Delete trainer_state.json
Browse files- trainer_state.json +0 -579
trainer_state.json
DELETED
|
@@ -1,579 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 3.0,
|
| 5 |
-
"eval_steps": 10,
|
| 6 |
-
"global_step": 81,
|
| 7 |
-
"is_hyper_param_search": false,
|
| 8 |
-
"is_local_process_zero": true,
|
| 9 |
-
"is_world_process_zero": true,
|
| 10 |
-
"log_history": [
|
| 11 |
-
{
|
| 12 |
-
"epoch": 0.04,
|
| 13 |
-
"learning_rate": 0.0,
|
| 14 |
-
"loss": 1.0172,
|
| 15 |
-
"step": 1
|
| 16 |
-
},
|
| 17 |
-
{
|
| 18 |
-
"epoch": 0.07,
|
| 19 |
-
"learning_rate": 0.00018927892607143717,
|
| 20 |
-
"loss": 0.9587,
|
| 21 |
-
"step": 2
|
| 22 |
-
},
|
| 23 |
-
{
|
| 24 |
-
"epoch": 0.11,
|
| 25 |
-
"learning_rate": 0.0003,
|
| 26 |
-
"loss": 0.9422,
|
| 27 |
-
"step": 3
|
| 28 |
-
},
|
| 29 |
-
{
|
| 30 |
-
"epoch": 0.15,
|
| 31 |
-
"learning_rate": 0.0003,
|
| 32 |
-
"loss": 0.9651,
|
| 33 |
-
"step": 4
|
| 34 |
-
},
|
| 35 |
-
{
|
| 36 |
-
"epoch": 0.19,
|
| 37 |
-
"learning_rate": 0.0003,
|
| 38 |
-
"loss": 1.0197,
|
| 39 |
-
"step": 5
|
| 40 |
-
},
|
| 41 |
-
{
|
| 42 |
-
"epoch": 0.22,
|
| 43 |
-
"learning_rate": 0.0003,
|
| 44 |
-
"loss": 0.9455,
|
| 45 |
-
"step": 6
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.26,
|
| 49 |
-
"learning_rate": 0.0003,
|
| 50 |
-
"loss": 0.9029,
|
| 51 |
-
"step": 7
|
| 52 |
-
},
|
| 53 |
-
{
|
| 54 |
-
"epoch": 0.3,
|
| 55 |
-
"learning_rate": 0.0003,
|
| 56 |
-
"loss": 0.9045,
|
| 57 |
-
"step": 8
|
| 58 |
-
},
|
| 59 |
-
{
|
| 60 |
-
"epoch": 0.33,
|
| 61 |
-
"learning_rate": 0.0003,
|
| 62 |
-
"loss": 0.8424,
|
| 63 |
-
"step": 9
|
| 64 |
-
},
|
| 65 |
-
{
|
| 66 |
-
"epoch": 0.37,
|
| 67 |
-
"learning_rate": 0.0003,
|
| 68 |
-
"loss": 0.882,
|
| 69 |
-
"step": 10
|
| 70 |
-
},
|
| 71 |
-
{
|
| 72 |
-
"epoch": 0.37,
|
| 73 |
-
"eval_accuracy": 0.22187782513237764,
|
| 74 |
-
"eval_loss": 0.8447974324226379,
|
| 75 |
-
"eval_runtime": 44.6213,
|
| 76 |
-
"eval_samples_per_second": 0.807,
|
| 77 |
-
"eval_steps_per_second": 0.045,
|
| 78 |
-
"step": 10
|
| 79 |
-
},
|
| 80 |
-
{
|
| 81 |
-
"epoch": 0.41,
|
| 82 |
-
"learning_rate": 0.0003,
|
| 83 |
-
"loss": 0.8986,
|
| 84 |
-
"step": 11
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"epoch": 0.44,
|
| 88 |
-
"learning_rate": 0.0003,
|
| 89 |
-
"loss": 0.9005,
|
| 90 |
-
"step": 12
|
| 91 |
-
},
|
| 92 |
-
{
|
| 93 |
-
"epoch": 0.48,
|
| 94 |
-
"learning_rate": 0.0003,
|
| 95 |
-
"loss": 0.8788,
|
| 96 |
-
"step": 13
