End of training
Browse files- all_results.json +11 -11
- eval_results.json +6 -6
- runs/Jul22_09-04-07_9fd5cb2bf5e9/events.out.tfevents.1690025808.9fd5cb2bf5e9.6476.1 +3 -0
- train_results.json +6 -6
- trainer_state.json +521 -1250
all_results.json
CHANGED
|
@@ -1,13 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy":
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 2.
|
| 8 |
-
"total_flos":
|
| 9 |
-
"train_loss": 0.
|
| 10 |
-
"train_runtime":
|
| 11 |
-
"train_samples_per_second": 4.
|
| 12 |
-
"train_steps_per_second": 0.
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.9975369458128078,
|
| 4 |
+
"eval_loss": 0.006894771475344896,
|
| 5 |
+
"eval_runtime": 51.8967,
|
| 6 |
+
"eval_samples_per_second": 23.47,
|
| 7 |
+
"eval_steps_per_second": 2.948,
|
| 8 |
+
"total_flos": 8.390377142257582e+18,
|
| 9 |
+
"train_loss": 0.465066796070651,
|
| 10 |
+
"train_runtime": 8763.5167,
|
| 11 |
+
"train_samples_per_second": 4.166,
|
| 12 |
+
"train_steps_per_second": 0.13
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy":
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 2.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.9975369458128078,
|
| 4 |
+
"eval_loss": 0.006894771475344896,
|
| 5 |
+
"eval_runtime": 51.8967,
|
| 6 |
+
"eval_samples_per_second": 23.47,
|
| 7 |
+
"eval_steps_per_second": 2.948
|
| 8 |
}
|
runs/Jul22_09-04-07_9fd5cb2bf5e9/events.out.tfevents.1690025808.9fd5cb2bf5e9.6476.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f29eb23e1ae73912dca83857ae3267db1f04629006f718396e106c20c098bd1
|
| 3 |
+
size 411
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second": 4.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"total_flos": 8.390377142257582e+18,
|
| 4 |
+
"train_loss": 0.465066796070651,
|
| 5 |
+
"train_runtime": 8763.5167,
|
| 6 |
+
"train_samples_per_second": 4.166,
|
| 7 |
+
"train_steps_per_second": 0.13
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,2212 +1,1483 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "swinv2-large-patch4-window12to16-192to256-22kto1k-ft-finetuned-Lesion-Classification-HAM10000-S/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.04,
|
| 12 |
-
"learning_rate":
|
| 13 |
-
"loss": 2.
|
| 14 |
"step": 5
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.09,
|
| 18 |
-
"learning_rate":
|
| 19 |
-
"loss":
|
| 20 |
"step": 10
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.13,
|
| 24 |
-
"learning_rate":
|
| 25 |
-
"loss":
|
| 26 |
"step": 15
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.18,
|
| 30 |
-
"learning_rate": 1.
|
| 31 |
-
"loss":
|
| 32 |
"step": 20
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.22,
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss": 1.
|
| 38 |
"step": 25
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.26,
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss": 1.
|
| 44 |
"step": 30
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.31,
|
| 48 |
-
"learning_rate":
|
| 49 |
-
"loss": 1.
|
| 50 |
"step": 35
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.35,
|
| 54 |
-
"learning_rate":
|
| 55 |
-
"loss": 1.
|
| 56 |
"step": 40
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.39,
|
| 60 |
-
"learning_rate":
|
| 61 |
-
"loss": 1.
|
| 62 |
"step": 45
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.44,
|
| 66 |
-
"learning_rate":
|
| 67 |
-
"loss": 1.
|
| 68 |
"step": 50
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.48,
|
| 72 |
-
"learning_rate":
|
| 73 |
-
"loss": 1.
|
| 74 |
"step": 55
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 0.53,
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss": 1.
|
| 80 |
"step": 60
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.57,
|
| 84 |
-
"learning_rate":
|
| 85 |
-
"loss": 1.
|
| 86 |
"step": 65
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.61,
|
| 90 |
-
"learning_rate":
|
| 91 |
-
"loss": 1.
|
| 92 |
"step": 70
|
| 93 |
},
|
| 94 |
{
|
| 95 |
"epoch": 0.66,
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 1.
|
| 98 |
"step": 75
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.7,
|
| 102 |
-
"learning_rate":
|
| 103 |
-
"loss": 1.
|
| 104 |
"step": 80
|
| 105 |
},
|
| 106 |
{
|
| 107 |
"epoch": 0.74,
|
| 108 |
-
"learning_rate":
|
| 109 |
-
"loss": 1.
|
| 110 |
"step": 85
|
| 111 |
},
|
| 112 |
{
|
| 113 |
"epoch": 0.79,
|
| 114 |
-
"learning_rate":
|
| 115 |
-
"loss": 1.
|
| 116 |
"step": 90
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.83,
|
| 120 |
-
"learning_rate":
|
| 121 |
-
"loss": 1.
|
| 122 |
"step": 95
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.88,
|
| 126 |
-
"learning_rate":
|
| 127 |
-
"loss": 1.
|
| 128 |
"step": 100
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.92,
|
| 132 |
-
"learning_rate":
|
| 133 |
-
"loss": 1.
|
| 134 |
"step": 105
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 0.96,
|
| 138 |
-
"learning_rate":
|
| 139 |
-
"loss": 1.
|
| 140 |
"step": 110
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 1.0,
|
| 144 |
-
"eval_accuracy": 0.
|
| 145 |
-
"eval_loss":
|
| 146 |
-
"eval_runtime":
|
| 147 |
-
"eval_samples_per_second": 1.
|
| 148 |
-
"eval_steps_per_second": 0.
|
| 149 |
"step": 114
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 1.01,
|
| 153 |
-
"learning_rate":
|
| 154 |
-
"loss": 1.
|
| 155 |
"step": 115
|
| 156 |
},
|
| 157 |
{
|
| 158 |
"epoch": 1.05,
|
| 159 |
-
"learning_rate":
|
| 160 |
-
"loss": 1.
|
| 161 |
"step": 120
|
| 162 |
},
|
| 163 |
{
|
| 164 |
"epoch": 1.09,
|
| 165 |
-
"learning_rate":
|
| 166 |
-
"loss":
|
| 167 |
"step": 125
|
| 168 |
},
|
| 169 |
{
|
| 170 |
"epoch": 1.14,
|
| 171 |
-
"learning_rate":
|
| 172 |
-
"loss":
|
| 173 |
"step": 130
|
| 174 |
},
|
| 175 |
{
|
| 176 |
"epoch": 1.18,
|
| 177 |
-
"learning_rate":
|
| 178 |
-
"loss": 1.
|
| 179 |
"step": 135
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 1.23,
|
| 183 |
-
"learning_rate":
|
| 184 |
-
"loss":
|
| 185 |
"step": 140
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 1.27,
|
| 189 |
-
"learning_rate":
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 145
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 1.31,
|
| 195 |
-
"learning_rate":
|
| 196 |
-
"loss": 0.
|
| 197 |
"step": 150
|
| 198 |
},
|
| 199 |
{
|
| 200 |
"epoch": 1.36,
|
| 201 |
-
"learning_rate":
|
| 202 |
-
"loss":
|
| 203 |
"step": 155
|
| 204 |
},
|
| 205 |
{
|
| 206 |
"epoch": 1.4,
|
| 207 |
-
"learning_rate":
|
| 208 |
-
"loss": 0.
|
| 209 |
"step": 160
|
| 210 |
},
|
| 211 |
{
|
| 212 |
"epoch": 1.44,
|
| 213 |
-
"learning_rate":
|
| 214 |
-
"loss": 0.
|
| 215 |
"step": 165
|
| 216 |
},
|
| 217 |
{
|
| 218 |
"epoch": 1.49,
|
| 219 |
-
"learning_rate":
|
| 220 |
-
"loss": 0.
|
| 221 |
"step": 170
|
| 222 |
},
|
| 223 |
{
|
| 224 |
"epoch": 1.53,
|
| 225 |
-
"learning_rate": 1.
|
| 226 |
-
"loss": 0.
|
| 227 |
"step": 175
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 1.58,
|
| 231 |
-
"learning_rate": 1.
