + "step": 33099 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33100 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33101 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33102 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33103 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33104 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33105 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 3.8063, + "step": 33106 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33107 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33108 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 3.3968, + "step": 33109 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33110 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33111 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.402, + "step": 33112 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 3.8122, + "step": 33113 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.3646, + "step": 33114 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33115 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.1367, + "step": 33116 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.0306, + "step": 33117 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 2.1759, + "step": 33118 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33119 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33120 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33121 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33122 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.925, + "step": 33123 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33124 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33125 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.6124, + "step": 33126 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33127 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33128 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.5777, + "step": 33129 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.9139, + "step": 33130 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33131 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 9.9704, + "step": 33132 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 2.2213, + "step": 33133 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33134 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33135 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33136 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.2275, + "step": 33137 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.8266, + "step": 33138 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 4.0352, + "step": 33139 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33140 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33141 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33142 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 2.2122, + "step": 33143 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.9152, + "step": 33144 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 3.1256, + "step": 33145 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33146 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33147 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33148 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33149 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 3.9844, + "step": 33150 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33151 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33152 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33153 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 2.6727, + "step": 33154 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 3.8946, + "step": 33155 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 3.6648, + "step": 33156 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33157 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.8614, + "step": 33158 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33159 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.1781, + "step": 33160 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 5.36, + "step": 33161 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33162 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.8567, + "step": 33163 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33164 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33165 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33166 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 5.5631, + "step": 33167 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.5728, + "step": 33168 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 6.0563, + "step": 33169 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33170 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33171 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 4.4607, + "step": 33172 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 4.7402, + "step": 33173 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33174 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33175 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.9612, + "step": 33176 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33177 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33178 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33179 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33180 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33181 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33182 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33183 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33184 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33185 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33186 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.9344, + "step": 33187 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33188 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.0439, + "step": 33189 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33190 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.0962, + "step": 33191 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33192 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 2.1871, + "step": 33193 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.8657, + "step": 33194 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 1.0016, + "step": 33195 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33196 + }, + { + "epoch": 2.77, + "learning_rate": 0.0001998495276841196, + "loss": 8.5486, + "step": 33197 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.8738, + "step": 33198 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33199 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33200 + }, + { + "epoch": 2.78, + "eval_loss": NaN, + "eval_runtime": 39.838, + "eval_samples_per_second": 62.754, + "eval_steps_per_second": 7.857, + "step": 33200 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33201 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.9168, + "step": 33202 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.8929, + "step": 33203 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33204 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.0347, + "step": 33205 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.3428, + "step": 33206 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.1401, + "step": 33207 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33208 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.1916, + "step": 33209 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.1594, + "step": 33210 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33211 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.2601, + "step": 33212 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 8.0624, + "step": 33213 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33214 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.6581, + "step": 33215 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33216 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.3215, + "step": 33217 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 6.963, + "step": 33218 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 6.4004, + "step": 33219 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.0348, + "step": 33220 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.2341, + "step": 33221 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.6233, + "step": 33222 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.2112, + "step": 33223 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 5.2251, + "step": 33224 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33225 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33226 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33227 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.4327, + "step": 33228 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33229 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33230 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33231 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33232 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.5222, + "step": 33233 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33234 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33235 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33236 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.0117, + "step": 33237 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33238 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.1414, + "step": 33239 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.8504, + "step": 33240 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33241 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.1043, + "step": 33242 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.7388, + "step": 33243 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33244 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33245 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.5745, + "step": 33246 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33247 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.3136, + "step": 33248 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33249 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33250 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.0354, + "step": 33251 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.6225, + "step": 33252 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.0845, + "step": 33253 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33254 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.0534, + "step": 33255 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 8.9415, + "step": 33256 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33257 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33258 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33259 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.6721, + "step": 33260 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33261 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 8.4122, + "step": 33262 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.6999, + "step": 33263 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33264 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.0614, + "step": 33265 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33266 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33267 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.0692, + "step": 33268 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.4176, + "step": 33269 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33270 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33271 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.7779, + "step": 33272 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 7.4591, + "step": 33273 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33274 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.1948, + "step": 33275 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33276 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33277 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 4.28, + "step": 33278 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.0268, + "step": 33279 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33280 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33281 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33282 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.0759, + "step": 33283 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 8.458, + "step": 33284 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33285 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.2251, + "step": 33286 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33287 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.8451, + "step": 33288 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 6.3111, + "step": 33289 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.8561, + "step": 33290 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.1955, + "step": 33291 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 8.9047, + "step": 33292 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33293 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.3334, + "step": 33294 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33295 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.4943, + "step": 33296 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.0235, + "step": 33297 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33298 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.9082, + "step": 33299 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.6694, + "step": 33300 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33301 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.8021, + "step": 33302 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.2663, + "step": 33303 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33304 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.665, + "step": 33305 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.913, + "step": 33306 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33307 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.5318, + "step": 33308 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33309 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 2.3296, + "step": 33310 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33311 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33312 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.0623, + "step": 33313 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33314 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 1.8024, + "step": 33315 + }, + { + "epoch": 2.78, + "learning_rate": 0.0001998495276841196, + "loss": 3.8161, + "step": 33316 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 7.4717, + "step": 33317 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.5399, + "step": 33318 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 4.9543, + "step": 33319 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33320 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33321 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33322 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 2.5935, + "step": 33323 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33324 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33325 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.5134, + "step": 33326 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33327 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33328 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.8877, + "step": 33329 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.3694, + "step": 33330 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 2.1027, + "step": 33331 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33332 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33333 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 5.9144, + "step": 33334 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.1886, + "step": 33335 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.9546, + "step": 33336 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33337 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33338 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.8414, + "step": 33339 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33340 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 4.161, + "step": 33341 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33342 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 5.3661, + "step": 33343 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33344 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 2.2245, + "step": 33345 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33346 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33347 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33348 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33349 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33350 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33351 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33352 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33353 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.9297, + "step": 33354 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33355 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.263, + "step": 33356 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 9.7582, + "step": 33357 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33358 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33359 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 8.0437, + "step": 33360 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.8592, + "step": 33361 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33362 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33363 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.4079, + "step": 33364 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 2.0005, + "step": 33365 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33366 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.8245, + "step": 33367 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 5.7976, + "step": 33368 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33369 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33370 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 2.7031, + "step": 33371 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33372 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 8.2258, + "step": 33373 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.8027, + "step": 33374 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 8.3078, + "step": 33375 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33376 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.6417, + "step": 33377 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33378 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 6.5667, + "step": 33379 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33380 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33381 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33382 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.8018, + "step": 33383 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 5.1019, + "step": 33384 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33385 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33386 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.9517, + "step": 33387 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.3794, + "step": 33388 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33389 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 6.1193, + "step": 33390 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33391 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33392 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 6.2172, + "step": 33393 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33394 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.3856, + "step": 33395 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 6.7067, + "step": 33396 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 2.1171, + "step": 33397 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33398 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33399 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.9374, + "step": 33400 + }, + { + "epoch": 2.79, + "eval_loss": NaN, + "eval_runtime": 39.7592, + "eval_samples_per_second": 62.878, + "eval_steps_per_second": 7.872, + "step": 33400 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33401 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.8605, + "step": 33402 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 7.9581, + "step": 33403 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33404 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33405 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.0109, + "step": 33406 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33407 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 8.1423, + "step": 33408 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33409 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.6836, + "step": 33410 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33411 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 1.8742, + "step": 33412 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 8.8467, + "step": 33413 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33414 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33415 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33416 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.8608, + "step": 33417 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 7.1947, + "step": 33418 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 7.9323, + "step": 33419 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 4.1238, + "step": 33420 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33421 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33422 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33423 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 12.4194, + "step": 33424 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33425 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33426 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 7.2038, + "step": 33427 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.9719, + "step": 33428 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33429 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33430 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 5.0058, + "step": 33431 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.7342, + "step": 33432 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 3.8459, + "step": 33433 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 6.904, + "step": 33434 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 4.1734, + "step": 33435 + }, + { + "epoch": 2.79, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33436 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.6483, + "step": 33437 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 5.2305, + "step": 33438 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 13.4419, + "step": 33439 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.6332, + "step": 33440 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.5754, + "step": 