|
| 97 |
-
},
|
| 98 |
-
{
|
| 99 |
-
"epoch": 0.52,
|
| 100 |
-
"learning_rate": 0.0003,
|
| 101 |
-
"loss": 0.8391,
|
| 102 |
-
"step": 14
|
| 103 |
-
},
|
| 104 |
-
{
|
| 105 |
-
"epoch": 0.56,
|
| 106 |
-
"learning_rate": 0.0003,
|
| 107 |
-
"loss": 0.8564,
|
| 108 |
-
"step": 15
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.59,
|
| 112 |
-
"learning_rate": 0.0003,
|
| 113 |
-
"loss": 0.84,
|
| 114 |
-
"step": 16
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.63,
|
| 118 |
-
"learning_rate": 0.0003,
|
| 119 |
-
"loss": 0.8479,
|
| 120 |
-
"step": 17
|
| 121 |
-
},
|
| 122 |
-
{
|
| 123 |
-
"epoch": 0.67,
|
| 124 |
-
"learning_rate": 0.0003,
|
| 125 |
-
"loss": 0.8174,
|
| 126 |
-
"step": 18
|
| 127 |
-
},
|
| 128 |
-
{
|
| 129 |
-
"epoch": 0.7,
|
| 130 |
-
"learning_rate": 0.0003,
|
| 131 |
-
"loss": 0.8431,
|
| 132 |
-
"step": 19
|
| 133 |
-
},
|
| 134 |
-
{
|
| 135 |
-
"epoch": 0.74,
|
| 136 |
-
"learning_rate": 0.0003,
|
| 137 |
-
"loss": 0.8456,
|
| 138 |
-
"step": 20
|
| 139 |
-
},
|
| 140 |
-
{
|
| 141 |
-
"epoch": 0.74,
|
| 142 |
-
"eval_accuracy": 0.2230832149468337,
|
| 143 |
-
"eval_loss": 0.8235660195350647,
|
| 144 |
-
"eval_runtime": 43.3512,
|
| 145 |
-
"eval_samples_per_second": 0.83,
|
| 146 |
-
"eval_steps_per_second": 0.046,
|
| 147 |
-
"step": 20
|
| 148 |
-
},
|
| 149 |
-
{
|
| 150 |
-
"epoch": 0.78,
|
| 151 |
-
"learning_rate": 0.0003,
|
| 152 |
-
"loss": 0.8614,
|
| 153 |
-
"step": 21
|
| 154 |
-
},
|
| 155 |
-
{
|
| 156 |
-
"epoch": 0.81,
|
| 157 |
-
"learning_rate": 0.0003,
|
| 158 |
-
"loss": 0.8756,
|
| 159 |
-
"step": 22
|
| 160 |
-
},
|
| 161 |
-
{
|
| 162 |
-
"epoch": 0.85,
|
| 163 |
-
"learning_rate": 0.0003,
|
| 164 |
-
"loss": 0.7752,
|
| 165 |
-
"step": 23
|
| 166 |
-
},
|
| 167 |
-
{
|
| 168 |
-
"epoch": 0.89,
|
| 169 |
-
"learning_rate": 0.0003,
|
| 170 |
-
"loss": 0.7891,
|
| 171 |
-
"step": 24
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.93,
|
| 175 |
-
"learning_rate": 0.0003,
|
| 176 |
-
"loss": 0.8094,
|
| 177 |
-
"step": 25
|
| 178 |
-
},
|
| 179 |
-
{
|
| 180 |
-
"epoch": 0.96,
|
| 181 |
-
"learning_rate": 0.0003,
|
| 182 |
-
"loss": 0.8546,
|
| 183 |
-
"step": 26
|
| 184 |
-
},
|
| 185 |
-
{
|
| 186 |
-
"epoch": 1.0,
|
| 187 |
-
"learning_rate": 0.0003,
|
| 188 |
-
"loss": 0.812,
|
| 189 |
-
"step": 27
|
| 190 |
-
},
|
| 191 |
-
{
|
| 192 |
-
"epoch": 1.04,
|
| 193 |
-
"learning_rate": 0.0003,
|
| 194 |
-
"loss": 0.769,
|
| 195 |
-
"step": 28
|
| 196 |
-