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 180
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 1.62,
|
| 237 |
-
"learning_rate": 1.
|
| 238 |
-
"loss": 0.
|
| 239 |
"step": 185
|
| 240 |
},
|
| 241 |
{
|
| 242 |
"epoch": 1.66,
|
| 243 |
-
"learning_rate": 1.
|
| 244 |
-
"loss": 0.
|
| 245 |
"step": 190
|
| 246 |
},
|
| 247 |
{
|
| 248 |
"epoch": 1.71,
|
| 249 |
-
"learning_rate": 1.
|
| 250 |
-
"loss": 0.
|
| 251 |
"step": 195
|
| 252 |
},
|
| 253 |
{
|
| 254 |
"epoch": 1.75,
|
| 255 |
-
"learning_rate": 1.
|
| 256 |
-
"loss": 0.
|
| 257 |
"step": 200
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 1.79,
|
| 261 |
-
"learning_rate": 1.
|
| 262 |
-
"loss": 0.
|
| 263 |
"step": 205
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 1.84,
|
| 267 |
-
"learning_rate": 1.
|
| 268 |
-
"loss": 0.
|
| 269 |
"step": 210
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 1.88,
|
| 273 |
-
"learning_rate": 1.
|
| 274 |
-
"loss": 0.
|
| 275 |
"step": 215
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 1.93,
|
| 279 |
-
"learning_rate": 1.
|
| 280 |
-
"loss": 0.
|
| 281 |
"step": 220
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 1.97,
|
| 285 |
-
"learning_rate": 1.
|
| 286 |
-
"loss": 0.
|
| 287 |
"step": 225
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 2.0,
|
| 291 |
-
"eval_accuracy": 0.
|
| 292 |
-
"eval_loss": 0.
|
| 293 |
-
"eval_runtime":
|
| 294 |
-
"eval_samples_per_second":
|
| 295 |
-
"eval_steps_per_second": 2.
|
| 296 |
"step": 228
|
| 297 |
},
|
| 298 |
{
|
| 299 |
"epoch": 2.01,
|
| 300 |
-
"learning_rate":
|
| 301 |
-
"loss": 0.
|
| 302 |
"step": 230
|
| 303 |
},
|
| 304 |
{
|
| 305 |
"epoch": 2.06,
|
| 306 |
-
"learning_rate":
|
| 307 |
-
"loss": 0.
|
| 308 |
"step": 235
|
| 309 |
},
|
| 310 |
{
|
| 311 |
"epoch": 2.1,
|
| 312 |
-
"learning_rate":
|
| 313 |
-
"loss": 0.
|
| 314 |
"step": 240
|
| 315 |
},
|
| 316 |
{
|
| 317 |
"epoch": 2.14,
|
| 318 |
-
"learning_rate":
|
| 319 |
-
"loss": 0.
|
| 320 |
"step": 245
|
| 321 |
},
|
| 322 |
{
|
| 323 |
"epoch": 2.19,
|
| 324 |
-
"learning_rate":
|
| 325 |
-
"loss": 0.
|
| 326 |
"step": 250
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 2.23,
|
| 330 |
-
"learning_rate":
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 255
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 2.28,
|
| 336 |
-
"learning_rate":
|
| 337 |
-
"loss": 0.
|
| 338 |
"step": 260
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 2.32,
|
| 342 |
-
"learning_rate":
|
| 343 |
-
"loss": 0.
|
| 344 |
"step": 265
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 2.36,
|
| 348 |
-
"learning_rate":
|
| 349 |
-
"loss": 0.
|
| 350 |
"step": 270
|
| 351 |
},
|
| 352 |
{
|
| 353 |
"epoch": 2.41,
|
| 354 |
-
"learning_rate":
|
| 355 |
-
"loss": 0.
|
| 356 |
"step": 275
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 2.45,
|
| 360 |
-
"learning_rate":
|
| 361 |
-
"loss": 0.
|
| 362 |
"step": 280
|
| 363 |
},
|
| 364 |
{
|
| 365 |
"epoch": 2.49,
|
| 366 |
-
"learning_rate":
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 285
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 2.54,
|
| 372 |
-
"learning_rate":
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 290
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 2.58,
|
| 378 |
-
"learning_rate":
|
| 379 |
-
"loss": 0.
|
| 380 |
"step": 295
|
| 381 |
},
|
| 382 |
{
|
| 383 |
"epoch": 2.63,
|
| 384 |
-
"learning_rate":
|
| 385 |
-
"loss": 0.
|
| 386 |
"step": 300
|
| 387 |
},
|
| 388 |
{
|
| 389 |
"epoch": 2.67,
|
| 390 |
-
"learning_rate":
|
| 391 |
-
"loss": 0.
|
| 392 |
"step": 305
|
| 393 |
},
|
| 394 |
{
|
| 395 |
"epoch": 2.71,
|
| 396 |
-
"learning_rate":
|
| 397 |
-
"loss": 0.
|
| 398 |
"step": 310
|
| 399 |
},
|
| 400 |
{
|
| 401 |
"epoch": 2.76,
|
| 402 |
-
"learning_rate":
|
| 403 |
-
"loss": 0.
|
| 404 |
"step": 315
|
| 405 |
},
|
| 406 |
{
|
| 407 |
"epoch": 2.8,
|
| 408 |
-
"learning_rate":
|
| 409 |
-
"loss": 0.
|
| 410 |
"step": 320
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 2.84,
|
| 414 |
-
"learning_rate":
|
| 415 |
-
"loss": 0.
|
| 416 |
"step": 325
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 2.89,
|
| 420 |
-
"learning_rate":
|
| 421 |
-
"loss": 0.
|
| 422 |
"step": 330
|
| 423 |
},
|
| 424 |
{
|
| 425 |
"epoch": 2.93,
|
| 426 |
-
"learning_rate":
|
| 427 |
-
"loss": 0.
|
| 428 |
"step": 335
|
| 429 |
},
|
| 430 |
{
|
| 431 |
"epoch": 2.98,
|
| 432 |
-
"learning_rate":
|
| 433 |
-
"loss": 0.
|
| 434 |
"step": 340
|
| 435 |
},
|
| 436 |
{
|
| 437 |
"epoch": 2.99,
|
| 438 |
-
"eval_accuracy": 0.
|
| 439 |
-
"eval_loss": 0.
|
| 440 |
-
"eval_runtime":
|
| 441 |
-
"eval_samples_per_second":
|
| 442 |
-
"eval_steps_per_second": 2.
|
| 443 |
"step": 342
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"epoch": 3.02,
|
| 447 |
-
"learning_rate":
|
| 448 |
-
"loss": 0.
|
| 449 |
"step": 345
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 3.06,
|
| 453 |
-
"learning_rate":
|
| 454 |
-
"loss": 0.
|
| 455 |
"step": 350
|
| 456 |
},
|
| 457 |
{
|
| 458 |
"epoch": 3.11,
|
| 459 |
-
"learning_rate":
|
| 460 |
-
"loss": 0.
|
| 461 |
"step": 355
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 3.15,
|
| 465 |
-
"learning_rate":
|
| 466 |
-
"loss": 0.
|
| 467 |
"step": 360
|
| 468 |
},
|
| 469 |
{
|
| 470 |
"epoch": 3.19,
|
| 471 |
-
"learning_rate":
|
| 472 |
-
"loss": 0.
|
| 473 |
"step": 365
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 3.24,
|
| 477 |
-
"learning_rate":
|
| 478 |
-
"loss": 0.
|
| 479 |
"step": 370
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 3.28,
|
| 483 |
-
"learning_rate":
|
| 484 |
-
"loss": 0.
|
| 485 |
"step": 375
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"epoch": 3.33,
|
| 489 |
-
"learning_rate":
|
| 490 |
-
"loss": 0.
|
| 491 |
"step": 380
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 3.37,
|
| 495 |
-
"learning_rate":
|
| 496 |
-
"loss": 0.
|
| 497 |
"step": 385
|
| 498 |
},
|
| 499 |
{
|
| 500 |
"epoch": 3.41,
|
| 501 |
-
"learning_rate":
|
| 502 |
-
"loss": 0.
|
| 503 |
"step": 390
|
| 504 |
},
|
| 505 |
{
|
| 506 |
"epoch": 3.46,
|
| 507 |
-
"learning_rate":
|
| 508 |
-
"loss": 0.
|
| 509 |
"step": 395
|
| 510 |
},
|
| 511 |
{
|
| 512 |
"epoch": 3.5,
|
| 513 |
-
"learning_rate":
|
| 514 |
-
"loss": 0.
|
| 515 |
"step": 400
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 3.54,
|
| 519 |
-
"learning_rate":
|
| 520 |
-
"loss": 0.
|
| 521 |
"step": 405
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 3.59,
|
| 525 |
-
"learning_rate":
|
| 526 |
-
"loss": 0.
|
| 527 |
"step": 410
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"epoch": 3.63,
|
| 531 |
-
"learning_rate":
|
| 532 |
-
"loss": 0.
|
| 533 |
"step": 415
|
| 534 |
},
|
| 535 |
{
|
| 536 |
"epoch": 3.68,
|
| 537 |
-
"learning_rate":
|
| 538 |
-
"loss": 0.
|
| 539 |
"step": 420
|
| 540 |
},
|
| 541 |
{
|
| 542 |
"epoch": 3.72,
|
| 543 |
-
"learning_rate":
|
| 544 |
-
"loss": 0.
|
| 545 |
"step": 425
|
| 546 |
},
|
| 547 |
{
|
| 548 |
"epoch": 3.76,
|
| 549 |
-
"learning_rate":
|
| 550 |
-
"loss": 0.
|
| 551 |
"step": 430
|
| 552 |
},
|
| 553 |
{
|
| 554 |
"epoch": 3.81,
|
| 555 |
-
"learning_rate":
|
| 556 |
-
"loss": 0.
|
| 557 |
"step": 435
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 3.85,
|
| 561 |
-
"learning_rate":
|
| 562 |
-
"loss": 0.
|
| 563 |
"step": 440
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 3.89,
|
| 567 |
-
"learning_rate":
|
| 568 |
-
"loss": 0.
|
| 569 |
"step": 445
|
| 570 |
},
|
| 571 |
{
|
| 572 |
"epoch": 3.94,
|
| 573 |
-
"learning_rate":
|
| 574 |
-
"loss": 0.
|
| 575 |
"step": 450
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"epoch": 3.98,
|
| 579 |
-
"learning_rate":
|
| 580 |
-
"loss": 0.
|
| 581 |
"step": 455
|
| 582 |
},
|
| 583 |
{
|
| 584 |
"epoch": 4.0,
|
| 585 |
-
"eval_accuracy": 0.
|
| 586 |
-
"eval_loss": 0.
|
| 587 |
-
"eval_runtime":
|
| 588 |
-
"eval_samples_per_second":
|
| 589 |
-
"eval_steps_per_second": 2.