33441 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33442 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33443 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.0678, + "step": 33444 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33445 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.9655, + "step": 33446 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33447 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 9.3314, + "step": 33448 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33449 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.8634, + "step": 33450 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.2206, + "step": 33451 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.9842, + "step": 33452 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 4.2518, + "step": 33453 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 8.5063, + "step": 33454 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33455 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33456 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.7188, + "step": 33457 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33458 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 4.4661, + "step": 33459 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33460 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 8.5947, + "step": 33461 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33462 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33463 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.0848, + "step": 33464 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33465 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.2949, + "step": 33466 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.8337, + "step": 33467 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.3969, + "step": 33468 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.7057, + "step": 33469 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33470 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 5.0688, + "step": 33471 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.2781, + "step": 33472 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.0893, + "step": 33473 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.0901, + "step": 33474 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33475 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33476 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.9389, + "step": 33477 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33478 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.0025, + "step": 33479 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.7014, + "step": 33480 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33481 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.1858, + "step": 33482 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33483 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 4.4026, + "step": 33484 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33485 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33486 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.9441, + "step": 33487 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.6915, + "step": 33488 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33489 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.2052, + "step": 33490 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 7.0654, + "step": 33491 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.9601, + "step": 33492 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 8.2716, + "step": 33493 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33494 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33495 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33496 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 4.4858, + "step": 33497 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33498 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.4356, + "step": 33499 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33500 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.234, + "step": 33501 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33502 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33503 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.4975, + "step": 33504 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.7173, + "step": 33505 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.5224, + "step": 33506 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33507 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.9948, + "step": 33508 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33509 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.8144, + "step": 33510 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 4.6071, + "step": 33511 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.4069, + "step": 33512 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 5.1606, + "step": 33513 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33514 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33515 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33516 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 6.7562, + "step": 33517 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.9759, + "step": 33518 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33519 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33520 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33521 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33522 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33523 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.6694, + "step": 33524 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33525 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 11.5732, + "step": 33526 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33527 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 5.4987, + "step": 33528 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 2.1845, + "step": 33529 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.6772, + "step": 33530 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.0516, + "step": 33531 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33532 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 10.7382, + "step": 33533 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.9943, + "step": 33534 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.964, + "step": 33535 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33536 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33537 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33538 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 4.4171, + "step": 33539 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33540 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33541 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.2196, + "step": 33542 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33543 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33544 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 7.0744, + "step": 33545 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 6.9796, + "step": 33546 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 4.0168, + "step": 33547 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 7.7255, + "step": 33548 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33549 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.8436, + "step": 33550 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33551 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 1.5671, + "step": 33552 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33553 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33554 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33555 + }, + { + "epoch": 2.8, + "learning_rate": 0.0001998495276841196, + "loss": 3.7164, + "step": 33556 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.684, + "step": 33557 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33558 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.943, + "step": 33559 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 11.1527, + "step": 33560 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33561 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33562 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 9.1361, + "step": 33563 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.6702, + "step": 33564 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33565 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33566 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.9268, + "step": 33567 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33568 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 2.2068, + "step": 33569 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.5213, + "step": 33570 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33571 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33572 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33573 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.8013, + "step": 33574 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33575 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 4.5993, + "step": 33576 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 4.7349, + "step": 33577 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 5.5122, + "step": 33578 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33579 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.915, + "step": 33580 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 7.7073, + "step": 33581 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33582 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33583 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33584 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 11.9127, + "step": 33585 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33586 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.9632, + "step": 33587 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 8.6685, + "step": 33588 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 4.1633, + "step": 33589 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33590 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.0097, + "step": 33591 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.1461, + "step": 33592 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33593 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33594 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.044, + "step": 33595 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33596 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33597 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33598 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 7.4195, + "step": 33599 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.6992, + "step": 33600 + }, + { + "epoch": 2.81, + "eval_loss": NaN, + "eval_runtime": 39.8221, + "eval_samples_per_second": 62.779, + "eval_steps_per_second": 7.86, + "step": 33600 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33601 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 2.961, + "step": 33602 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33603 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.5683, + "step": 33604 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33605 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.9238, + "step": 33606 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 8.6357, + "step": 33607 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33608 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.0456, + "step": 33609 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33610 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.9066, + "step": 33611 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33612 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33613 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33614 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33615 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.8385, + "step": 33616 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33617 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.4446, + "step": 33618 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33619 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 6.9134, + "step": 33620 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 7.5124, + "step": 33621 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 5.7695, + "step": 33622 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 4.0141, + "step": 33623 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33624 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 6.9254, + "step": 33625 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33626 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33627 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 5.7745, + "step": 33628 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 6.8675, + "step": 33629 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33630 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33631 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33632 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 2.7455, + "step": 33633 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 2.849, + "step": 33634 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33635 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.6097, + "step": 33636 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 11.314, + "step": 33637 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33638 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33639 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 4.4381, + "step": 33640 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33641 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33642 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 2.3073, + "step": 33643 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33644 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33645 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33646 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 4.7163, + "step": 33647 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.9948, + "step": 33648 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33649 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33650 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 2.1599, + "step": 33651 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33652 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33653 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.4888, + "step": 33654 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33655 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.2459, + "step": 33656 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33657 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 2.6686, + "step": 33658 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33659 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 4.8, + "step": 33660 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 7.4189, + "step": 33661 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.9228, + "step": 33662 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33663 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33664 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 3.6642, + "step": 33665 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33666 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33667 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33668 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.8839, + "step": 33669 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33670 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33671 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 1.1129, + "step": 33672 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33673 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33674 + }, + { + "epoch": 2.81, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33675 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 8.6109, + "step": 33676 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33677 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 8.3577, + "step": 33678 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33679 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.9543, + "step": 33680 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33681 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.5129, + "step": 33682 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.6522, + "step": 33683 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.1187, + "step": 33684 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 8.5612, + "step": 33685 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 5.1323, + "step": 33686 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 6.8489, + "step": 33687 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33688 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 6.9969, + "step": 33689 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 6.0558, + "step": 33690 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.5253, + "step": 33691 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 6.8233, + "step": 33692 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33693 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.5368, + "step": 33694 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33695 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33696 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33697 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 5.4603, + "step": 33698 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.7501, + "step": 33699 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.2021, + "step": 33700 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 8.0366, + "step": 33701 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33702 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33703 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.6389, + "step": 33704 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.8271, + "step": 33705 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.8374, + "step": 33706 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.6792, + "step": 33707 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.0999, + "step": 33708 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 8.831, + "step": 33709 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.373, + "step": 33710 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 6.096, + "step": 33711 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 5.8838, + "step": 33712 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.3561, + "step": 33713 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.8874, + "step": 33714 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.9318, + "step": 33715 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33716 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.6705, + "step": 33717 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.1945, + "step": 33718 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.8413, + "step": 33719 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33720 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.9652, + "step": 33721 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33722 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33723 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.225, + "step": 33724 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33725 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33726 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.6725, + "step": 33727 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33728 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.8068, + "step": 33729 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.7164, + "step": 33730 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33731 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.0113, + "step": 33732 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.0693, + "step": 33733 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.3917, + "step": 33734 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 3.5263, + "step": 33735 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.0425, + "step": 33736 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33737 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.2845, + "step": 33738 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.4019, + "step": 33739 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.5351, + "step": 33740 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33741 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.9969, + "step": 33742 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 9.4622, + "step": 33743 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33744 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 3.8953, + "step": 33745 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.1633, + "step": 33746 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 5.2285, + "step": 33747 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33748 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33749 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.0274, + "step": 33750 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33751 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.624, + "step": 33752 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.877, + "step": 33753 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33754 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 3.6341, + "step": 33755 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33756 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.997, + "step": 33757 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.9216, + "step": 33758 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 4.7446, + "step": 33759 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 6.8547, + "step": 33760 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33761 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33762 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.7687, + "step": 33763 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33764 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33765 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33766 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.0315, + "step": 33767 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33768 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.9758, + "step": 33769 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33770 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 8.4284, + "step": 33771 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33772 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33773 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 5.0712, + "step": 33774 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33775 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.0821, + "step": 33776 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 2.1149, + "step": 33777 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 3.8754, + "step": 33778 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.8882, + "step": 33779 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33780 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.1335, + "step": 