},
|
| 197 |
-
{
|
| 198 |
-
"epoch": 1.07,
|
| 199 |
-
"learning_rate": 0.0003,
|
| 200 |
-
"loss": 0.7538,
|
| 201 |
-
"step": 29
|
| 202 |
-
},
|
| 203 |
-
{
|
| 204 |
-
"epoch": 1.11,
|
| 205 |
-
"learning_rate": 0.0003,
|
| 206 |
-
"loss": 0.7431,
|
| 207 |
-
"step": 30
|
| 208 |
-
},
|
| 209 |
-
{
|
| 210 |
-
"epoch": 1.11,
|
| 211 |
-
"eval_accuracy": 0.2239011580352146,
|
| 212 |
-
"eval_loss": 0.8174943923950195,
|
| 213 |
-
"eval_runtime": 43.2867,
|
| 214 |
-
"eval_samples_per_second": 0.832,
|
| 215 |
-
"eval_steps_per_second": 0.046,
|
| 216 |
-
"step": 30
|
| 217 |
-
},
|
| 218 |
-
{
|
| 219 |
-
"epoch": 1.15,
|
| 220 |
-
"learning_rate": 0.0003,
|
| 221 |
-
"loss": 0.812,
|
| 222 |
-
"step": 31
|
| 223 |
-
},
|
| 224 |
-
{
|
| 225 |
-
"epoch": 1.19,
|
| 226 |
-
"learning_rate": 0.0003,
|
| 227 |
-
"loss": 0.8473,
|
| 228 |
-
"step": 32
|
| 229 |
-
},
|
| 230 |
-
{
|
| 231 |
-
"epoch": 1.22,
|
| 232 |
-
"learning_rate": 0.0003,
|
| 233 |
-
"loss": 0.7627,
|
| 234 |
-
"step": 33
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 1.26,
|
| 238 |
-
"learning_rate": 0.0003,
|
| 239 |
-
"loss": 0.7231,
|
| 240 |
-
"step": 34
|
| 241 |
-
},
|
| 242 |
-
{
|
| 243 |
-
"epoch": 1.3,
|
| 244 |
-
"learning_rate": 0.0003,
|
| 245 |
-
"loss": 0.7546,
|
| 246 |
-
"step": 35
|
| 247 |
-
},
|
| 248 |
-
{
|
| 249 |
-
"epoch": 1.33,
|
| 250 |
-
"learning_rate": 0.0003,
|
| 251 |
-
"loss": 0.6798,
|
| 252 |
-
"step": 36
|
| 253 |
-
},
|
| 254 |
-
{
|
| 255 |
-
"epoch": 1.37,
|
| 256 |
-
"learning_rate": 0.0003,
|
| 257 |
-
"loss": 0.7154,
|
| 258 |
-
"step": 37
|
| 259 |
-
},
|
| 260 |
-
{
|
| 261 |
-
"epoch": 1.41,
|
| 262 |
-
"learning_rate": 0.0003,
|
| 263 |
-
"loss": 0.8383,
|
| 264 |
-
"step": 38
|
| 265 |
-
},
|
| 266 |
-
{
|
| 267 |
-
"epoch": 1.44,
|
| 268 |
-
"learning_rate": 0.0003,
|
| 269 |
-
"loss": 0.7747,
|
| 270 |
-
"step": 39
|
| 271 |
-
},
|
| 272 |
-
{
|
| 273 |
-
"epoch": 1.48,
|
| 274 |
-
"learning_rate": 0.0003,
|
| 275 |
-
"loss": 0.7121,
|
| 276 |
-
"step": 40
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 1.48,
|
| 280 |
-
"eval_accuracy": 0.22361057299065823,
|
| 281 |
-
"eval_loss": 0.830994188785553,
|
| 282 |
-
"eval_runtime": 43.2574,
|
| 283 |
-
"eval_samples_per_second": 0.832,
|
| 284 |
-
"eval_steps_per_second": 0.046,
|
| 285 |
-
"step": 40
|
| 286 |
-
},
|
| 287 |
-
{
|
| 288 |
-
"epoch": 1.52,
|
| 289 |
-
"learning_rate": 0.0003,
|
| 290 |
-
"loss": 0.7714,
|
| 291 |
-
"step": 41
|