|
| 590 |
"step": 457
|
| 591 |
},
|
| 592 |
{
|
| 593 |
"epoch": 4.03,
|
| 594 |
-
"learning_rate":
|
| 595 |
-
"loss": 0.
|
| 596 |
"step": 460
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 4.07,
|
| 600 |
-
"learning_rate":
|
| 601 |
-
"loss": 0.
|
| 602 |
"step": 465
|
| 603 |
},
|
| 604 |
{
|
| 605 |
"epoch": 4.11,
|
| 606 |
-
"learning_rate":
|
| 607 |
-
"loss": 0.
|
| 608 |
"step": 470
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 4.16,
|
| 612 |
-
"learning_rate":
|
| 613 |
-
"loss": 0.
|
| 614 |
"step": 475
|
| 615 |
},
|
| 616 |
{
|
| 617 |
"epoch": 4.2,
|
| 618 |
-
"learning_rate":
|
| 619 |
"loss": 0.3722,
|
| 620 |
"step": 480
|
| 621 |
},
|
| 622 |
{
|
| 623 |
"epoch": 4.25,
|
| 624 |
-
"learning_rate":
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 485
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 4.29,
|
| 630 |
-
"learning_rate":
|
| 631 |
-
"loss": 0.
|
| 632 |
"step": 490
|
| 633 |
},
|
| 634 |
{
|
| 635 |
"epoch": 4.33,
|
| 636 |
-
"learning_rate":
|
| 637 |
-
"loss": 0.
|
| 638 |
"step": 495
|
| 639 |
},
|
| 640 |
{
|
| 641 |
"epoch": 4.38,
|
| 642 |
-
"learning_rate":
|
| 643 |
-
"loss": 0.
|
| 644 |
"step": 500
|
| 645 |
},
|
| 646 |
{
|
| 647 |
"epoch": 4.42,
|
| 648 |
-
"learning_rate":
|
| 649 |
-
"loss": 0.
|
| 650 |
"step": 505
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 4.46,
|
| 654 |
-
"learning_rate":
|
| 655 |
-
"loss": 0.
|
| 656 |
"step": 510
|
| 657 |
},
|
| 658 |
{
|
| 659 |
"epoch": 4.51,
|
| 660 |
-
"learning_rate":
|
| 661 |
-
"loss": 0.
|
| 662 |
"step": 515
|
| 663 |
},
|
| 664 |
{
|
| 665 |
"epoch": 4.55,
|
| 666 |
-
"learning_rate":
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 520
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 4.6,
|
| 672 |
-
"learning_rate":
|
| 673 |
-
"loss": 0.
|
| 674 |
"step": 525
|
| 675 |
},
|
| 676 |
{
|
| 677 |
"epoch": 4.64,
|
| 678 |
-
"learning_rate":
|
| 679 |
-
"loss": 0.
|
| 680 |
"step": 530
|
| 681 |
},
|
| 682 |
{
|
| 683 |
"epoch": 4.68,
|
| 684 |
-
"learning_rate":
|
| 685 |
-
"loss": 0.
|
| 686 |
"step": 535
|
| 687 |
},
|
| 688 |
{
|
| 689 |
"epoch": 4.73,
|
| 690 |
-
"learning_rate":
|
| 691 |
-
"loss": 0.
|
| 692 |
"step": 540
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 4.77,
|
| 696 |
-
"learning_rate":
|
| 697 |
-
"loss": 0.
|
| 698 |
"step": 545
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 4.81,
|
| 702 |
-
"learning_rate":
|
| 703 |
-
"loss": 0.
|
| 704 |
"step": 550
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 4.86,
|
| 708 |
-
"learning_rate":
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 555
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 4.9,
|
| 714 |
-
"learning_rate":
|
| 715 |
-
"loss": 0.
|
| 716 |
"step": 560
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 4.95,
|
| 720 |
-
"learning_rate":
|
| 721 |
-
"loss": 0.
|
| 722 |
"step": 565
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 4.99,
|
| 726 |
-
"learning_rate":
|
| 727 |
-
"loss": 0.
|
| 728 |
"step": 570
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 5.0,
|
| 732 |
-
"eval_accuracy": 0.
|
| 733 |
-
"eval_loss": 0.
|
| 734 |
-
"eval_runtime":
|
| 735 |
-
"eval_samples_per_second":
|
| 736 |
-
"eval_steps_per_second": 2.
|
| 737 |
"step": 571
|
| 738 |
},
|
| 739 |
{
|
| 740 |
"epoch": 5.03,
|
| 741 |
-
"learning_rate":
|
| 742 |
-
"loss": 0.
|
| 743 |
"step": 575
|
| 744 |
},
|
| 745 |
{
|
| 746 |
"epoch": 5.08,
|
| 747 |
-
"learning_rate":
|
| 748 |
-
"loss": 0.
|
| 749 |
"step": 580
|
| 750 |
},
|
| 751 |
{
|
| 752 |
"epoch": 5.12,
|
| 753 |
-
"learning_rate":
|
| 754 |
-
"loss": 0.
|
| 755 |
"step": 585
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 5.16,
|
| 759 |
-
"learning_rate":
|
| 760 |
-
"loss": 0.
|
| 761 |
"step": 590
|
| 762 |
},
|
| 763 |
{
|
| 764 |
"epoch": 5.21,
|
| 765 |
-
"learning_rate":
|
| 766 |
-
"loss": 0.
|
| 767 |
"step": 595
|
| 768 |
},
|
| 769 |
{
|
| 770 |
"epoch": 5.25,
|
| 771 |
-
"learning_rate":
|
| 772 |
-
"loss": 0.
|
| 773 |
"step": 600
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 5.3,
|
| 777 |
-
"learning_rate":
|
| 778 |
-
"loss": 0.
|
| 779 |
"step": 605
|
| 780 |
},
|
| 781 |
{
|
| 782 |
"epoch": 5.34,
|
| 783 |
-
"learning_rate":
|
| 784 |
-
"loss": 0.
|
| 785 |
"step": 610
|
| 786 |
},
|
| 787 |
{
|
| 788 |
"epoch": 5.38,
|
| 789 |
-
"learning_rate":
|
| 790 |
-
"loss": 0.
|
| 791 |
"step": 615
|
| 792 |
},
|
| 793 |
{
|
| 794 |
"epoch": 5.43,
|
| 795 |
-
"learning_rate":
|
| 796 |
-
"loss": 0.
|
| 797 |
"step": 620
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 5.47,
|
| 801 |
-
"learning_rate":
|
| 802 |
-
"loss": 0.
|
| 803 |
"step": 625
|
| 804 |
},
|
| 805 |
{
|
| 806 |
"epoch": 5.51,
|
| 807 |
-
"learning_rate":
|
| 808 |
-
"loss": 0.
|
| 809 |
"step": 630
|
| 810 |
},
|
| 811 |
{
|
| 812 |
"epoch": 5.56,
|
| 813 |
-
"learning_rate":
|
| 814 |
-
"loss": 0.
|
| 815 |
"step": 635
|
| 816 |
},
|
| 817 |
{
|
| 818 |
"epoch": 5.6,
|
| 819 |
-
"learning_rate":
|
| 820 |
-
"loss": 0.
|
| 821 |
"step": 640
|
| 822 |
},
|
| 823 |
{
|
| 824 |
"epoch": 5.65,
|
| 825 |
-
"learning_rate":
|
| 826 |
-
"loss": 0.
|
| 827 |
"step": 645
|
| 828 |
},
|
| 829 |
{
|
| 830 |
"epoch": 5.69,
|
| 831 |
-
"learning_rate":
|
| 832 |
-
"loss": 0.
|
| 833 |
"step": 650
|
| 834 |
},
|
| 835 |
{
|
| 836 |
"epoch": 5.73,
|
| 837 |
-
"learning_rate":
|
| 838 |
-
"loss": 0.
|
| 839 |
"step": 655
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 5.78,
|
| 843 |
-
"learning_rate":
|
| 844 |
-
"loss": 0.
|
| 845 |
"step": 660
|
| 846 |
},
|
| 847 |
{
|
| 848 |
"epoch": 5.82,
|
| 849 |
-
"learning_rate":
|
| 850 |
-
"loss": 0.
|
| 851 |
"step": 665
|
| 852 |
},
|
| 853 |
{
|
| 854 |
"epoch": 5.86,
|
| 855 |
-
"learning_rate":
|
| 856 |
-
"loss": 0.
|
| 857 |
"step": 670
|
| 858 |
},
|
| 859 |
{
|
| 860 |
"epoch": 5.91,
|
| 861 |
-
"learning_rate":
|
| 862 |
-
"loss": 0.
|
| 863 |
"step": 675
|
| 864 |
},
|
| 865 |
{
|
| 866 |
"epoch": 5.95,
|
| 867 |
-
"learning_rate":
|
| 868 |
-
"loss": 0.
|
| 869 |
"step": 680
|
| 870 |
},
|
| 871 |
{
|
| 872 |
"epoch": 6.0,
|
| 873 |
-
"learning_rate":
|
| 874 |
-
"loss": 0.
|
| 875 |
"step": 685
|
| 876 |
},
|
| 877 |
{
|
| 878 |
"epoch": 6.0,
|
| 879 |
-
"eval_accuracy": 0.
|
| 880 |
-
"eval_loss": 0.
|
| 881 |
-
"eval_runtime":
|
| 882 |
-
"eval_samples_per_second":
|
| 883 |
-
"eval_steps_per_second": 2.