33781 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33782 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33783 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33784 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 7.347, + "step": 33785 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 1.0426, + "step": 33786 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 6.0055, + "step": 33787 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 3.6794, + "step": 33788 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33789 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33790 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33791 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 5.1682, + "step": 33792 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 5.8103, + "step": 33793 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 3.9385, + "step": 33794 + }, + { + "epoch": 2.82, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33795 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33796 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 1.8641, + "step": 33797 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 1.8673, + "step": 33798 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33799 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 3.2985, + "step": 33800 + }, + { + "epoch": 2.83, + "eval_loss": NaN, + "eval_runtime": 39.84, + "eval_samples_per_second": 62.751, + "eval_steps_per_second": 7.856, + "step": 33800 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 2.1317, + "step": 33801 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 1.935, + "step": 33802 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 6.7888, + "step": 33803 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33804 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33805 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 8.2131, + "step": 33806 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 2.1926, + "step": 33807 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 6.9358, + "step": 33808 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 1.8293, + "step": 33809 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 2.1057, + "step": 33810 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33811 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 0.9941, + "step": 33812 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998495276841196, + "loss": 0.0, + "step": 33813 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.0389, + "step": 33814 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 5.7888, + "step": 33815 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 3.6317, + "step": 33816 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33817 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33818 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 7.1638, + "step": 33819 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 2.7625, + "step": 33820 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33821 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33822 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.902, + "step": 33823 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 6.2746, + "step": 33824 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33825 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33826 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33827 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33828 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.0334, + "step": 33829 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33830 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33831 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 7.1345, + "step": 33832 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33833 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.7344, + "step": 33834 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.8476, + "step": 33835 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 8.2954, + "step": 33836 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.8365, + "step": 33837 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 7.8905, + "step": 33838 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33839 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33840 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33841 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33842 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33843 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.1382, + "step": 33844 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.4067, + "step": 33845 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 8.0549, + "step": 33846 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33847 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.6218, + "step": 33848 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 2.2107, + "step": 33849 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.7774, + "step": 33850 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33851 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.1478, + "step": 33852 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 6.3651, + "step": 33853 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.8826, + "step": 33854 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33855 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33856 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 3.4989, + "step": 33857 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 8.2158, + "step": 33858 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33859 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.8298, + "step": 33860 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 2.0576, + "step": 33861 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 2.384, + "step": 33862 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 8.3226, + "step": 33863 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33864 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33865 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 7.8434, + "step": 33866 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33867 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.8302, + "step": 33868 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 2.1095, + "step": 33869 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33870 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33871 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 6.6092, + "step": 33872 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33873 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 5.5646, + "step": 33874 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33875 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 5.836, + "step": 33876 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33877 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33878 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 3.2141, + "step": 33879 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.1343, + "step": 33880 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33881 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.0079, + "step": 33882 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.6092, + "step": 33883 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 7.2606, + "step": 33884 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 8.61, + "step": 33885 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 9.2297, + "step": 33886 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 2.1116, + "step": 33887 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33888 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.8723, + "step": 33889 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33890 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.9444, + "step": 33891 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.7665, + "step": 33892 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 9.145, + "step": 33893 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.1221, + "step": 33894 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.375, + "step": 33895 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33896 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.8107, + "step": 33897 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33898 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 2.296, + "step": 33899 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33900 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33901 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33902 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.3571, + "step": 33903 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 4.3971, + "step": 33904 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33905 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 1.0908, + "step": 33906 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33907 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33908 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33909 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33910 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 7.4735, + "step": 33911 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33912 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 10.5341, + "step": 33913 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33914 + }, + { + "epoch": 2.83, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33915 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 11.6211, + "step": 33916 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.0153, + "step": 33917 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.6762, + "step": 33918 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.7585, + "step": 33919 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.5634, + "step": 33920 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.8551, + "step": 33921 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33922 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 5.7512, + "step": 33923 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33924 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33925 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33926 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.2303, + "step": 33927 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33928 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.6099, + "step": 33929 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.4049, + "step": 33930 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33931 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.6722, + "step": 33932 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.3012, + "step": 33933 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33934 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33935 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 7.4914, + "step": 33936 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.0028, + "step": 33937 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.0484, + "step": 33938 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33939 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.6517, + "step": 33940 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33941 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33942 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.0879, + "step": 33943 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.8999, + "step": 33944 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.7851, + "step": 33945 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 8.9631, + "step": 33946 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33947 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 7.0299, + "step": 33948 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.0375, + "step": 33949 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.0986, + "step": 33950 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33951 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33952 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.8874, + "step": 33953 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 7.8489, + "step": 33954 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33955 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.8892, + "step": 33956 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33957 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33958 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 8.6032, + "step": 33959 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33960 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33961 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.7544, + "step": 33962 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.7499, + "step": 33963 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33964 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33965 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.8738, + "step": 33966 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33967 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.2582, + "step": 33968 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33969 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33970 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33971 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33972 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 7.2933, + "step": 33973 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33974 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33975 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33976 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.9757, + "step": 33977 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.1901, + "step": 33978 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.7802, + "step": 33979 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.3669, + "step": 33980 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.2079, + "step": 33981 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.0557, + "step": 33982 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.1235, + "step": 33983 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 8.1819, + "step": 33984 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 5.9149, + "step": 33985 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33986 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.7813, + "step": 33987 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33988 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33989 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33990 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.5996, + "step": 33991 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.9117, + "step": 33992 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.9871, + "step": 33993 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33994 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33995 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.0061, + "step": 33996 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33997 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 7.0776, + "step": 33998 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 33999 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34000 + }, + { + "epoch": 2.84, + "eval_loss": NaN, + "eval_runtime": 39.8832, + "eval_samples_per_second": 62.683, + "eval_steps_per_second": 7.848, + "step": 34000 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.155, + "step": 34001 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34002 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.7458, + "step": 34003 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.7655, + "step": 34004 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 3.4279, + "step": 34005 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34006 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 8.0807, + "step": 34007 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.2638, + "step": 34008 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.8869, + "step": 34009 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34010 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.1964, + "step": 34011 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.3132, + "step": 34012 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34013 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 5.4833, + "step": 34014 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 5.3907, + "step": 34015 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.9273, + "step": 34016 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34017 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34018 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34019 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.8452, + "step": 34020 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34021 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 5.4503, + "step": 34022 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 8.092, + "step": 34023 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 4.1865, + "step": 34024 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 2.7423, + "step": 34025 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34026 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34027 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.521, + "step": 34028 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34029 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 8.8199, + "step": 34030 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 5.0412, + "step": 34031 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34032 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 1.626, + "step": 34033 + }, + { + "epoch": 2.84, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34034 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34035 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34036 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.8973, + "step": 34037 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34038 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.7949, + "step": 34039 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34040 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 2.0554, + "step": 34041 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34042 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34043 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34044 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.9077, + "step": 34045 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34046 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34047 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34048 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34049 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 14.883, + "step": 34050 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34051 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34052 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.2733, + "step": 34053 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34054 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34055 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34056 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.7254, + "step": 34057 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 4.7515, + "step": 34058 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 4.2866, + "step": 34059 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34060 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 6.0733, + "step": 34061 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 7.4041, + "step": 34062 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34063 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.597, + "step": 34064 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34065 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 8.4947, + "step": 34066 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 7.1339, + "step": 34067 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 6.7246, + "step": 34068 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34069 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 4.3152, + "step": 34070 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 7.2738, + "step": 34071 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 8.3556, + "step": 34072 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34073 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34074 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34075 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.753, + "step": 34076 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34077 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34078 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 2.0148, + "step": 34079 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.0949, + "step": 34080 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34081 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34082 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 7.1553, + "step": 34083 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34084 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 5.1035, + "step": 34085 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 9.3296, + "step": 34086 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 8.3782, + "step": 34087 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 6.0745, + "step": 34088 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34089 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34090 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.7922, + "step": 34091 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34092 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.9256, + "step": 34093 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34094 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 2.1007, + "step": 34095 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.0095, + "step": 34096 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.8545, + "step": 34097 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 8.2943, + "step": 34098 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.9064, + "step": 34099 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.8585, + "step": 34100 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34101 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34102 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 9.2275, + "step": 34103 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 5.1965, + "step": 34104 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.849, + "step": 34105 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34106 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34107 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.8521, + "step": 34108 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34109 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34110 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34111 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 4.9464, + "step": 34112 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34113 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 8.7651, + "step": 34114 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 2.6769, + "step": 34115 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34116 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.7939, + "step": 