| 292 |
-
},
|
| 293 |
-
{
|
| 294 |
-
"epoch": 1.56,
|
| 295 |
-
"learning_rate": 0.0003,
|
| 296 |
-
"loss": 0.7679,
|
| 297 |
-
"step": 42
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 1.59,
|
| 301 |
-
"learning_rate": 0.0003,
|
| 302 |
-
"loss": 0.6597,
|
| 303 |
-
"step": 43
|
| 304 |
-
},
|
| 305 |
-
{
|
| 306 |
-
"epoch": 1.63,
|
| 307 |
-
"learning_rate": 0.0003,
|
| 308 |
-
"loss": 0.73,
|
| 309 |
-
"step": 44
|
| 310 |
-
},
|
| 311 |
-
{
|
| 312 |
-
"epoch": 1.67,
|
| 313 |
-
"learning_rate": 0.0003,
|
| 314 |
-
"loss": 0.7872,
|
| 315 |
-
"step": 45
|
| 316 |
-
},
|
| 317 |
-
{
|
| 318 |
-
"epoch": 1.7,
|
| 319 |
-
"learning_rate": 0.0003,
|
| 320 |
-
"loss": 0.7107,
|
| 321 |
-
"step": 46
|
| 322 |
-
},
|
| 323 |
-
{
|
| 324 |
-
"epoch": 1.74,
|
| 325 |
-
"learning_rate": 0.0003,
|
| 326 |
-
"loss": 0.6862,
|
| 327 |
-
"step": 47
|
| 328 |
-
},
|
| 329 |
-
{
|
| 330 |
-
"epoch": 1.78,
|
| 331 |
-
"learning_rate": 0.0003,
|
| 332 |
-
"loss": 0.6835,
|
| 333 |
-
"step": 48
|
| 334 |
-
},
|
| 335 |
-
{
|
| 336 |
-
"epoch": 1.81,
|
| 337 |
-
"learning_rate": 0.0003,
|
| 338 |
-
"loss": 0.8554,
|
| 339 |
-
"step": 49
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 1.85,
|
| 343 |
-
"learning_rate": 0.0003,
|
| 344 |
-
"loss": 0.6281,
|
| 345 |
-
"step": 50
|
| 346 |
-
},
|
| 347 |
-
{
|
| 348 |
-
"epoch": 1.85,
|
| 349 |
-
"eval_accuracy": 0.22325541349175598,
|
| 350 |
-
"eval_loss": 0.8365404605865479,
|
| 351 |
-
"eval_runtime": 43.3797,
|
| 352 |
-
"eval_samples_per_second": 0.83,
|
| 353 |
-
"eval_steps_per_second": 0.046,
|
| 354 |
-
"step": 50
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 1.89,
|
| 358 |
-
"learning_rate": 0.0003,
|
| 359 |
-
"loss": 0.6688,
|
| 360 |
-
"step": 51
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 1.93,
|
| 364 |
-
"learning_rate": 0.0003,
|
| 365 |
-
"loss": 0.7317,
|
| 366 |
-
"step": 52
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 1.96,
|
| 370 |
-
"learning_rate": 0.0003,
|
| 371 |
-
"loss": 0.7165,
|
| 372 |
-
"step": 53
|
| 373 |
-
},
|
| 374 |
-
{
|
| 375 |
-
"epoch": 2.0,
|
| 376 |
-
"learning_rate": 0.0003,
|
| 377 |
-
"loss": 0.5791,
|
| 378 |
-
"step": 54
|
| 379 |
-
},
|
| 380 |
-
{
|
| 381 |
-
"epoch": 2.04,
|
| 382 |
-
"learning_rate": 0.0003,
|
| 383 |
-
"loss": 0.5359,
|
| 384 |
-
"step": 55
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 2.07,
|
| 388 |
-
"learning_rate": 0.0003,
|
| 389 |
-
"loss": 0.636,
|
| 390 |
-