|
| 884 |
"step": 685
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 6.04,
|
| 888 |
-
"learning_rate":
|
| 889 |
-
"loss": 0.
|
| 890 |
"step": 690
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 6.08,
|
| 894 |
-
"learning_rate":
|
| 895 |
-
"loss": 0.
|
| 896 |
"step": 695
|
| 897 |
},
|
| 898 |
{
|
| 899 |
"epoch": 6.13,
|
| 900 |
-
"learning_rate":
|
| 901 |
-
"loss": 0.
|
| 902 |
"step": 700
|
| 903 |
},
|
| 904 |
{
|
| 905 |
"epoch": 6.17,
|
| 906 |
-
"learning_rate":
|
| 907 |
-
"loss": 0.
|
| 908 |
"step": 705
|
| 909 |
},
|
| 910 |
{
|
| 911 |
"epoch": 6.21,
|
| 912 |
-
"learning_rate":
|
| 913 |
-
"loss": 0.
|
| 914 |
"step": 710
|
| 915 |
},
|
| 916 |
{
|
| 917 |
"epoch": 6.26,
|
| 918 |
-
"learning_rate":
|
| 919 |
-
"loss": 0.
|
| 920 |
"step": 715
|
| 921 |
},
|
| 922 |
{
|
| 923 |
"epoch": 6.3,
|
| 924 |
-
"learning_rate":
|
| 925 |
-
"loss": 0.
|
| 926 |
"step": 720
|
| 927 |
},
|
| 928 |
{
|
| 929 |
"epoch": 6.35,
|
| 930 |
-
"learning_rate":
|
| 931 |
-
"loss": 0.
|
| 932 |
"step": 725
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 6.39,
|
| 936 |
-
"learning_rate":
|
| 937 |
-
"loss": 0.
|
| 938 |
"step": 730
|
| 939 |
},
|
| 940 |
{
|
| 941 |
"epoch": 6.43,
|
| 942 |
-
"learning_rate":
|
| 943 |
-
"loss": 0.
|
| 944 |
"step": 735
|
| 945 |
},
|
| 946 |
{
|
| 947 |
"epoch": 6.48,
|
| 948 |
-
"learning_rate":
|
| 949 |
-
"loss": 0.
|
| 950 |
"step": 740
|
| 951 |
},
|
| 952 |
{
|
| 953 |
"epoch": 6.52,
|
| 954 |
-
"learning_rate":
|
| 955 |
-
"loss": 0.
|
| 956 |
"step": 745
|
| 957 |
},
|
| 958 |
{
|
| 959 |
"epoch": 6.56,
|
| 960 |
-
"learning_rate":
|
| 961 |
-
"loss": 0.
|
| 962 |
"step": 750
|
| 963 |
},
|
| 964 |
{
|
| 965 |
"epoch": 6.61,
|
| 966 |
-
"learning_rate":
|
| 967 |
-
"loss": 0.
|
| 968 |
"step": 755
|
| 969 |
},
|
| 970 |
{
|
| 971 |
"epoch": 6.65,
|
| 972 |
-
"learning_rate":
|
| 973 |
-
"loss": 0.
|
| 974 |
"step": 760
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 6.7,
|
| 978 |
-
"learning_rate":
|
| 979 |
-
"loss": 0.
|
| 980 |
"step": 765
|
| 981 |
},
|
| 982 |
{
|
| 983 |
"epoch": 6.74,
|
| 984 |
-
"learning_rate":
|
| 985 |
-
"loss": 0.
|
| 986 |
"step": 770
|
| 987 |
},
|
| 988 |
{
|
| 989 |
"epoch": 6.78,
|
| 990 |
-
"learning_rate":
|
| 991 |
-
"loss": 0.
|
| 992 |
"step": 775
|
| 993 |
},
|
| 994 |
{
|
| 995 |
"epoch": 6.83,
|
| 996 |
-
"learning_rate":
|
| 997 |
-
"loss": 0.
|
| 998 |
"step": 780
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
"epoch": 6.87,
|
| 1002 |
-
"learning_rate":
|
| 1003 |
-
"loss": 0.
|
| 1004 |
"step": 785
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 6.91,
|
| 1008 |
-
"learning_rate":
|
| 1009 |
-
"loss": 0.
|
| 1010 |
"step": 790
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
"epoch": 6.96,
|
| 1014 |
-
"learning_rate":
|
| 1015 |
-
"loss": 0.
|
| 1016 |
"step": 795
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 6.99,
|
| 1020 |
-
"eval_accuracy": 0.
|
| 1021 |
-
"eval_loss": 0.
|
| 1022 |
-
"eval_runtime":
|
| 1023 |
-
"eval_samples_per_second":
|
| 1024 |
-
"eval_steps_per_second": 2.
|
| 1025 |
"step": 799
|
| 1026 |
},
|
| 1027 |
{
|
| 1028 |
"epoch": 7.0,
|
| 1029 |
-
"learning_rate":
|
| 1030 |
-
"loss": 0.
|
| 1031 |
"step": 800
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
"epoch": 7.05,
|
| 1035 |
-
"learning_rate":
|
| 1036 |
-
"loss": 0.
|
| 1037 |
"step": 805
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 7.09,
|
| 1041 |
-
"learning_rate":
|
| 1042 |
-
"loss": 0.
|
| 1043 |
"step": 810
|
| 1044 |
},
|
| 1045 |
{
|
| 1046 |
"epoch": 7.13,
|
| 1047 |
-
"learning_rate":
|
| 1048 |
-
"loss": 0.
|
| 1049 |
"step": 815
|
| 1050 |
},
|
| 1051 |
{
|
| 1052 |
"epoch": 7.18,
|
| 1053 |
-
"learning_rate":
|
| 1054 |
-
"loss": 0.
|
| 1055 |
"step": 820
|
| 1056 |
},
|
| 1057 |
{
|
| 1058 |
"epoch": 7.22,
|
| 1059 |
-
"learning_rate":
|
| 1060 |
-
"loss": 0.
|
| 1061 |
"step": 825
|
| 1062 |
},
|
| 1063 |
{
|
| 1064 |
"epoch": 7.26,
|
| 1065 |
-
"learning_rate":
|
| 1066 |
-
"loss": 0.
|
| 1067 |
"step": 830
|
| 1068 |
},
|
| 1069 |
{
|
| 1070 |
"epoch": 7.31,
|
| 1071 |
-
"learning_rate":
|
| 1072 |
-
"loss": 0.
|
| 1073 |
"step": 835
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
"epoch": 7.35,
|
| 1077 |
-
"learning_rate":
|
| 1078 |
-
"loss": 0.
|
| 1079 |
"step": 840
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 7.4,
|
| 1083 |
-
"learning_rate":
|
| 1084 |
-
"loss": 0.
|
| 1085 |
"step": 845
|
| 1086 |
},
|
| 1087 |
{
|
| 1088 |
"epoch": 7.44,
|
| 1089 |
-
"learning_rate":
|
| 1090 |
-
"loss": 0.
|
| 1091 |
"step": 850
|
| 1092 |
},
|
| 1093 |
{
|
| 1094 |
"epoch": 7.48,
|
| 1095 |
-
"learning_rate": 5e-05,
|
| 1096 |
-
"loss": 0.
|
| 1097 |
"step": 855
|
| 1098 |
},
|
| 1099 |
{
|
| 1100 |
"epoch": 7.53,
|
| 1101 |
-
"learning_rate":
|
| 1102 |
-
"loss": 0.
|
| 1103 |
"step": 860
|
| 1104 |
},
|
| 1105 |
{
|
| 1106 |
"epoch": 7.57,
|
| 1107 |
-
"learning_rate":
|
| 1108 |
-
"loss": 0.
|
| 1109 |
"step": 865
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
"epoch": 7.61,
|
| 1113 |
-
"learning_rate":
|
| 1114 |
-
"loss": 0.
|
| 1115 |
"step": 870
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
"epoch": 7.66,
|
| 1119 |
-
"learning_rate":
|
| 1120 |
-
"loss": 0.
|
| 1121 |
"step": 875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 7.7,
|
| 1125 |
-
"learning_rate":
|
| 1126 |
-
"loss": 0.
|
| 1127 |
"step": 880
|
| 1128 |
},
|
| 1129 |
{
|
| 1130 |
"epoch": 7.75,
|
| 1131 |
-
"learning_rate":
|
| 1132 |
-
"loss": 0.
|
| 1133 |
"step": 885
|
| 1134 |
},
|
| 1135 |
{
|
| 1136 |
"epoch": 7.79,
|
| 1137 |
-
"learning_rate":
|
| 1138 |
-
"loss": 0.
|
| 1139 |
"step": 890
|
| 1140 |
},
|
| 1141 |
{
|
| 1142 |
"epoch": 7.83,
|
| 1143 |
-
"learning_rate":
|
| 1144 |
-
"loss": 0.
|
| 1145 |
"step": 895
|
| 1146 |
},
|
| 1147 |
{
|
| 1148 |
"epoch": 7.88,
|
| 1149 |
-
"learning_rate":
|
| 1150 |
-
"loss": 0.
|
| 1151 |
"step": 900
|
| 1152 |
},
|
| 1153 |
{
|
| 1154 |
"epoch": 7.92,
|
| 1155 |
-
"learning_rate":
|
| 1156 |
-
"loss": 0.
|
| 1157 |
"step": 905
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
"epoch": 7.96,
|
| 1161 |
-
"learning_rate":
|
| 1162 |
-
"loss": 0.