34117 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34118 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 8.2706, + "step": 34119 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34120 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.0218, + "step": 34121 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 5.6086, + "step": 34122 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34123 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 8.5988, + "step": 34124 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.9381, + "step": 34125 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.526, + "step": 34126 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34127 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34128 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34129 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34130 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34131 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34132 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.0619, + "step": 34133 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 2.2167, + "step": 34134 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.6165, + "step": 34135 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34136 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34137 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 4.8453, + "step": 34138 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34139 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34140 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.7386, + "step": 34141 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34142 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34143 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 1.1342, + "step": 34144 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34145 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34146 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 14.6917, + "step": 34147 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 3.5106, + "step": 34148 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34149 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34150 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34151 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.8945, + "step": 34152 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34153 + }, + { + "epoch": 2.85, + "learning_rate": 0.0001998439546353833, + "loss": 2.4259, + "step": 34154 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 2.1438, + "step": 34155 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34156 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.1367, + "step": 34157 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34158 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34159 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34160 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34161 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.1067, + "step": 34162 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34163 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.5721, + "step": 34164 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 3.5177, + "step": 34165 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34166 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.174, + "step": 34167 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34168 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 5.9472, + "step": 34169 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34170 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 3.519, + "step": 34171 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 5.204, + "step": 34172 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34173 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 10.2261, + "step": 34174 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34175 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.1764, + "step": 34176 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34177 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.1184, + "step": 34178 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 5.932, + "step": 34179 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34180 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34181 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34182 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34183 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 2.1455, + "step": 34184 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.796, + "step": 34185 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.2717, + "step": 34186 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34187 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.9736, + "step": 34188 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 2.5354, + "step": 34189 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 2.9896, + "step": 34190 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.9073, + "step": 34191 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.0108, + "step": 34192 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 8.0871, + "step": 34193 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.9561, + "step": 34194 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 3.6623, + "step": 34195 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.8159, + "step": 34196 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 6.7335, + "step": 34197 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34198 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34199 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34200 + }, + { + "epoch": 2.86, + "eval_loss": NaN, + "eval_runtime": 39.9179, + "eval_samples_per_second": 62.629, + "eval_steps_per_second": 7.841, + "step": 34200 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 7.7552, + "step": 34201 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 6.9445, + "step": 34202 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 3.718, + "step": 34203 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 8.9915, + "step": 34204 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.977, + "step": 34205 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34206 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34207 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.6123, + "step": 34208 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.6885, + "step": 34209 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34210 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34211 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.2408, + "step": 34212 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.1241, + "step": 34213 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 8.5612, + "step": 34214 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34215 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 6.5009, + "step": 34216 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34217 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34218 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.4406, + "step": 34219 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34220 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34221 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34222 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34223 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34224 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.1043, + "step": 34225 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 8.2162, + "step": 34226 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.0588, + "step": 34227 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 5.4348, + "step": 34228 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34229 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 5.0683, + "step": 34230 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 9.1607, + "step": 34231 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 6.6844, + "step": 34232 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34233 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34234 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34235 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34236 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34237 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 8.0896, + "step": 34238 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 9.3833, + "step": 34239 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34240 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 2.5154, + "step": 34241 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34242 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34243 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 8.4152, + "step": 34244 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.1962, + "step": 34245 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34246 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34247 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.8161, + "step": 34248 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34249 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34250 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 3.5333, + "step": 34251 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 7.967, + "step": 34252 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 6.9367, + "step": 34253 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.5842, + "step": 34254 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34255 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34256 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 6.3647, + "step": 34257 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34258 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34259 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 2.9594, + "step": 34260 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 3.1423, + "step": 34261 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34262 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34263 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34264 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34265 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 10.5672, + "step": 34266 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.6254, + "step": 34267 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 3.8214, + "step": 34268 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 1.0266, + "step": 34269 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.2919, + "step": 34270 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 4.3924, + "step": 34271 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34272 + }, + { + "epoch": 2.86, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34273 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34274 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.6572, + "step": 34275 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.7319, + "step": 34276 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.8739, + "step": 34277 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 5.0935, + "step": 34278 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.97, + "step": 34279 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.9106, + "step": 34280 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.7696, + "step": 34281 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.0324, + "step": 34282 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 2.0788, + "step": 34283 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 8.2148, + "step": 34284 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34285 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 5.6499, + "step": 34286 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 7.5743, + "step": 34287 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34288 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.9875, + "step": 34289 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.7119, + "step": 34290 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.04, + "step": 34291 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 2.0359, + "step": 34292 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34293 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34294 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 2.3331, + "step": 34295 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34296 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.6031, + "step": 34297 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.0609, + "step": 34298 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34299 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34300 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.1603, + "step": 34301 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.1843, + "step": 34302 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 2.4314, + "step": 34303 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 8.3692, + "step": 34304 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.9691, + "step": 34305 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34306 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 2.7945, + "step": 34307 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34308 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.7837, + "step": 34309 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34310 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34311 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.9039, + "step": 34312 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34313 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34314 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 9.2869, + "step": 34315 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34316 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.1359, + "step": 34317 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34318 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.9296, + "step": 34319 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 6.2166, + "step": 34320 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.4622, + "step": 34321 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34322 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34323 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.1092, + "step": 34324 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34325 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34326 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34327 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.0003, + "step": 34328 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 8.3174, + "step": 34329 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34330 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.6826, + "step": 34331 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34332 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34333 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34334 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.7704, + "step": 34335 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34336 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.7356, + "step": 34337 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34338 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34339 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.8349, + "step": 34340 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 9.3112, + "step": 34341 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34342 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 2.7847, + "step": 34343 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34344 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 12.157, + "step": 34345 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34346 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.6285, + "step": 34347 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.8119, + "step": 34348 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.8252, + "step": 34349 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.9752, + "step": 34350 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.8735, + "step": 34351 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.2425, + "step": 34352 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34353 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 7.872, + "step": 34354 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34355 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34356 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34357 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34358 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 5.5115, + "step": 34359 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 8.965, + "step": 34360 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 1.0534, + "step": 34361 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.1845, + "step": 34362 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 7.3875, + "step": 34363 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 7.8987, + "step": 34364 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 4.8328, + "step": 34365 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.5952, + "step": 34366 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.3029, + "step": 34367 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 6.4437, + "step": 34368 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34369 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34370 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 2.1353, + "step": 34371 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 3.9934, + "step": 34372 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.7367, + "step": 34373 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 0.0, + "step": 34374 + }, + { + "epoch": 2.87, + "learning_rate": 0.0001998439546353833, + "loss": 9.0135, + "step": 34375 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 3.5427, + "step": 34376 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34377 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34378 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 8.1525, + "step": 34379 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 0.9875, + "step": 34380 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 1.982, + "step": 34381 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 2.2415, + "step": 34382 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 3.0112, + "step": 34383 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 1.5452, + "step": 34384 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34385 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 2.0861, + "step": 34386 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 1.2015, + "step": 34387 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 4.1668, + "step": 34388 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 1.9585, + "step": 34389 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34390 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 1.7432, + "step": 34391 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34392 + }, + { + "epoch": 2.87, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34393 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34394 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34395 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 8.8354, + "step": 34396 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34397 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34398 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.832, + "step": 34399 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34400 + }, + { + "epoch": 2.88, + "eval_loss": NaN, + "eval_runtime": 39.9427, + "eval_samples_per_second": 62.59, + "eval_steps_per_second": 7.836, + "step": 34400 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.9754, + "step": 34401 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34402 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34403 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34404 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34405 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.9845, + "step": 34406 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 3.1947, + "step": 34407 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34408 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34409 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 3.1827, + "step": 34410 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.3699, + "step": 34411 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34412 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34413 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.2195, + "step": 34414 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.8722, + "step": 34415 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34416 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34417 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.501, + "step": 34418 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 8.4068, + "step": 34419 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 7.5513, + "step": 34420 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34421 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.858, + "step": 34422 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.8206, + "step": 34423 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34424 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34425 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34426 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.6936, + "step": 34427 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.9918, + "step": 34428 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.4532, + "step": 34429 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34430 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34431 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34432 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34433 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34434 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.8384, + "step": 34435 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34436 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 7.5608, + "step": 34437 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34438 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.9429, + "step": 34439 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.2983, + "step": 34440 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34441 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34442 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.8004, + "step": 34443 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34444 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 14.4537, + "step": 34445 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34446 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.1726, + "step": 34447 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34448 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 3.7434, + "step": 34449 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.2131, + "step": 34450 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.6709, + "step": 34451 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34452 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34453 