Delete trainer_state.json
Browse files- trainer_state.json +0 -315
trainer_state.json
DELETED
|
@@ -1,315 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.9998338042213728,
|
| 5 |
-
"eval_steps": 500,
|
| 6 |
-
"global_step": 47,
|
| 7 |
-
"is_hyper_param_search": false,
|
| 8 |
-
"is_local_process_zero": true,
|
| 9 |
-
"is_world_process_zero": true,
|
| 10 |
-
"log_history": [
|
| 11 |
-
{
|
| 12 |
-
"epoch": 0.021273059664284527,
|
| 13 |
-
"learning_rate": 3.3333333333333333e-06,
|
| 14 |
-
"loss": 0.6113,
|
| 15 |
-
"step": 1
|
| 16 |
-
},
|
| 17 |
-
{
|
| 18 |
-
"epoch": 0.042546119328569054,
|
| 19 |
-
"learning_rate": 6.666666666666667e-06,
|
| 20 |
-
"loss": 0.6099,
|
| 21 |
-
"step": 2
|
| 22 |
-
},
|
| 23 |
-
{
|
| 24 |
-
"epoch": 0.06381917899285358,
|
| 25 |
-
"learning_rate": 1e-05,
|
| 26 |
-
"loss": 0.607,
|
| 27 |
-
"step": 3
|
| 28 |
-
},
|
| 29 |
-
{
|
| 30 |
-
"epoch": 0.08509223865713811,
|
| 31 |
-
"learning_rate": 1.3333333333333333e-05,
|
| 32 |
-
"loss": 0.5918,
|
| 33 |
-
"step": 4
|
| 34 |
-
},
|
| 35 |
-
{
|
| 36 |
-
"epoch": 0.10636529832142264,
|
| 37 |
-
"learning_rate": 1.6666666666666667e-05,
|
| 38 |
-
"loss": 0.5932,
|
| 39 |
-
"step": 5
|
| 40 |
-
},
|
| 41 |
-
{
|
| 42 |
-
"epoch": 0.12763835798570716,
|
| 43 |
-
"learning_rate": 2e-05,
|
| 44 |
-
"loss": 0.5708,
|
| 45 |
-
"step": 6
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.1489114176499917,
|
| 49 |
-
"learning_rate": 2.3333333333333336e-05,
|
| 50 |
-
"loss": 0.5353,
|
| 51 |
-
"step": 7
|
| 52 |
-
},
|
| 53 |
-
{
|
| 54 |
-
"epoch": 0.17018447731427622,
|
| 55 |
-
"learning_rate": 2.6666666666666667e-05,
|
| 56 |
-
"loss": 0.5249,
|
| 57 |
-
"step": 8
|
| 58 |
-
},
|
| 59 |
-
{
|
| 60 |
-
"epoch": 0.19145753697856074,
|
| 61 |
-
"learning_rate": 3e-05,
|
| 62 |
-
"loss": 0.524,
|
| 63 |
-
"step": 9
|
| 64 |
-
},
|
| 65 |
-
{
|
| 66 |
-
"epoch": 0.21273059664284527,
|
| 67 |
-
"learning_rate": 3.3333333333333335e-05,
|
| 68 |
-
"loss": 0.5507,
|
| 69 |
-
"step": 10
|
| 70 |
-
},
|
| 71 |
-
{
|
| 72 |
-
"epoch": 0.2340036563071298,
|
| 73 |
-
"learning_rate": 3.6666666666666666e-05,
|
| 74 |
-
"loss": 0.5335,
|
| 75 |
-
"step": 11
|
| 76 |
-
},
|
| 77 |
-
{
|
| 78 |
-
"epoch": 0.2552767159714143,
|
| 79 |
-
"learning_rate": 4e-05,
|
| 80 |
-
"loss": 0.5438,
|
| 81 |
-
"step": 12
|
| 82 |
-
},
|
| 83 |
-
{
|
| 84 |
-
"epoch": 0.2765497756356989,
|
| 85 |
-
"learning_rate": 4.3333333333333334e-05,
|
| 86 |
-
"loss": 0.5628,
|
| 87 |
-
"step": 13
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.2978228352999834,
|
| 91 |
-
"learning_rate": 4.666666666666667e-05,
|
| 92 |
-
"loss": 0.5479,
|
| 93 |
-
"step": 14
|
| 94 |
-
},
|
| 95 |
-
{
|
| 96 |
-
"epoch": 0.31909589496426793,
|
| 97 |
-
"learning_rate": 5e-05,
|
| 98 |
-
"loss": 0.5387,
|
| 99 |
-
"step": 15
|
| 100 |
-
},
|
| 101 |
-
{
|
| 102 |
-
"epoch": 0.34036895462855243,
|
| 103 |
-
"learning_rate": 4.8437500000000005e-05,
|
| 104 |
-
"loss": 0.5836,
|
| 105 |