|
| 1163 |
"step": 910
|
| 1164 |
},
|
| 1165 |
{
|
| 1166 |
"epoch": 8.0,
|
| 1167 |
-
"eval_accuracy": 0.
|
| 1168 |
-
"eval_loss": 0.
|
| 1169 |
-
"eval_runtime":
|
| 1170 |
-
"eval_samples_per_second":
|
| 1171 |
-
"eval_steps_per_second": 2.
|
| 1172 |
"step": 914
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 8.01,
|
| 1176 |
-
"learning_rate":
|
| 1177 |
-
"loss": 0.
|
| 1178 |
"step": 915
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
"epoch": 8.05,
|
| 1182 |
-
"learning_rate":
|
| 1183 |
-
"loss": 0.
|
| 1184 |
"step": 920
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
"epoch": 8.1,
|
| 1188 |
-
"learning_rate":
|
| 1189 |
-
"loss": 0.
|
| 1190 |
"step": 925
|
| 1191 |
},
|
| 1192 |
{
|
| 1193 |
"epoch": 8.14,
|
| 1194 |
-
"learning_rate":
|
| 1195 |
-
"loss": 0.
|
| 1196 |
"step": 930
|
| 1197 |
},
|
| 1198 |
{
|
| 1199 |
"epoch": 8.18,
|
| 1200 |
-
"learning_rate":
|
| 1201 |
-
"loss": 0.
|
| 1202 |
"step": 935
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
"epoch": 8.23,
|
| 1206 |
-
"learning_rate":
|
| 1207 |
-
"loss": 0.
|
| 1208 |
"step": 940
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
"epoch": 8.27,
|
| 1212 |
-
"learning_rate":
|
| 1213 |
-
"loss": 0.
|
| 1214 |
"step": 945
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 8.32,
|
| 1218 |
-
"learning_rate":
|
| 1219 |
-
"loss": 0.
|
| 1220 |
"step": 950
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
"epoch": 8.36,
|
| 1224 |
-
"learning_rate":
|
| 1225 |
-
"loss": 0.
|
| 1226 |
"step": 955
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
"epoch": 8.4,
|
| 1230 |
-
"learning_rate":
|
| 1231 |
-
"loss": 0.
|
| 1232 |
"step": 960
|
| 1233 |
},
|
| 1234 |
{
|
| 1235 |
"epoch": 8.45,
|
| 1236 |
-
"learning_rate":
|
| 1237 |
-
"loss": 0.
|
| 1238 |
"step": 965
|
| 1239 |
},
|
| 1240 |
{
|
| 1241 |
"epoch": 8.49,
|
| 1242 |
-
"learning_rate":
|
| 1243 |
-
"loss": 0.
|
| 1244 |
"step": 970
|
| 1245 |
},
|
| 1246 |
{
|
| 1247 |
"epoch": 8.53,
|
| 1248 |
-
"learning_rate":
|
| 1249 |
-
"loss": 0.
|
| 1250 |
"step": 975
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
"epoch": 8.58,
|
| 1254 |
-
"learning_rate":
|
| 1255 |
-
"loss": 0.
|
| 1256 |
"step": 980
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 8.62,
|
| 1260 |
-
"learning_rate":
|
| 1261 |
-
"loss": 0.
|
| 1262 |
"step": 985
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
"epoch": 8.67,
|
| 1266 |
-
"learning_rate":
|
| 1267 |
-
"loss": 0.
|
| 1268 |
"step": 990
|
| 1269 |
},
|
| 1270 |
{
|
| 1271 |
"epoch": 8.71,
|
| 1272 |
-
"learning_rate":
|
| 1273 |
-
"loss": 0.
|
| 1274 |
"step": 995
|
| 1275 |
},
|
| 1276 |
{
|
| 1277 |
"epoch": 8.75,
|
| 1278 |
-
"learning_rate":
|
| 1279 |
-
"loss": 0.
|
| 1280 |
"step": 1000
|
| 1281 |
},
|
| 1282 |
{
|
| 1283 |
"epoch": 8.8,
|
| 1284 |
-
"learning_rate":
|
| 1285 |
-
"loss": 0.
|
| 1286 |
"step": 1005
|
| 1287 |
},
|
| 1288 |
{
|
| 1289 |
"epoch": 8.84,
|
| 1290 |
-
"learning_rate":
|
| 1291 |
-
"loss": 0.
|
| 1292 |
"step": 1010
|
| 1293 |
},
|
| 1294 |
{
|
| 1295 |
"epoch": 8.88,
|
| 1296 |
-
"learning_rate":
|
| 1297 |
-
"loss": 0.
|
| 1298 |
"step": 1015
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 8.93,
|
| 1302 |
-
"learning_rate":
|
| 1303 |
-
"loss": 0.
|
| 1304 |
"step": 1020
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
"epoch": 8.97,
|
| 1308 |
-
"learning_rate":
|
| 1309 |
-
"loss": 0.
|
| 1310 |
"step": 1025
|
| 1311 |
},
|
| 1312 |
{
|
| 1313 |
"epoch": 9.0,
|
| 1314 |
-
"eval_accuracy": 0.
|
| 1315 |
-
"eval_loss": 0.
|
| 1316 |
-
"eval_runtime":
|
| 1317 |
-
"eval_samples_per_second":
|
| 1318 |
-
"eval_steps_per_second": 2.
|
| 1319 |
"step": 1028
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 9.02,
|
| 1323 |
-
"learning_rate":
|
| 1324 |
-
"loss": 0.
|
| 1325 |
"step": 1030
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
"epoch": 9.06,
|
| 1329 |
-
"learning_rate":
|
| 1330 |
-
"loss": 0.
|
| 1331 |
"step": 1035
|
| 1332 |
},
|
| 1333 |
{
|
| 1334 |
"epoch": 9.1,
|
| 1335 |
-
"learning_rate":
|
| 1336 |
-
"loss": 0.
|
| 1337 |
"step": 1040
|
| 1338 |
},
|
| 1339 |
{
|
| 1340 |
"epoch": 9.15,
|
| 1341 |
-
"learning_rate":
|
| 1342 |
-
"loss": 0.
|
| 1343 |
"step": 1045
|
| 1344 |
},
|
| 1345 |
{
|
| 1346 |
"epoch": 9.19,
|
| 1347 |
-
"learning_rate":
|
| 1348 |
-
"loss": 0.
|
| 1349 |
"step": 1050
|
| 1350 |
},
|
| 1351 |
{
|
| 1352 |
"epoch": 9.23,
|
| 1353 |
-
"learning_rate":
|
| 1354 |
-
"loss": 0.
|
| 1355 |
"step": 1055
|
| 1356 |
},
|
| 1357 |
{
|
| 1358 |
"epoch": 9.28,
|
| 1359 |
-
"learning_rate":
|
| 1360 |
-
"loss": 0.
|
| 1361 |
"step": 1060
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 9.32,
|
| 1365 |
-
"learning_rate":
|
| 1366 |
-
"loss": 0.
|
| 1367 |
"step": 1065
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
"epoch": 9.37,
|
| 1371 |
-
"learning_rate":
|
| 1372 |
-
"loss": 0.
|
| 1373 |
"step": 1070
|
| 1374 |
},
|
| 1375 |
{
|
| 1376 |
"epoch": 9.41,
|
| 1377 |
-
"learning_rate":
|
| 1378 |
-
"loss": 0.
|
| 1379 |
"step": 1075
|
| 1380 |
},
|
| 1381 |
{
|
| 1382 |
"epoch": 9.45,
|
| 1383 |
-
"learning_rate":
|
| 1384 |
-
"loss": 0.
|
| 1385 |
"step": 1080
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
"epoch": 9.5,
|
| 1389 |
-
"learning_rate":
|
| 1390 |
-
"loss": 0.
|
| 1391 |
"step": 1085
|
| 1392 |
},
|
| 1393 |
{
|
| 1394 |
"epoch": 9.54,
|
| 1395 |
-
"learning_rate":
|
| 1396 |
-
"loss": 0.
|
| 1397 |
"step": 1090
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
"epoch": 9.58,
|
| 1401 |
-
"learning_rate": 3.
|
| 1402 |
-
"loss": 0.
|
| 1403 |
"step": 1095
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 9.63,
|
| 1407 |
-
"learning_rate": 3.
|
| 1408 |
-
"loss": 0.
|
| 1409 |
"step": 1100
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
"epoch": 9.67,
|
| 1413 |
-
"learning_rate": 3.
|
| 1414 |
-
"loss": 0.
|
| 1415 |
"step": 1105
|
| 1416 |
},
|
| 1417 |
{
|
| 1418 |
"epoch": 9.72,
|
| 1419 |
-
"learning_rate":
|
| 1420 |
-
"loss": 0.
|
| 1421 |
"step": 1110
|
| 1422 |
},
|
| 1423 |
{
|
| 1424 |
"epoch": 9.76,
|
| 1425 |
-
"learning_rate":
|
| 1426 |
-
"loss": 0.
|
| 1427 |
"step": 1115
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 9.8,
|
| 1431 |
-
"learning_rate":
|
| 1432 |
-
"loss": 0.
|
| 1433 |
"step": 1120
|
| 1434 |
},
|
| 1435 |
{
|
| 1436 |
"epoch": 9.85,
|
| 1437 |
-
"learning_rate":
|
| 1438 |
-
"loss": 0.
|
| 1439 |
"step": 1125
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
"epoch": 9.89,
|
| 1443 |
-
"learning_rate":
|
| 1444 |
-
"loss": 0.