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.2126, + "step": 34454 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34455 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34456 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34457 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34458 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34459 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34460 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.9147, + "step": 34461 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34462 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.2039, + "step": 34463 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34464 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34465 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 3.5389, + "step": 34466 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34467 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34468 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34469 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.2077, + "step": 34470 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.1779, + "step": 34471 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34472 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34473 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.499, + "step": 34474 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.3651, + "step": 34475 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 8.0113, + "step": 34476 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 7.8358, + "step": 34477 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.8268, + "step": 34478 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34479 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.7907, + "step": 34480 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.8764, + "step": 34481 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 8.6579, + "step": 34482 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34483 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.1357, + "step": 34484 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34485 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 6.8403, + "step": 34486 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 7.8008, + "step": 34487 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 6.6586, + "step": 34488 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 4.3485, + "step": 34489 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34490 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.9323, + "step": 34491 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.7181, + "step": 34492 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34493 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34494 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34495 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34496 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.9254, + "step": 34497 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 8.486, + "step": 34498 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34499 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.8238, + "step": 34500 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34501 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 9.3824, + "step": 34502 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.0186, + "step": 34503 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34504 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34505 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34506 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34507 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 2.4117, + "step": 34508 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 8.6466, + "step": 34509 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34510 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 0.9887, + "step": 34511 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 1.1035, + "step": 34512 + }, + { + "epoch": 2.88, + "learning_rate": 0.00019983838158664697, + "loss": 6.6802, + "step": 34513 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34514 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.8188, + "step": 34515 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.8318, + "step": 34516 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34517 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.9223, + "step": 34518 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 2.3424, + "step": 34519 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 7.6911, + "step": 34520 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34521 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 8.3612, + "step": 34522 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 5.0196, + "step": 34523 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34524 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.8325, + "step": 34525 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.8561, + "step": 34526 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.7592, + "step": 34527 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.5282, + "step": 34528 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.9716, + "step": 34529 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.2961, + "step": 34530 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 7.4518, + "step": 34531 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34532 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 2.3288, + "step": 34533 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.0917, + "step": 34534 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34535 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34536 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34537 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.2186, + "step": 34538 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.6247, + "step": 34539 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.9261, + "step": 34540 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.8774, + "step": 34541 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 2.1232, + "step": 34542 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 9.1013, + "step": 34543 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.9505, + "step": 34544 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34545 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34546 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34547 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34548 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.8217, + "step": 34549 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.002, + "step": 34550 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34551 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34552 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.7978, + "step": 34553 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34554 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34555 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34556 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34557 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34558 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34559 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34560 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34561 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34562 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.916, + "step": 34563 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 8.4099, + "step": 34564 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34565 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34566 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.0068, + "step": 34567 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34568 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 2.7986, + "step": 34569 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34570 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.935, + "step": 34571 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34572 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.4309, + "step": 34573 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.8238, + "step": 34574 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34575 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.4003, + "step": 34576 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34577 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34578 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34579 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34580 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 7.1173, + "step": 34581 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34582 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.0436, + "step": 34583 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 5.164, + "step": 34584 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34585 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34586 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.3005, + "step": 34587 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.006, + "step": 34588 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34589 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34590 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 2.649, + "step": 34591 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 6.0872, + "step": 34592 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.0046, + "step": 34593 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.2732, + "step": 34594 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.9312, + "step": 34595 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34596 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 5.1024, + "step": 34597 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34598 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 8.8922, + "step": 34599 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.4852, + "step": 34600 + }, + { + "epoch": 2.89, + "eval_loss": NaN, + "eval_runtime": 39.8868, + "eval_samples_per_second": 62.677, + "eval_steps_per_second": 7.847, + "step": 34600 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.8656, + "step": 34601 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.8785, + "step": 34602 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.9112, + "step": 34603 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34604 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 7.7432, + "step": 34605 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 8.049, + "step": 34606 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34607 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 9.4289, + "step": 34608 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34609 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34610 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.0682, + "step": 34611 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 6.7139, + "step": 34612 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.6608, + "step": 34613 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.3182, + "step": 34614 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.4827, + "step": 34615 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.9404, + "step": 34616 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34617 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34618 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 9.2232, + "step": 34619 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 4.4217, + "step": 34620 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 2.6663, + "step": 34621 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34622 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34623 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.395, + "step": 34624 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 3.8436, + "step": 34625 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34626 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34627 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34628 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34629 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34630 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 1.6567, + "step": 34631 + }, + { + "epoch": 2.89, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34632 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34633 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.967, + "step": 34634 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.1914, + "step": 34635 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 7.2209, + "step": 34636 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34637 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 6.7151, + "step": 34638 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.1006, + "step": 34639 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34640 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 7.7218, + "step": 34641 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34642 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34643 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34644 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34645 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.996, + "step": 34646 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 7.3702, + "step": 34647 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.9504, + "step": 34648 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34649 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34650 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.3388, + "step": 34651 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34652 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 3.759, + "step": 34653 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34654 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34655 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 3.7712, + "step": 34656 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34657 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.9768, + "step": 34658 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.3685, + "step": 34659 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34660 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 8.7941, + "step": 34661 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 3.9346, + "step": 34662 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34663 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.4234, + "step": 34664 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34665 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 3.1152, + "step": 34666 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34667 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 6.3323, + "step": 34668 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34669 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.8298, + "step": 34670 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.1909, + "step": 34671 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34672 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34673 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 8.7395, + "step": 34674 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 8.9916, + "step": 34675 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 2.2601, + "step": 34676 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34677 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34678 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34679 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.3421, + "step": 34680 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34681 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34682 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 5.2313, + "step": 34683 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34684 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34685 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 2.1841, + "step": 34686 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.6603, + "step": 34687 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34688 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34689 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.0743, + "step": 34690 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34691 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 9.5433, + "step": 34692 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34693 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34694 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 3.0517, + "step": 34695 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34696 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34697 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34698 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34699 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34700 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.0654, + "step": 34701 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34702 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34703 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.3503, + "step": 34704 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34705 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.0164, + "step": 34706 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34707 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.9756, + "step": 34708 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.6047, + "step": 34709 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34710 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.0879, + "step": 34711 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34712 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34713 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 6.3677, + "step": 34714 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 5.6681, + "step": 34715 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34716 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 7.6031, + "step": 34717 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 2.6919, + "step": 34718 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.783, + "step": 34719 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34720 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 8.6042, + "step": 34721 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34722 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34723 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.9481, + "step": 34724 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 5.2531, + "step": 34725 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.3318, + "step": 34726 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 6.5864, + "step": 34727 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34728 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 3.8617, + "step": 34729 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.6958, + "step": 34730 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34731 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34732 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 7.5046, + "step": 34733 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.1495, + "step": 34734 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.1344, + "step": 34735 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34736 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34737 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 5.0803, + "step": 34738 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34739 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 3.1032, + "step": 34740 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.8969, + "step": 34741 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 6.0918, + "step": 34742 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 1.1567, + "step": 34743 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34744 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 2.3635, + "step": 34745 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34746 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 4.1501, + "step": 34747 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 7.6539, + "step": 34748 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34749 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34750 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 11.6018, + "step": 34751 + }, + { + "epoch": 2.9, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34752 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34753 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.8349, + "step": 34754 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.9415, + "step": 34755 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.9504, + "step": 34756 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.9821, + "step": 34757 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.7754, + "step": 34758 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34759 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34760 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 5.3094, + "step": 34761 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 8.0424, + "step": 34762 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.0644, + "step": 34763 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 8.2576, + "step": 34764 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34765 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.0508, + "step": 34766 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34767 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.5768, + "step": 34768 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34769 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.738, + "step": 34770 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34771 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.9795, + "step": 34772 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34773 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.243, + "step": 34774 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 7.6956, + "step": 34775 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.1095, + "step": 34776 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34777 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 4.1495, + "step": 34778 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 6.6866, + "step": 34779 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34780 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 10.6984, + "step": 34781 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.0492, + "step": 34782 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.1047, + "step": 34783 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.6865, + "step": 34784 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.9601, + "step": 34785 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34786 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 5.1002, + "step": 34787 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34788 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.7943, + "step": 34789 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 9.226, + "step": 34790 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34791 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34792 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34793 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 