"step": 56
|
| 391 |
-
},
|
| 392 |
-
{
|
| 393 |
-
"epoch": 2.11,
|
| 394 |
-
"learning_rate": 0.0003,
|
| 395 |
-
"loss": 0.5707,
|
| 396 |
-
"step": 57
|
| 397 |
-
},
|
| 398 |
-
{
|
| 399 |
-
"epoch": 2.15,
|
| 400 |
-
"learning_rate": 0.0003,
|
| 401 |
-
"loss": 0.7385,
|
| 402 |
-
"step": 58
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 2.19,
|
| 406 |
-
"learning_rate": 0.0003,
|
| 407 |
-
"loss": 0.6318,
|
| 408 |
-
"step": 59
|
| 409 |
-
},
|
| 410 |
-
{
|
| 411 |
-
"epoch": 2.22,
|
| 412 |
-
"learning_rate": 0.0003,
|
| 413 |
-
"loss": 0.6325,
|
| 414 |
-
"step": 60
|
| 415 |
-
},
|
| 416 |
-
{
|
| 417 |
-
"epoch": 2.22,
|
| 418 |
-
"eval_accuracy": 0.22314778940117955,
|
| 419 |
-
"eval_loss": 0.8458040952682495,
|
| 420 |
-
"eval_runtime": 43.5251,
|
| 421 |
-
"eval_samples_per_second": 0.827,
|
| 422 |
-
"eval_steps_per_second": 0.046,
|
| 423 |
-
"step": 60
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 2.26,
|
| 427 |
-
"learning_rate": 0.0003,
|
| 428 |
-
"loss": 0.6493,
|
| 429 |
-
"step": 61
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 2.3,
|
| 433 |
-
"learning_rate": 0.0003,
|
| 434 |
-
"loss": 0.5792,
|
| 435 |
-
"step": 62
|
| 436 |
-
},
|
| 437 |
-
{
|
| 438 |
-
"epoch": 2.33,
|
| 439 |
-
"learning_rate": 0.0003,
|
| 440 |
-
"loss": 0.5246,
|
| 441 |
-
"step": 63
|
| 442 |
-
},
|
| 443 |
-
{
|
| 444 |
-
"epoch": 2.37,
|
| 445 |
-
"learning_rate": 0.0003,
|
| 446 |
-
"loss": 0.5364,
|
| 447 |
-
"step": 64
|
| 448 |
-
},
|
| 449 |
-
{
|
| 450 |
-
"epoch": 2.41,
|
| 451 |
-
"learning_rate": 0.0003,
|
| 452 |
-
"loss": 0.829,
|
| 453 |
-
"step": 65
|
| 454 |
-
},
|
| 455 |
-
{
|
| 456 |
-
"epoch": 2.44,
|
| 457 |
-
"learning_rate": 0.0003,
|
| 458 |
-
"loss": 0.5891,
|
| 459 |
-
"step": 66
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 2.48,
|
| 463 |
-
"learning_rate": 0.0003,
|
| 464 |
-
"loss": 0.5092,
|
| 465 |
-
"step": 67
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"epoch": 2.52,
|
| 469 |
-
"learning_rate": 0.0003,
|
| 470 |
-
"loss": 0.7564,
|
| 471 |
-
"step": 68
|
| 472 |
-
},
|
| 473 |
-
{
|
| 474 |
-
"epoch": 2.56,
|
| 475 |
-
"learning_rate": 0.0003,
|
| 476 |
-
"loss": 0.747,
|
| 477 |
-
"step": 69
|
| 478 |
-
},
|
| 479 |
-
{
|
| 480 |
-
"epoch": 2.59,
|
| 481 |
-
"learning_rate": 0.0003,
|
| 482 |
-
"loss": 0.6826,
|
| 483 |
-
"step": 70
|
| 484 |
-
},
|
| 485 |
-
{
|
| 486 |
-
"epoch": 2.59,
|
| 487 |
-
"eval_accuracy": 0.222168410176934,
|
| 488 |
-