|
| 1445 |
"step": 1130
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 9.93,
|
| 1449 |
-
"learning_rate":
|
| 1450 |
-
"loss": 0.
|
| 1451 |
"step": 1135
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
"epoch": 9.98,
|
| 1455 |
-
"learning_rate":
|
| 1456 |
-
"loss": 0.
|
| 1457 |
"step": 1140
|
| 1458 |
},
|
| 1459 |
{
|
| 1460 |
-
"epoch":
|
| 1461 |
-
"eval_accuracy": 0.
|
| 1462 |
-
"eval_loss": 0.
|
| 1463 |
-
"eval_runtime":
|
| 1464 |
-
"eval_samples_per_second":
|
| 1465 |
-
"eval_steps_per_second": 2.
|
| 1466 |
-
"step":
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 10.02,
|
| 1470 |
-
"learning_rate": 3.304093567251462e-05,
|
| 1471 |
-
"loss": 0.1459,
|
| 1472 |
-
"step": 1145
|
| 1473 |
-
},
|
| 1474 |
-
{
|
| 1475 |
-
"epoch": 10.07,
|
| 1476 |
-
"learning_rate": 3.274853801169591e-05,
|
| 1477 |
-
"loss": 0.1303,
|
| 1478 |
-
"step": 1150
|
| 1479 |
-
},
|
| 1480 |
-
{
|
| 1481 |
-
"epoch": 10.11,
|
| 1482 |
-
"learning_rate": 3.24561403508772e-05,
|
| 1483 |
-
"loss": 0.2256,
|
| 1484 |
-
"step": 1155
|
| 1485 |
-
},
|
| 1486 |
-
{
|
| 1487 |
-
"epoch": 10.15,
|
| 1488 |
-
"learning_rate": 3.216374269005848e-05,
|
| 1489 |
-
"loss": 0.2763,
|
| 1490 |
-
"step": 1160
|
| 1491 |
-
},
|
| 1492 |
-
{
|
| 1493 |
-
"epoch": 10.2,
|
| 1494 |
-
"learning_rate": 3.187134502923977e-05,
|
| 1495 |
-
"loss": 0.1605,
|
| 1496 |
-
"step": 1165
|
| 1497 |
-
},
|
| 1498 |
-
{
|
| 1499 |
-
"epoch": 10.24,
|
| 1500 |
-
"learning_rate": 3.157894736842105e-05,
|
| 1501 |
-
"loss": 0.14,
|
| 1502 |
-
"step": 1170
|
| 1503 |
-
},
|
| 1504 |
-
{
|
| 1505 |
-
"epoch": 10.28,
|
| 1506 |
-
"learning_rate": 3.128654970760234e-05,
|
| 1507 |
-
"loss": 0.1832,
|
| 1508 |
-
"step": 1175
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 10.33,
|
| 1512 |
-
"learning_rate": 3.0994152046783626e-05,
|
| 1513 |
-
"loss": 0.1387,
|
| 1514 |
-
"step": 1180
|
| 1515 |
-
},
|
| 1516 |
-
{
|
| 1517 |
-
"epoch": 10.37,
|
| 1518 |
-
"learning_rate": 3.0701754385964913e-05,
|
| 1519 |
-
"loss": 0.1055,
|
| 1520 |
-
"step": 1185
|
| 1521 |
-
},
|
| 1522 |
-
{
|
| 1523 |
-
"epoch": 10.42,
|
| 1524 |
-
"learning_rate": 3.0409356725146197e-05,
|
| 1525 |
-
"loss": 0.1302,
|
| 1526 |
-
"step": 1190
|
| 1527 |
-
},
|
| 1528 |
-
{
|
| 1529 |
-
"epoch": 10.46,
|
| 1530 |
-
"learning_rate": 3.0116959064327488e-05,
|
| 1531 |
-
"loss": 0.0993,
|
| 1532 |
-
"step": 1195
|
| 1533 |
-
},
|
| 1534 |
-
{
|
| 1535 |
-
"epoch": 10.5,
|
| 1536 |
-
"learning_rate": 2.9824561403508772e-05,
|
| 1537 |
-
"loss": 0.174,
|
| 1538 |
-
"step": 1200
|
| 1539 |
-
},
|
| 1540 |
-
{
|
| 1541 |
-
"epoch": 10.55,
|
| 1542 |
-
"learning_rate": 2.9532163742690062e-05,
|
| 1543 |
-
"loss": 0.1348,
|
| 1544 |
-
"step": 1205
|
| 1545 |
-
},
|
| 1546 |
-
{
|
| 1547 |
-
"epoch": 10.59,
|
| 1548 |
-
"learning_rate": 2.9239766081871346e-05,
|
| 1549 |
-
"loss": 0.12,
|
| 1550 |
-
"step": 1210
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 10.63,
|
| 1554 |
-
"learning_rate": 2.8947368421052634e-05,
|
| 1555 |
-
"loss": 0.2483,
|
| 1556 |
-
"step": 1215
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 10.68,
|
| 1560 |
-
"learning_rate": 2.8654970760233917e-05,
|
| 1561 |
-
"loss": 0.0965,
|
| 1562 |
-
"step": 1220
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 10.72,
|
| 1566 |
-
"learning_rate": 2.8362573099415208e-05,
|
| 1567 |
-
"loss": 0.1754,
|
| 1568 |
-
"step": 1225
|
| 1569 |
-
},
|
| 1570 |
-
{
|
| 1571 |
-
"epoch": 10.77,
|
| 1572 |
-
"learning_rate": 2.8070175438596492e-05,
|
| 1573 |
-
"loss": 0.1192,
|
| 1574 |
-
"step": 1230
|
| 1575 |
-
},
|
| 1576 |
-
{
|
| 1577 |
-
"epoch": 10.81,
|
| 1578 |
-
"learning_rate": 2.777777777777778e-05,
|
| 1579 |
-
"loss": 0.1869,
|
| 1580 |
-
"step": 1235
|
| 1581 |
-
},
|
| 1582 |
-
{
|
| 1583 |
-
"epoch": 10.85,
|
| 1584 |
-
"learning_rate": 2.7485380116959063e-05,
|
| 1585 |
-
"loss": 0.0549,
|
| 1586 |
-
"step": 1240
|
| 1587 |
-
},
|
| 1588 |
-
{
|
| 1589 |
-
"epoch": 10.9,
|
| 1590 |
-
"learning_rate": 2.7192982456140354e-05,
|
| 1591 |
-
"loss": 0.1127,
|
| 1592 |
-
"step": 1245
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 10.94,
|
| 1596 |
-
"learning_rate": 2.6900584795321637e-05,
|
| 1597 |
-
"loss": 0.1294,
|
| 1598 |
-
"step": 1250
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 10.98,
|
| 1602 |
-
"learning_rate": 2.6608187134502928e-05,
|
| 1603 |
-
"loss": 0.1147,
|
| 1604 |
-
"step": 1255
|
| 1605 |
-
},
|
| 1606 |
-
{
|
| 1607 |
-
"epoch": 10.99,
|
| 1608 |
-
"eval_accuracy": 0.9835796387520526,
|
| 1609 |
-
"eval_loss": 0.05173669755458832,
|
| 1610 |
-
"eval_runtime": 54.2024,
|
| 1611 |
-
"eval_samples_per_second": 22.471,
|
| 1612 |
-
"eval_steps_per_second": 2.823,
|
| 1613 |
-
"step": 1256
|
| 1614 |
-
},
|
| 1615 |
-
{
|
| 1616 |
-
"epoch": 11.03,
|
| 1617 |
-
"learning_rate": 2.6315789473684212e-05,
|
| 1618 |
-
"loss": 0.1873,
|
| 1619 |
-
"step": 1260
|
| 1620 |
-
},
|
| 1621 |
-
{
|
| 1622 |
-
"epoch": 11.07,
|
| 1623 |
-
"learning_rate": 2.60233918128655e-05,
|
| 1624 |
-
"loss": 0.1212,
|
| 1625 |
-
"step": 1265
|
| 1626 |
-
},
|
| 1627 |
-
{
|
| 1628 |
-
"epoch": 11.12,
|
| 1629 |
-
"learning_rate": 2.5730994152046783e-05,
|
| 1630 |
-
"loss": 0.0422,
|
| 1631 |
-
"step": 1270
|
| 1632 |
-
},
|
| 1633 |
-
{
|
| 1634 |
-
"epoch": 11.16,
|
| 1635 |
-
"learning_rate": 2.5438596491228074e-05,
|
| 1636 |
-
"loss": 0.1971,
|
| 1637 |
-
"step": 1275
|
| 1638 |
-
},
|
| 1639 |
-
{
|
| 1640 |
-
"epoch": 11.2,
|
| 1641 |
-
"learning_rate": 2.5146198830409358e-05,
|
| 1642 |
-
"loss": 0.0876,
|
| 1643 |
-
"step": 1280
|
| 1644 |
-
},
|
| 1645 |
-
{
|
| 1646 |
-
"epoch": 11.25,
|
| 1647 |
-