5.6762, + "step": 34794 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 7.7662, + "step": 34795 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34796 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.9296, + "step": 34797 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34798 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34799 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34800 + }, + { + "epoch": 2.91, + "eval_loss": NaN, + "eval_runtime": 39.8806, + "eval_samples_per_second": 62.687, + "eval_steps_per_second": 7.848, + "step": 34800 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 8.7392, + "step": 34801 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34802 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34803 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34804 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34805 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 4.7258, + "step": 34806 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.7257, + "step": 34807 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 4.0285, + "step": 34808 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34809 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.2708, + "step": 34810 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34811 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34812 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 6.5172, + "step": 34813 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.125, + "step": 34814 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.855, + "step": 34815 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.8504, + "step": 34816 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34817 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.0223, + "step": 34818 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.2182, + "step": 34819 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 10.2194, + "step": 34820 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.8984, + "step": 34821 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34822 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.8535, + "step": 34823 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.9111, + "step": 34824 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.9412, + "step": 34825 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34826 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.8933, + "step": 34827 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34828 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.0381, + "step": 34829 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34830 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.9432, + "step": 34831 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34832 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34833 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.6465, + "step": 34834 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34835 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 5.3453, + "step": 34836 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 4.6455, + "step": 34837 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.8214, + "step": 34838 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34839 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34840 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.837, + "step": 34841 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 4.2614, + "step": 34842 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34843 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 4.4137, + "step": 34844 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.1558, + "step": 34845 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34846 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.9539, + "step": 34847 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.7061, + "step": 34848 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.2306, + "step": 34849 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34850 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.8046, + "step": 34851 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.4229, + "step": 34852 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.6269, + "step": 34853 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34854 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 7.5794, + "step": 34855 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34856 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34857 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34858 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34859 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.0201, + "step": 34860 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 2.2229, + "step": 34861 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34862 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 3.6819, + "step": 34863 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.7335, + "step": 34864 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34865 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.5325, + "step": 34866 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34867 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 1.9081, + "step": 34868 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.8276, + "step": 34869 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34870 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34871 + }, + { + "epoch": 2.91, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34872 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34873 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.1814, + "step": 34874 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34875 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34876 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 5.0546, + "step": 34877 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 8.2437, + "step": 34878 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 7.189, + "step": 34879 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34880 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34881 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 7.0161, + "step": 34882 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.1479, + "step": 34883 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34884 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34885 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.1906, + "step": 34886 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.9129, + "step": 34887 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34888 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34889 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.9344, + "step": 34890 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 5.8595, + "step": 34891 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34892 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.8409, + "step": 34893 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.3893, + "step": 34894 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34895 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.7601, + "step": 34896 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34897 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.6215, + "step": 34898 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34899 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 5.1559, + "step": 34900 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.91, + "step": 34901 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.0847, + "step": 34902 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.1932, + "step": 34903 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.2564, + "step": 34904 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34905 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34906 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34907 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.9434, + "step": 34908 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34909 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.8662, + "step": 34910 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34911 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.0687, + "step": 34912 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34913 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.2023, + "step": 34914 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34915 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34916 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34917 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.8972, + "step": 34918 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.8253, + "step": 34919 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 5.5987, + "step": 34920 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34921 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.7763, + "step": 34922 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 7.3662, + "step": 34923 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 6.936, + "step": 34924 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.9827, + "step": 34925 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.1627, + "step": 34926 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34927 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.7088, + "step": 34928 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.4662, + "step": 34929 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34930 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 8.2485, + "step": 34931 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.9967, + "step": 34932 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34933 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34934 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34935 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.4337, + "step": 34936 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34937 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.0244, + "step": 34938 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34939 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34940 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.4537, + "step": 34941 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.7226, + "step": 34942 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34943 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34944 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.165, + "step": 34945 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.6232, + "step": 34946 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.0494, + "step": 34947 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.154, + "step": 34948 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34949 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 5.6927, + "step": 34950 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34951 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34952 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34953 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.1216, + "step": 34954 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.0242, + "step": 34955 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 4.133, + "step": 34956 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34957 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 1.9248, + "step": 34958 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.763, + "step": 34959 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 6.461, + "step": 34960 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.8883, + "step": 34961 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34962 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34963 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 6.1753, + "step": 34964 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34965 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.7574, + "step": 34966 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34967 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34968 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.1333, + "step": 34969 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 6.4575, + "step": 34970 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34971 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.4819, + "step": 34972 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.0556, + "step": 34973 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34974 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34975 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 6.8989, + "step": 34976 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34977 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 8.9556, + "step": 34978 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 2.9155, + "step": 34979 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.6105, + "step": 34980 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34981 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34982 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 8.1336, + "step": 34983 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34984 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.3865, + "step": 34985 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34986 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34987 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 3.3902, + "step": 34988 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34989 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 5.4914, + "step": 34990 + }, + { + "epoch": 2.92, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34991 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34992 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 4.4939, + "step": 34993 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34994 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 9.5848, + "step": 34995 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 34996 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.874, + "step": 34997 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.1811, + "step": 34998 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 5.9971, + "step": 34999 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.9297, + "step": 35000 + }, + { + "epoch": 2.93, + "eval_loss": NaN, + "eval_runtime": 39.8996, + "eval_samples_per_second": 62.657, + "eval_steps_per_second": 7.845, + "step": 35000 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35001 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35002 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.9098, + "step": 35003 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.6321, + "step": 35004 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 7.4815, + "step": 35005 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.9659, + "step": 35006 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35007 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35008 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.4616, + "step": 35009 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35010 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 5.1099, + "step": 35011 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35012 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35013 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35014 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.5281, + "step": 35015 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35016 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35017 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.6715, + "step": 35018 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.8728, + "step": 35019 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.4193, + "step": 35020 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35021 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.2832, + "step": 35022 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35023 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.8183, + "step": 35024 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.1071, + "step": 35025 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35026 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35027 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.4697, + "step": 35028 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.9119, + "step": 35029 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35030 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35031 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35032 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.78, + "step": 35033 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.9321, + "step": 35034 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 7.7771, + "step": 35035 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.7967, + "step": 35036 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35037 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.937, + "step": 35038 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.8671, + "step": 35039 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35040 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.0758, + "step": 35041 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.0684, + "step": 35042 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.8628, + "step": 35043 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.9513, + "step": 35044 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 11.3689, + "step": 35045 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 7.4805, + "step": 35046 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.6338, + "step": 35047 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 4.1658, + "step": 35048 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 8.1795, + "step": 35049 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.5879, + "step": 35050 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 8.1063, + "step": 35051 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 5.5568, + "step": 35052 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35053 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.6491, + "step": 35054 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 8.6077, + "step": 35055 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 9.8096, + "step": 35056 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35057 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35058 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35059 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 5.3722, + "step": 35060 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 4.1661, + "step": 35061 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.9488, + "step": 35062 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.0304, + "step": 35063 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35064 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35065 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35066 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35067 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 6.2828, + "step": 35068 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35069 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.0893, + "step": 35070 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.0193, + "step": 35071 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.1352, + "step": 35072 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 2.0957, + "step": 35073 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 6.8175, + "step": 35074 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35075 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35076 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35077 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 4.2414, + "step": 35078 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.9783, + "step": 35079 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.1676, + "step": 35080 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.6249, + "step": 35081 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.9799, + "step": 35082 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35083 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35084 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 6.3335, + "step": 35085 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35086 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.9166, + "step": 35087 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35088 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35089 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35090 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 7.5143, + "step": 35091 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35092 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.6597, + "step": 35093 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 7.8839, + "step": 35094 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35095 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 3.8936, + "step": 35096 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.9039, + "step": 35097 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.832, + "step": 35098 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35099 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35100 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 4.3085, + "step": 35101 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35102 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.9952, + "step": 35103 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35104 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 10.6997, + "step": 35105 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 8.3172, + "step": 35106 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 7.3189, + "step": 35107 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35108 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 1.0558, + "step": 35109 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 9.6309, + "step": 35110 + }, + { + "epoch": 2.93, + "learning_rate": 0.00019983838158664697, + "loss": 7.8154, + "step": 35111 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35112 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 10.594, + "step": 35113 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35114 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35115 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35116 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35117 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.4781, + "step": 35118 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.2122, + "step": 35119 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 7.1142, + "step": 35120 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.9465, + "step": 35121 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.9127, + "step": 35122 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.0677, + "step": 35123 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35124 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.9124, + "step": 35125 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35126 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.6213, + "step": 35127 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.3638, + "step": 35128 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35129 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 