-
"step": 16
|
| 106 |
-
},
|
| 107 |
-
{
|
| 108 |
-
"epoch": 0.361642014292837,
|
| 109 |
-
"learning_rate": 4.6875e-05,
|
| 110 |
-
"loss": 0.5331,
|
| 111 |
-
"step": 17
|
| 112 |
-
},
|
| 113 |
-
{
|
| 114 |
-
"epoch": 0.3829150739571215,
|
| 115 |
-
"learning_rate": 4.5312500000000004e-05,
|
| 116 |
-
"loss": 0.528,
|
| 117 |
-
"step": 18
|
| 118 |
-
},
|
| 119 |
-
{
|
| 120 |
-
"epoch": 0.40418813362140604,
|
| 121 |
-
"learning_rate": 4.375e-05,
|
| 122 |
-
"loss": 0.5767,
|
| 123 |
-
"step": 19
|
| 124 |
-
},
|
| 125 |
-
{
|
| 126 |
-
"epoch": 0.42546119328569054,
|
| 127 |
-
"learning_rate": 4.21875e-05,
|
| 128 |
-
"loss": 0.5232,
|
| 129 |
-
"step": 20
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.4467342529499751,
|
| 133 |
-
"learning_rate": 4.0625000000000005e-05,
|
| 134 |
-
"loss": 0.4937,
|
| 135 |
-
"step": 21
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"epoch": 0.4680073126142596,
|
| 139 |
-
"learning_rate": 3.90625e-05,
|
| 140 |
-
"loss": 0.4802,
|
| 141 |
-
"step": 22
|
| 142 |
-
},
|
| 143 |
-
{
|
| 144 |
-
"epoch": 0.48928037227854415,
|
| 145 |
-
"learning_rate": 3.7500000000000003e-05,
|
| 146 |
-
"loss": 0.4845,
|
| 147 |
-
"step": 23
|
| 148 |
-
},
|
| 149 |
-
{
|
| 150 |
-
"epoch": 0.5105534319428287,
|
| 151 |
-
"learning_rate": 3.59375e-05,
|
| 152 |
-
"loss": 0.4749,
|
| 153 |
-
"step": 24
|
| 154 |
-
},
|
| 155 |
-
{
|
| 156 |
-
"epoch": 0.5318264916071131,
|
| 157 |
-
"learning_rate": 3.4375e-05,
|
| 158 |
-
"loss": 0.4771,
|
| 159 |
-
"step": 25
|
| 160 |
-
},
|
| 161 |
-
{
|
| 162 |
-
"epoch": 0.5530995512713978,
|
| 163 |
-
"learning_rate": 3.2812500000000005e-05,
|
| 164 |
-
"loss": 0.4785,
|
| 165 |
-
"step": 26
|
| 166 |
-
},
|
| 167 |
-
{
|
| 168 |
-
"epoch": 0.5743726109356823,
|
| 169 |
-
"learning_rate": 3.125e-05,
|
| 170 |
-
"loss": 0.4827,
|
| 171 |
-
"step": 27
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.5956456705999668,
|
| 175 |
-
"learning_rate": 2.96875e-05,
|
| 176 |
-
"loss": 0.4682,
|
| 177 |
-
"step": 28
|
| 178 |
-
},
|
| 179 |
-
{
|
| 180 |
-
"epoch": 0.6169187302642513,
|
| 181 |
-
"learning_rate": 2.8125000000000003e-05,
|
| 182 |
-
"loss": 0.4669,
|
| 183 |
-
"step": 29
|
| 184 |
-
},
|
| 185 |
-
{
|
| 186 |
-
"epoch": 0.6381917899285359,
|
| 187 |
-
"learning_rate": 2.6562500000000002e-05,
|
| 188 |
-
"loss": 0.4716,
|
| 189 |
-
"step": 30
|
| 190 |
-
},
|
| 191 |
-
{
|
| 192 |
-
"epoch": 0.6594648495928204,
|
| 193 |
-
"learning_rate": 2.5e-05,
|
| 194 |
-
"loss": 0.4722,
|
| 195 |
-
"step": 31
|
| 196 |
-
},
|
| 197 |
-
{
|
| 198 |
-
"epoch": 0.6807379092571049,
|
| 199 |
-
"learning_rate": 2.34375e-05,
|
| 200 |
-
"loss": 0.4636,
|
| 201 |
-
"step": 32
|
| 202 |
-
},
|
| 203 |
-
{
|
| 204 |
-
"epoch": 0.7020109689213894,
|
| 205 |
-
"learning_rate": 2.1875e-05,
|
| 206 |
-
"loss": 0.4625,
|
| 207 |
-
"step": 33
|
| 208 |
-
},
|
| 209 |
-
{
|
| 210 |
-
"epoch": 0.723284028585674,
|
| 211 |
-
"learning_rate": 2.0312500000000002e-05,