"learning_rate": 2.485380116959064e-05,
|
| 1648 |
-
"loss": 0.1424,
|
| 1649 |
-
"step": 1285
|
| 1650 |
-
},
|
| 1651 |
-
{
|
| 1652 |
-
"epoch": 11.29,
|
| 1653 |
-
"learning_rate": 2.456140350877193e-05,
|
| 1654 |
-
"loss": 0.0589,
|
| 1655 |
-
"step": 1290
|
| 1656 |
-
},
|
| 1657 |
-
{
|
| 1658 |
-
"epoch": 11.33,
|
| 1659 |
-
"learning_rate": 2.4269005847953216e-05,
|
| 1660 |
-
"loss": 0.0975,
|
| 1661 |
-
"step": 1295
|
| 1662 |
-
},
|
| 1663 |
-
{
|
| 1664 |
-
"epoch": 11.38,
|
| 1665 |
-
"learning_rate": 2.3976608187134503e-05,
|
| 1666 |
-
"loss": 0.1219,
|
| 1667 |
-
"step": 1300
|
| 1668 |
-
},
|
| 1669 |
-
{
|
| 1670 |
-
"epoch": 11.42,
|
| 1671 |
-
"learning_rate": 2.368421052631579e-05,
|
| 1672 |
-
"loss": 0.1359,
|
| 1673 |
-
"step": 1305
|
| 1674 |
-
},
|
| 1675 |
-
{
|
| 1676 |
-
"epoch": 11.47,
|
| 1677 |
-
"learning_rate": 2.3391812865497074e-05,
|
| 1678 |
-
"loss": 0.0449,
|
| 1679 |
-
"step": 1310
|
| 1680 |
-
},
|
| 1681 |
-
{
|
| 1682 |
-
"epoch": 11.51,
|
| 1683 |
-
"learning_rate": 2.309941520467836e-05,
|
| 1684 |
-
"loss": 0.1362,
|
| 1685 |
-
"step": 1315
|
| 1686 |
-
},
|
| 1687 |
-
{
|
| 1688 |
-
"epoch": 11.55,
|
| 1689 |
-
"learning_rate": 2.280701754385965e-05,
|
| 1690 |
-
"loss": 0.1156,
|
| 1691 |
-
"step": 1320
|
| 1692 |
-
},
|
| 1693 |
-
{
|
| 1694 |
-
"epoch": 11.6,
|
| 1695 |
-
"learning_rate": 2.2514619883040936e-05,
|
| 1696 |
-
"loss": 0.0743,
|
| 1697 |
-
"step": 1325
|
| 1698 |
-
},
|
| 1699 |
-
{
|
| 1700 |
-
"epoch": 11.64,
|
| 1701 |
-
"learning_rate": 2.2222222222222223e-05,
|
| 1702 |
-
"loss": 0.2625,
|
| 1703 |
-
"step": 1330
|
| 1704 |
-
},
|
| 1705 |
-
{
|
| 1706 |
-
"epoch": 11.68,
|
| 1707 |
-
"learning_rate": 2.1929824561403507e-05,
|
| 1708 |
-
"loss": 0.065,
|
| 1709 |
-
"step": 1335
|
| 1710 |
-
},
|
| 1711 |
-
{
|
| 1712 |
-
"epoch": 11.73,
|
| 1713 |
-
"learning_rate": 2.1637426900584794e-05,
|
| 1714 |
-
"loss": 0.1152,
|
| 1715 |
-
"step": 1340
|
| 1716 |
-
},
|
| 1717 |
-
{
|
| 1718 |
-
"epoch": 11.77,
|
| 1719 |
-
"learning_rate": 2.134502923976608e-05,
|
| 1720 |
-
"loss": 0.1055,
|
| 1721 |
-
"step": 1345
|
| 1722 |
-
},
|
| 1723 |
-
{
|
| 1724 |
-
"epoch": 11.82,
|
| 1725 |
-
"learning_rate": 2.105263157894737e-05,
|
| 1726 |
-
"loss": 0.0941,
|
| 1727 |
-
"step": 1350
|
| 1728 |
-
},
|
| 1729 |
-
{
|
| 1730 |
-
"epoch": 11.86,
|
| 1731 |
-
"learning_rate": 2.0760233918128656e-05,
|
| 1732 |
-
"loss": 0.1564,
|
| 1733 |
-
"step": 1355
|
| 1734 |
-
},
|
| 1735 |
-
{
|
| 1736 |
-
"epoch": 11.9,
|
| 1737 |
-
"learning_rate": 2.046783625730994e-05,
|
| 1738 |
-
"loss": 0.151,
|
| 1739 |
-
"step": 1360
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 11.95,
|
| 1743 |
-
"learning_rate": 2.0175438596491227e-05,
|
| 1744 |
-
"loss": 0.0964,
|
| 1745 |
-
"step": 1365
|
| 1746 |
-
},
|
| 1747 |
-
{
|
| 1748 |
-
"epoch": 11.99,
|
| 1749 |
-
"learning_rate": 1.9883040935672515e-05,
|
| 1750 |
-
"loss": 0.1663,
|
| 1751 |
-
"step": 1370
|
| 1752 |
-
},
|
| 1753 |
-
{
|
| 1754 |
-
"epoch": 12.0,
|
| 1755 |
-
"eval_accuracy": 0.9958949096880131,
|
| 1756 |
-
"eval_loss": 0.005566800944507122,
|
| 1757 |
-
"eval_runtime": 54.1153,
|
| 1758 |
-
"eval_samples_per_second": 22.507,
|
| 1759 |
-
"eval_steps_per_second": 2.827,
|
| 1760 |
-
"step": 1371
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 12.04,
|
| 1764 |
-
"learning_rate": 1.9590643274853802e-05,
|
| 1765 |
-
"loss": 0.0835,
|
| 1766 |
-
"step": 1375
|
| 1767 |
-
},
|
| 1768 |
-
{
|
| 1769 |
-
"epoch": 12.08,
|
| 1770 |
-
"learning_rate": 1.929824561403509e-05,
|
| 1771 |
-
"loss": 0.0809,
|
| 1772 |
-
"step": 1380
|
| 1773 |
-
},
|
| 1774 |
-
{
|
| 1775 |
-
"epoch": 12.12,
|
| 1776 |
-
"learning_rate": 1.9005847953216373e-05,
|
| 1777 |
-
"loss": 0.1657,
|
| 1778 |
-
"step": 1385
|
| 1779 |
-
},
|
| 1780 |
-
{
|
| 1781 |
-
"epoch": 12.17,
|
| 1782 |
-
"learning_rate": 1.871345029239766e-05,
|
| 1783 |
-
"loss": 0.0642,
|
| 1784 |
-
"step": 1390
|
| 1785 |
-
},
|
| 1786 |
-
{
|
| 1787 |
-
"epoch": 12.21,
|
| 1788 |
-
"learning_rate": 1.8421052631578947e-05,
|
| 1789 |
-
"loss": 0.1124,
|
| 1790 |
-
"step": 1395
|
| 1791 |
-
},
|
| 1792 |
-
{
|
| 1793 |
-
"epoch": 12.25,
|
| 1794 |
-
"learning_rate": 1.8128654970760235e-05,
|
| 1795 |
-
"loss": 0.1444,
|
| 1796 |
-
"step": 1400
|
| 1797 |
-
},
|
| 1798 |
-
{
|
| 1799 |
-
"epoch": 12.3,
|
| 1800 |
-
"learning_rate": 1.7836257309941522e-05,
|
| 1801 |
-
"loss": 0.0315,
|
| 1802 |
-
"step": 1405
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 12.34,
|
| 1806 |
-
"learning_rate": 1.7543859649122806e-05,
|
| 1807 |
-
"loss": 0.1471,
|
| 1808 |
-
"step": 1410
|
| 1809 |
-
},
|
| 1810 |
-
{
|
| 1811 |
-
"epoch": 12.39,
|
| 1812 |
-
"learning_rate": 1.7251461988304093e-05,
|
| 1813 |
-
"loss": 0.091,
|
| 1814 |
-
"step": 1415
|
| 1815 |
-
},
|
| 1816 |
-
{
|
| 1817 |
-
"epoch": 12.43,
|
| 1818 |
-
"learning_rate": 1.695906432748538e-05,
|
| 1819 |
-
"loss": 0.167,
|
| 1820 |
-
"step": 1420
|
| 1821 |
-
},
|
| 1822 |
-
{
|
| 1823 |
-
"epoch": 12.47,
|
| 1824 |
-
"learning_rate": 1.6666666666666667e-05,
|
| 1825 |
-
"loss": 0.0769,
|
| 1826 |
-
"step": 1425
|
| 1827 |
-
},
|
| 1828 |
-
{
|
| 1829 |
-
"epoch": 12.52,
|
| 1830 |
-
"learning_rate": 1.6374269005847955e-05,
|
| 1831 |
-
"loss": 0.1187,
|
| 1832 |
-
"step": 1430
|
| 1833 |
-
},
|
| 1834 |
-
{
|
| 1835 |
-
"epoch": 12.56,
|
| 1836 |
-
"learning_rate": 1.608187134502924e-05,
|
| 1837 |
-
"loss": 0.1473,
|
| 1838 |
-
"step": 1435
|
| 1839 |
-
},
|
| 1840 |
-
{
|
| 1841 |
-
"epoch": 12.6,
|
| 1842 |
-
"learning_rate": 1.5789473684210526e-05,
|
| 1843 |
-
"loss": 0.0737,
|
| 1844 |
-
"step": 1440
|
| 1845 |
-
},
|
| 1846 |
-
{
|
| 1847 |
-