"eval_loss": 0.8560571670532227,
|
| 489 |
-
"eval_runtime": 43.3441,
|
| 490 |
-
"eval_samples_per_second": 0.831,
|
| 491 |
-
"eval_steps_per_second": 0.046,
|
| 492 |
-
"step": 70
|
| 493 |
-
},
|
| 494 |
-
{
|
| 495 |
-
"epoch": 2.63,
|
| 496 |
-
"learning_rate": 0.0003,
|
| 497 |
-
"loss": 0.7208,
|
| 498 |
-
"step": 71
|
| 499 |
-
},
|
| 500 |
-
{
|
| 501 |
-
"epoch": 2.67,
|
| 502 |
-
"learning_rate": 0.0003,
|
| 503 |
-
"loss": 0.7304,
|
| 504 |
-
"step": 72
|
| 505 |
-
},
|
| 506 |
-
{
|
| 507 |
-
"epoch": 2.7,
|
| 508 |
-
"learning_rate": 0.0003,
|
| 509 |
-
"loss": 0.5485,
|
| 510 |
-
"step": 73
|
| 511 |
-
},
|
| 512 |
-
{
|
| 513 |
-
"epoch": 2.74,
|
| 514 |
-
"learning_rate": 0.0003,
|
| 515 |
-
"loss": 0.5899,
|
| 516 |
-
"step": 74
|
| 517 |
-
},
|
| 518 |
-
{
|
| 519 |
-
"epoch": 2.78,
|
| 520 |
-
"learning_rate": 0.0003,
|
| 521 |
-
"loss": 0.5291,
|
| 522 |
-
"step": 75
|
| 523 |
-
},
|
| 524 |
-
{
|
| 525 |
-
"epoch": 2.81,
|
| 526 |
-
"learning_rate": 0.0003,
|
| 527 |
-
"loss": 0.6907,
|
| 528 |
-
"step": 76
|
| 529 |
-
},
|
| 530 |
-
{
|
| 531 |
-
"epoch": 2.85,
|
| 532 |
-
"learning_rate": 0.0003,
|
| 533 |
-
"loss": 0.5826,
|
| 534 |
-
"step": 77
|
| 535 |
-
},
|
| 536 |
-
{
|
| 537 |
-
"epoch": 2.89,
|
| 538 |
-
"learning_rate": 0.0003,
|
| 539 |
-
"loss": 0.5553,
|
| 540 |
-
"step": 78
|
| 541 |
-
},
|
| 542 |
-
{
|
| 543 |
-
"epoch": 2.93,
|
| 544 |
-
"learning_rate": 0.0003,
|
| 545 |
-
"loss": 0.6393,
|
| 546 |
-
"step": 79
|
| 547 |
-
},
|
| 548 |
-
{
|
| 549 |
-
"epoch": 2.96,
|
| 550 |
-
"learning_rate": 0.0003,
|
| 551 |
-
"loss": 0.5822,
|
| 552 |
-
"step": 80
|
| 553 |
-
},
|
| 554 |
-
{
|
| 555 |
-
"epoch": 2.96,
|
| 556 |
-
"eval_accuracy": 0.2210275948168238,
|
| 557 |
-
"eval_loss": 0.8713862895965576,
|
| 558 |
-
"eval_runtime": 42.5816,
|
| 559 |
-
"eval_samples_per_second": 0.845,
|
| 560 |
-
"eval_steps_per_second": 0.047,
|
| 561 |
-
"step": 80
|
| 562 |
-
},
|
| 563 |
-
{
|
| 564 |
-
"epoch": 3.0,
|
| 565 |
-
"learning_rate": 0.0003,
|
| 566 |
-
"loss": 0.5126,
|
| 567 |
-
"step": 81
|
| 568 |
-
}
|
| 569 |
-
],
|
| 570 |
-
"logging_steps": 1.0,
|
| 571 |
-
"max_steps": 81,
|
| 572 |
-
"num_input_tokens_seen": 0,
|
| 573 |
-
"num_train_epochs": 3,
|
| 574 |
-
"save_steps": 500,
|
| 575 |
-
"total_flos": 10840779079680.0,
|
| 576 |
-
"train_batch_size": 2,
|
| 577 |
-
"trial_name": null,
|
| 578 |
-
"trial_params": null
|
| 579 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|