8.8144, + "step": 35130 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35131 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35132 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.4494, + "step": 35133 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35134 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.4855, + "step": 35135 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35136 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.8802, + "step": 35137 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35138 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35139 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35140 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.0455, + "step": 35141 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.866, + "step": 35142 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35143 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 8.5811, + "step": 35144 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35145 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.1373, + "step": 35146 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.617, + "step": 35147 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.0592, + "step": 35148 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.8492, + "step": 35149 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35150 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 5.7022, + "step": 35151 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 5.3196, + "step": 35152 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35153 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 6.4556, + "step": 35154 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35155 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.0984, + "step": 35156 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.1131, + "step": 35157 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.0927, + "step": 35158 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35159 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35160 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35161 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35162 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35163 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.9524, + "step": 35164 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35165 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35166 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.4096, + "step": 35167 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35168 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35169 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 8.0235, + "step": 35170 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.1514, + "step": 35171 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.1299, + "step": 35172 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 5.6313, + "step": 35173 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35174 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35175 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.8115, + "step": 35176 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.3379, + "step": 35177 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.3722, + "step": 35178 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.9738, + "step": 35179 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.9355, + "step": 35180 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35181 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.5227, + "step": 35182 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.7768, + "step": 35183 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 8.3659, + "step": 35184 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 5.0965, + "step": 35185 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.9674, + "step": 35186 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.0361, + "step": 35187 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 8.1945, + "step": 35188 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 6.9022, + "step": 35189 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35190 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35191 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.3524, + "step": 35192 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35193 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.5777, + "step": 35194 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.7705, + "step": 35195 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.3816, + "step": 35196 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35197 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35198 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.7246, + "step": 35199 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35200 + }, + { + "epoch": 2.94, + "eval_loss": NaN, + "eval_runtime": 39.8871, + "eval_samples_per_second": 62.677, + "eval_steps_per_second": 7.847, + "step": 35200 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35201 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35202 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.9932, + "step": 35203 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.0161, + "step": 35204 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 6.7568, + "step": 35205 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 8.9453, + "step": 35206 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.1165, + "step": 35207 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35208 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35209 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 7.1614, + "step": 35210 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.7376, + "step": 35211 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 8.2629, + "step": 35212 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35213 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.094, + "step": 35214 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 4.0082, + "step": 35215 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35216 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35217 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.2751, + "step": 35218 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35219 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 3.9663, + "step": 35220 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35221 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 7.4012, + "step": 35222 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 9.2621, + "step": 35223 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35224 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35225 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 5.9507, + "step": 35226 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35227 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35228 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35229 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 1.9964, + "step": 35230 + }, + { + "epoch": 2.94, + "learning_rate": 0.00019983838158664697, + "loss": 2.386, + "step": 35231 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.5717, + "step": 35232 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.3882, + "step": 35233 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35234 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 5.7551, + "step": 35235 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 7.9285, + "step": 35236 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35237 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35238 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35239 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.0442, + "step": 35240 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35241 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35242 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.0309, + "step": 35243 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35244 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35245 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 4.6958, + "step": 35246 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35247 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.044, + "step": 35248 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35249 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35250 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35251 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 8.1208, + "step": 35252 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 5.5335, + "step": 35253 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35254 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35255 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 4.0315, + "step": 35256 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.784, + "step": 35257 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.2399, + "step": 35258 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 7.8215, + "step": 35259 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 8.0749, + "step": 35260 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 4.6432, + "step": 35261 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.0128, + "step": 35262 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.8405, + "step": 35263 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 9.4927, + "step": 35264 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35265 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35266 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.5139, + "step": 35267 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35268 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.3402, + "step": 35269 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35270 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35271 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35272 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35273 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35274 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.8766, + "step": 35275 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 4.3854, + "step": 35276 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35277 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.0435, + "step": 35278 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35279 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.9818, + "step": 35280 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35281 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.9165, + "step": 35282 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.1001, + "step": 35283 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 7.9679, + "step": 35284 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35285 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 9.6474, + "step": 35286 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35287 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 9.2387, + "step": 35288 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 7.5773, + "step": 35289 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 14.7403, + "step": 35290 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.6855, + "step": 35291 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35292 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35293 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.8555, + "step": 35294 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.6584, + "step": 35295 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 8.1447, + "step": 35296 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35297 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.3711, + "step": 35298 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35299 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 8.4792, + "step": 35300 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.9773, + "step": 35301 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35302 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35303 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.0176, + "step": 35304 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.0295, + "step": 35305 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35306 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.9136, + "step": 35307 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.1525, + "step": 35308 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.8142, + "step": 35309 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 8.4244, + "step": 35310 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35311 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.9748, + "step": 35312 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35313 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35314 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35315 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35316 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35317 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35318 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.1353, + "step": 35319 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.0895, + "step": 35320 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35321 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 5.9846, + "step": 35322 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35323 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.9077, + "step": 35324 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35325 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35326 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35327 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35328 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35329 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 6.3218, + "step": 35330 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35331 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35332 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35333 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35334 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.9254, + "step": 35335 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.8336, + "step": 35336 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35337 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35338 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35339 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 15.1358, + "step": 35340 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 2.1144, + "step": 35341 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35342 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.7752, + "step": 35343 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 1.0591, + "step": 35344 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35345 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 3.8805, + "step": 35346 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 4.2402, + "step": 35347 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35348 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 4.7415, + "step": 35349 + }, + { + "epoch": 2.95, + "learning_rate": 0.00019983838158664697, + "loss": 7.8855, + "step": 35350 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 5.6701, + "step": 35351 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 4.1764, + "step": 35352 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35353 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 7.0061, + "step": 35354 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 4.8219, + "step": 35355 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 10.6211, + "step": 35356 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35357 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.5372, + "step": 35358 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 5.2767, + "step": 35359 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 7.1889, + "step": 35360 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35361 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 8.0655, + "step": 35362 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9625, + "step": 35363 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.785, + "step": 35364 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.0617, + "step": 35365 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 8.6511, + "step": 35366 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9345, + "step": 35367 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.5785, + "step": 35368 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9292, + "step": 35369 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 7.7767, + "step": 35370 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.8368, + "step": 35371 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.8985, + "step": 35372 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35373 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35374 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.7637, + "step": 35375 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9003, + "step": 35376 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.5437, + "step": 35377 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35378 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35379 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35380 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.0251, + "step": 35381 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35382 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.1297, + "step": 35383 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 5.8569, + "step": 35384 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.4952, + "step": 35385 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 7.8304, + "step": 35386 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 3.2069, + "step": 35387 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35388 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 4.4524, + "step": 35389 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 3.6352, + "step": 35390 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.1237, + "step": 35391 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35392 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.0096, + "step": 35393 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35394 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 4.0611, + "step": 35395 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35396 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 5.5746, + "step": 35397 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35398 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 4.1544, + "step": 35399 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.791, + "step": 35400 + }, + { + "epoch": 2.96, + "eval_loss": NaN, + "eval_runtime": 39.8705, + "eval_samples_per_second": 62.703, + "eval_steps_per_second": 7.85, + "step": 35400 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 5.1618, + "step": 35401 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.0409, + "step": 35402 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35403 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.1187, + "step": 35404 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 4.1985, + "step": 35405 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.4799, + "step": 35406 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9531, + "step": 35407 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 9.7058, + "step": 35408 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 3.8349, + "step": 35409 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 8.9983, + "step": 35410 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35411 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35412 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35413 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 7.924, + "step": 35414 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35415 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 3.928, + "step": 35416 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.8849, + "step": 35417 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35418 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35419 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35420 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35421 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35422 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.6771, + "step": 35423 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35424 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35425 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.0163, + "step": 35426 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35427 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35428 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.7259, + "step": 35429 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35430 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 3.5335, + "step": 35431 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 7.1634, + "step": 35432 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.8733, + "step": 35433 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.8348, + "step": 35434 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.3521, + "step": 35435 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35436 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9313, + "step": 35437 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35438 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 3.5511, + "step": 35439 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 8.0644, + "step": 35440 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9799, + "step": 35441 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 8.9227, + "step": 35442 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35443 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35444 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.3834, + "step": 35445 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 2.0503, + "step": 35446 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35447 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35448 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 9.2377, + "step": 35449 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.9088, + "step": 35450 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 6.5018, + "step": 35451 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35452 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35453 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 1.8927, + "step": 35454 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35455 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35456 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35457 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35458 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35459 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35460 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 8.2201, + "step": 35461 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35462 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 3.9055, + "step": 35463 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35464 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35465 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 4.0483, + "step": 35466 