|
| 212 |
-
"loss": 0.4675,
|
| 213 |
-
"step": 34
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.7445570882499585,
|
| 217 |
-
"learning_rate": 1.8750000000000002e-05,
|
| 218 |
-
"loss": 0.4736,
|
| 219 |
-
"step": 35
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"epoch": 0.765830147914243,
|
| 223 |
-
"learning_rate": 1.71875e-05,
|
| 224 |
-
"loss": 0.4585,
|
| 225 |
-
"step": 36
|
| 226 |
-
},
|
| 227 |
-
{
|
| 228 |
-
"epoch": 0.7871032075785275,
|
| 229 |
-
"learning_rate": 1.5625e-05,
|
| 230 |
-
"loss": 0.4623,
|
| 231 |
-
"step": 37
|
| 232 |
-
},
|
| 233 |
-
{
|
| 234 |
-
"epoch": 0.8083762672428121,
|
| 235 |
-
"learning_rate": 1.4062500000000001e-05,
|
| 236 |
-
"loss": 0.4653,
|
| 237 |
-
"step": 38
|
| 238 |
-
},
|
| 239 |
-
{
|
| 240 |
-
"epoch": 0.8296493269070966,
|
| 241 |
-
"learning_rate": 1.25e-05,
|
| 242 |
-
"loss": 0.4597,
|
| 243 |
-
"step": 39
|
| 244 |
-
},
|
| 245 |
-
{
|
| 246 |
-
"epoch": 0.8509223865713811,
|
| 247 |
-
"learning_rate": 1.09375e-05,
|
| 248 |
-
"loss": 0.4581,
|
| 249 |
-
"step": 40
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 0.8721954462356656,
|
| 253 |
-
"learning_rate": 9.375000000000001e-06,
|
| 254 |
-
"loss": 0.4612,
|
| 255 |
-
"step": 41
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.8934685058999502,
|
| 259 |
-
"learning_rate": 7.8125e-06,
|
| 260 |
-
"loss": 0.4654,
|
| 261 |
-
"step": 42
|
| 262 |
-
},
|
| 263 |
-
{
|
| 264 |
-
"epoch": 0.9147415655642347,
|
| 265 |
-
"learning_rate": 6.25e-06,
|
| 266 |
-
"loss": 0.4626,
|
| 267 |
-
"step": 43
|
| 268 |
-
},
|
| 269 |
-
{
|
| 270 |
-
"epoch": 0.9360146252285192,
|
| 271 |
-
"learning_rate": 4.6875000000000004e-06,
|
| 272 |
-
"loss": 0.4682,
|
| 273 |
-
"step": 44
|
| 274 |
-
},
|
| 275 |
-
{
|
| 276 |
-
"epoch": 0.9572876848928037,
|
| 277 |
-
"learning_rate": 3.125e-06,
|
| 278 |
-
"loss": 0.4761,
|
| 279 |
-
"step": 45
|
| 280 |
-
},
|
| 281 |
-
{
|
| 282 |
-
"epoch": 0.9785607445570883,
|
| 283 |
-
"learning_rate": 1.5625e-06,
|
| 284 |
-
"loss": 0.4777,
|
| 285 |
-
"step": 46
|
| 286 |
-
},
|
| 287 |
-
{
|
| 288 |
-
"epoch": 0.9998338042213728,
|
| 289 |
-
"learning_rate": 0.0,
|
| 290 |
-
"loss": 0.4638,
|
| 291 |
-
"step": 47
|
| 292 |
-
}
|
| 293 |
-
],
|
| 294 |
-
"logging_steps": 1.0,
|
| 295 |
-
"max_steps": 47,
|
| 296 |
-
"num_input_tokens_seen": 0,
|
| 297 |
-
"num_train_epochs": 1,
|
| 298 |
-
"save_steps": 500,
|
| 299 |
-
"stateful_callbacks": {
|
| 300 |
-
"TrainerControl": {
|
| 301 |
-
"args": {
|
| 302 |
-
"should_epoch_stop": false,
|
| 303 |
-
"should_evaluate": false,
|
| 304 |
-
"should_log": false,
|
| 305 |
-
"should_save": true,
|
| 306 |
-
"should_training_stop": true
|
| 307 |
-
},
|
| 308 |
-
"attributes": {}
|
| 309 |
-
}
|
| 310 |
-
},
|
| 311 |
-
"total_flos": 1.3679818322089083e+18,
|
| 312 |
-
"train_batch_size": 2,
|
| 313 |
-
"trial_name": null,
|
| 314 |
-
"trial_params": null
|
| 315 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|