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35467 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35468 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35469 + }, + { + "epoch": 2.96, + "learning_rate": 0.00019983838158664697, + "loss": 7.865, + "step": 35470 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35471 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35472 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.4213, + "step": 35473 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 8.7968, + "step": 35474 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35475 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 8.3418, + "step": 35476 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35477 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35478 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.7745, + "step": 35479 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.0804, + "step": 35480 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 2.0855, + "step": 35481 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35482 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 5.0245, + "step": 35483 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35484 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35485 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.9723, + "step": 35486 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 5.1659, + "step": 35487 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 4.1691, + "step": 35488 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35489 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 5.9949, + "step": 35490 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.1559, + "step": 35491 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.0532, + "step": 35492 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35493 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 7.183, + "step": 35494 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35495 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.0007, + "step": 35496 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.4926, + "step": 35497 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 10.4803, + "step": 35498 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.0414, + "step": 35499 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35500 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35501 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.8467, + "step": 35502 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35503 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35504 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.9667, + "step": 35505 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 8.0659, + "step": 35506 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35507 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35508 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35509 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.9131, + "step": 35510 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.2976, + "step": 35511 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.8414, + "step": 35512 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35513 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35514 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35515 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 8.5653, + "step": 35516 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35517 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 8.7918, + "step": 35518 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.971, + "step": 35519 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35520 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35521 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 2.6164, + "step": 35522 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 4.4037, + "step": 35523 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35524 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.1294, + "step": 35525 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.7761, + "step": 35526 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35527 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.1015, + "step": 35528 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 4.2889, + "step": 35529 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35530 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35531 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35532 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35533 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35534 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 4.606, + "step": 35535 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35536 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.9522, + "step": 35537 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 4.8716, + "step": 35538 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.0714, + "step": 35539 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 9.9307, + "step": 35540 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35541 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.5949, + "step": 35542 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 7.275, + "step": 35543 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35544 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35545 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.6775, + "step": 35546 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 4.4939, + "step": 35547 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.6004, + "step": 35548 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 2.1619, + "step": 35549 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35550 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 2.0877, + "step": 35551 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 5.6218, + "step": 35552 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35553 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 7.5558, + "step": 35554 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35555 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 5.1776, + "step": 35556 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 4.4107, + "step": 35557 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35558 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 5.399, + "step": 35559 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.7704, + "step": 35560 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35561 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.9167, + "step": 35562 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35563 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.7362, + "step": 35564 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.7025, + "step": 35565 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 2.3117, + "step": 35566 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 6.426, + "step": 35567 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35568 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 6.2367, + "step": 35569 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 2.8786, + "step": 35570 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35571 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.9267, + "step": 35572 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.8348, + "step": 35573 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 7.596, + "step": 35574 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35575 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.6671, + "step": 35576 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35577 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.9128, + "step": 35578 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35579 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 2.2239, + "step": 35580 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35581 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.3526, + "step": 35582 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35583 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.4495, + "step": 35584 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 3.3829, + "step": 35585 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35586 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.8863, + "step": 35587 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 1.0061, + "step": 35588 + }, + { + "epoch": 2.97, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35589 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.0325, + "step": 35590 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.922, + "step": 35591 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35592 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.1149, + "step": 35593 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.5498, + "step": 35594 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.2381, + "step": 35595 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35596 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 5.8981, + "step": 35597 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 4.2422, + "step": 35598 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 8.283, + "step": 35599 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35600 + }, + { + "epoch": 2.98, + "eval_loss": NaN, + "eval_runtime": 39.8657, + "eval_samples_per_second": 62.711, + "eval_steps_per_second": 7.851, + "step": 35600 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35601 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 4.3234, + "step": 35602 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35603 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35604 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.5642, + "step": 35605 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35606 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.9093, + "step": 35607 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.7232, + "step": 35608 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35609 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35610 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.961, + "step": 35611 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.3734, + "step": 35612 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35613 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.1213, + "step": 35614 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 6.332, + "step": 35615 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35616 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.274, + "step": 35617 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.9901, + "step": 35618 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.4515, + "step": 35619 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35620 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.0998, + "step": 35621 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 5.6586, + "step": 35622 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.9249, + "step": 35623 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.9392, + "step": 35624 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.8487, + "step": 35625 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35626 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35627 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35628 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35629 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35630 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.4405, + "step": 35631 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.6886, + "step": 35632 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.9378, + "step": 35633 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35634 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.9612, + "step": 35635 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35636 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.9552, + "step": 35637 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.0567, + "step": 35638 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35639 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35640 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 8.187, + "step": 35641 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35642 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.9332, + "step": 35643 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.7104, + "step": 35644 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35645 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.4645, + "step": 35646 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.169, + "step": 35647 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35648 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.9567, + "step": 35649 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 4.2804, + "step": 35650 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.7898, + "step": 35651 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35652 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35653 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.3202, + "step": 35654 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35655 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35656 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 8.9572, + "step": 35657 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35658 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.7783, + "step": 35659 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.8956, + "step": 35660 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35661 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 4.8477, + "step": 35662 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.2074, + "step": 35663 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35664 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.2461, + "step": 35665 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.1622, + "step": 35666 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35667 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35668 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.0478, + "step": 35669 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35670 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 9.5001, + "step": 35671 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35672 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 6.6355, + "step": 35673 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 8.6936, + "step": 35674 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.3853, + "step": 35675 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35676 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35677 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35678 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 6.4424, + "step": 35679 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35680 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35681 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.067, + "step": 35682 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35683 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35684 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 6.7883, + "step": 35685 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.8341, + "step": 35686 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35687 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35688 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 3.2777, + "step": 35689 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.0531, + "step": 35690 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.8041, + "step": 35691 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.9535, + "step": 35692 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.787, + "step": 35693 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.1359, + "step": 35694 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 7.2556, + "step": 35695 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.9374, + "step": 35696 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35697 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.1286, + "step": 35698 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35699 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 2.2591, + "step": 35700 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35701 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 8.7739, + "step": 35702 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35703 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35704 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35705 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35706 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 5.759, + "step": 35707 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 1.0587, + "step": 35708 + }, + { + "epoch": 2.98, + "learning_rate": 0.00019983838158664697, + "loss": 0.962, + "step": 35709 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.2099, + "step": 35710 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35711 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35712 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.2173, + "step": 35713 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.0257, + "step": 35714 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35715 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35716 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35717 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 4.1261, + "step": 35718 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35719 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.1743, + "step": 35720 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35721 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 7.5362, + "step": 35722 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 9.776, + "step": 35723 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35724 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35725 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.7586, + "step": 35726 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35727 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 3.3434, + "step": 35728 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.9944, + "step": 35729 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.0536, + "step": 35730 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 6.0116, + "step": 35731 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35732 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 3.3522, + "step": 35733 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 4.0289, + "step": 35734 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 7.7994, + "step": 35735 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35736 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.7898, + "step": 35737 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 8.4844, + "step": 35738 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35739 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.9247, + "step": 35740 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35741 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35742 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35743 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.7756, + "step": 35744 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 8.1793, + "step": 35745 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35746 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35747 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 8.6797, + "step": 35748 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35749 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35750 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35751 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.079, + "step": 35752 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.086, + "step": 35753 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 3.1229, + "step": 35754 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35755 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 7.1374, + "step": 35756 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35757 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.8785, + "step": 35758 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35759 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.0431, + "step": 35760 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35761 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35762 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35763 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35764 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.525, + "step": 35765 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35766 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.882, + "step": 35767 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35768 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35769 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35770 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.1085, + "step": 35771 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35772 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 3.5345, + "step": 35773 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 7.0497, + "step": 35774 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.5981, + "step": 35775 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35776 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35777 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35778 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35779 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.9114, + "step": 35780 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35781 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.9649, + "step": 35782 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.8417, + "step": 35783 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35784 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35785 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 4.3848, + "step": 35786 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 6.6792, + "step": 35787 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35788 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35789 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35790 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35791 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.421, + "step": 35792 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 2.1738, + "step": 35793 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35794 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35795 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35796 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 3.3286, + "step": 35797 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 1.3932, + "step": 35798 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 7.8793, + "step": 35799 + }, + { + "epoch": 2.99, + "learning_rate": 0.00019983838158664697, + "loss": 0.0, + "step": 35800 + }, + { + "epoch": 2.99, + "eval_loss": NaN, + "eval_runtime": 39.8711, + "eval_samples_per_second": 62.702, + "eval_steps_per_second": 7.85, + "step": 35800 + } + ], + "logging_steps": 1, + "max_steps": 35889, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 200, + "total_flos": 4.94178310029312e+16, + "train_batch_size": 1, + "trial_name": null, + "trial_params": null +}