"epoch": 12.65,
|
| 1848 |
-
"learning_rate": 1.5497076023391813e-05,
|
| 1849 |
-
"loss": 0.0668,
|
| 1850 |
-
"step": 1445
|
| 1851 |
-
},
|
| 1852 |
-
{
|
| 1853 |
-
"epoch": 12.69,
|
| 1854 |
-
"learning_rate": 1.5204678362573099e-05,
|
| 1855 |
-
"loss": 0.1488,
|
| 1856 |
-
"step": 1450
|
| 1857 |
-
},
|
| 1858 |
-
{
|
| 1859 |
-
"epoch": 12.74,
|
| 1860 |
-
"learning_rate": 1.4912280701754386e-05,
|
| 1861 |
-
"loss": 0.0413,
|
| 1862 |
-
"step": 1455
|
| 1863 |
-
},
|
| 1864 |
-
{
|
| 1865 |
-
"epoch": 12.78,
|
| 1866 |
-
"learning_rate": 1.4619883040935673e-05,
|
| 1867 |
-
"loss": 0.1062,
|
| 1868 |
-
"step": 1460
|
| 1869 |
-
},
|
| 1870 |
-
{
|
| 1871 |
-
"epoch": 12.82,
|
| 1872 |
-
"learning_rate": 1.4327485380116959e-05,
|
| 1873 |
-
"loss": 0.1047,
|
| 1874 |
-
"step": 1465
|
| 1875 |
-
},
|
| 1876 |
-
{
|
| 1877 |
-
"epoch": 12.87,
|
| 1878 |
-
"learning_rate": 1.4035087719298246e-05,
|
| 1879 |
-
"loss": 0.068,
|
| 1880 |
-
"step": 1470
|
| 1881 |
-
},
|
| 1882 |
-
{
|
| 1883 |
-
"epoch": 12.91,
|
| 1884 |
-
"learning_rate": 1.3742690058479531e-05,
|
| 1885 |
-
"loss": 0.03,
|
| 1886 |
-
"step": 1475
|
| 1887 |
-
},
|
| 1888 |
-
{
|
| 1889 |
-
"epoch": 12.95,
|
| 1890 |
-
"learning_rate": 1.3450292397660819e-05,
|
| 1891 |
-
"loss": 0.0708,
|
| 1892 |
-
"step": 1480
|
| 1893 |
-
},
|
| 1894 |
-
{
|
| 1895 |
-
"epoch": 13.0,
|
| 1896 |
-
"learning_rate": 1.3157894736842106e-05,
|
| 1897 |
-
"loss": 0.094,
|
| 1898 |
-
"step": 1485
|
| 1899 |
-
},
|
| 1900 |
-
{
|
| 1901 |
-
"epoch": 13.0,
|
| 1902 |
-
"eval_accuracy": 0.9991789819376026,
|
| 1903 |
-
"eval_loss": 0.002953051822260022,
|
| 1904 |
-
"eval_runtime": 54.2489,
|
| 1905 |
-
"eval_samples_per_second": 22.452,
|
| 1906 |
-
"eval_steps_per_second": 2.82,
|
| 1907 |
-
"step": 1485
|
| 1908 |
-
},
|
| 1909 |
-
{
|
| 1910 |
-
"epoch": 13.04,
|
| 1911 |
-
"learning_rate": 1.2865497076023392e-05,
|
| 1912 |
-
"loss": 0.1298,
|
| 1913 |
-
"step": 1490
|
| 1914 |
-
},
|
| 1915 |
-
{
|
| 1916 |
-
"epoch": 13.09,
|
| 1917 |
-
"learning_rate": 1.2573099415204679e-05,
|
| 1918 |
-
"loss": 0.1349,
|
| 1919 |
-
"step": 1495
|
| 1920 |
-
},
|
| 1921 |
-
{
|
| 1922 |
-
"epoch": 13.13,
|
| 1923 |
-
"learning_rate": 1.2280701754385964e-05,
|
| 1924 |
-
"loss": 0.1136,
|
| 1925 |
-
"step": 1500
|
| 1926 |
-
},
|
| 1927 |
-
{
|
| 1928 |
-
"epoch": 13.17,
|
| 1929 |
-
"learning_rate": 1.1988304093567252e-05,
|
| 1930 |
-
"loss": 0.0972,
|
| 1931 |
-
"step": 1505
|
| 1932 |
-
},
|
| 1933 |
-
{
|
| 1934 |
-
"epoch": 13.22,
|
| 1935 |
-
"learning_rate": 1.1695906432748537e-05,
|
| 1936 |
-
"loss": 0.0589,
|
| 1937 |
-
"step": 1510
|
| 1938 |
-
},
|
| 1939 |
-
{
|
| 1940 |
-
"epoch": 13.26,
|
| 1941 |
-
"learning_rate": 1.1403508771929824e-05,
|
| 1942 |
-
"loss": 0.0569,
|
| 1943 |
-
"step": 1515
|
| 1944 |
-
},
|
| 1945 |
-
{
|
| 1946 |
-
"epoch": 13.3,
|
| 1947 |
-
"learning_rate": 1.1111111111111112e-05,
|
| 1948 |
-
"loss": 0.085,
|
| 1949 |
-
"step": 1520
|
| 1950 |
-
},
|
| 1951 |
-
{
|
| 1952 |
-
"epoch": 13.35,
|
| 1953 |
-
"learning_rate": 1.0818713450292397e-05,
|
| 1954 |
-
"loss": 0.1057,
|
| 1955 |
-
"step": 1525
|
| 1956 |
-
},
|
| 1957 |
-
{
|
| 1958 |
-
"epoch": 13.39,
|
| 1959 |
-
"learning_rate": 1.0526315789473684e-05,
|
| 1960 |
-
"loss": 0.0953,
|
| 1961 |
-
"step": 1530
|
| 1962 |
-
},
|
| 1963 |
-
{
|
| 1964 |
-
"epoch": 13.44,
|
| 1965 |
-
"learning_rate": 1.023391812865497e-05,
|
| 1966 |
-
"loss": 0.085,
|
| 1967 |
-
"step": 1535
|
| 1968 |
-
},
|
| 1969 |
-
{
|
| 1970 |
-
"epoch": 13.48,
|
| 1971 |
-
"learning_rate": 9.941520467836257e-06,
|
| 1972 |
-
"loss": 0.043,
|
| 1973 |
-
"step": 1540
|
| 1974 |
-
},
|
| 1975 |
-
{
|
| 1976 |
-
"epoch": 13.52,
|
| 1977 |
-
"learning_rate": 9.649122807017545e-06,
|
| 1978 |
-
"loss": 0.0555,
|
| 1979 |
-
"step": 1545
|
| 1980 |
-
},
|
| 1981 |
-
{
|
| 1982 |
-
"epoch": 13.57,
|
| 1983 |
-
"learning_rate": 9.35672514619883e-06,
|
| 1984 |
-
"loss": 0.0452,
|
| 1985 |
-
"step": 1550
|
| 1986 |
-
},
|
| 1987 |
-
{
|
| 1988 |
-
"epoch": 13.61,
|
| 1989 |
-
"learning_rate": 9.064327485380117e-06,
|
| 1990 |
-
"loss": 0.0489,
|
| 1991 |
-
"step": 1555
|
| 1992 |
-
},
|
| 1993 |
-
{
|
| 1994 |
-
"epoch": 13.65,
|
| 1995 |
-
"learning_rate": 8.771929824561403e-06,
|
| 1996 |
-
"loss": 0.1405,
|
| 1997 |
-
"step": 1560
|
| 1998 |
-
},
|
| 1999 |
-
{
|
| 2000 |
-
"epoch": 13.7,
|
| 2001 |
-
"learning_rate": 8.47953216374269e-06,
|
| 2002 |
-
"loss": 0.0661,
|
| 2003 |
-
"step": 1565
|
| 2004 |
-
},
|
| 2005 |
-
{
|
| 2006 |
-
"epoch": 13.74,
|
| 2007 |
-
"learning_rate": 8.187134502923977e-06,
|
| 2008 |
-
"loss": 0.1128,
|
| 2009 |
-
"step": 1570
|
| 2010 |
-
},
|
| 2011 |
-
{
|
| 2012 |
-
"epoch": 13.79,
|
| 2013 |
-
"learning_rate": 7.894736842105263e-06,
|
| 2014 |
-
"loss": 0.0803,
|
| 2015 |
-
"step": 1575
|
| 2016 |
-
},
|
| 2017 |
-
{
|
| 2018 |
-
"epoch": 13.83,
|
| 2019 |
-
"learning_rate": 7.602339181286549e-06,
|
| 2020 |
-
"loss": 0.0843,
|
| 2021 |
-
"step": 1580
|
| 2022 |
-
},
|
| 2023 |
-
{
|
| 2024 |
-
"epoch": 13.87,
|
| 2025 |
-
"learning_rate": 7.3099415204678366e-06,
|
| 2026 |
-
"loss": 0.0902,
|
| 2027 |
-
"step": 1585
|
| 2028 |
-
},
|
| 2029 |
-
{
|
| 2030 |
-
"epoch": 13.92,
|
| 2031 |
-
"learning_rate": 7.017543859649123e-06,
|
| 2032 |
-
"loss": 0.094,
|
| 2033 |
-
"step": 1590
|
| 2034 |
-
},
|
| 2035 |
-
{
|
| 2036 |
-
"epoch": 13.96,
|
| 2037 |
-
"learning_rate": 6.725146198830409e-06,
|
| 2038 |
-
"loss": 0.1308,
|
| 2039 |
-
"step": 1595
|
| 2040 |
-
},
|
| 2041 |
-
{
|
| 2042 |
-
"epoch": 14.0,
|
| 2043 |
-
"eval_accuracy": 0.9991789819376026,
|
| 2044 |
-
"eval_loss": 0.0011125708697363734,
|
| 2045 |
-
"eval_runtime": 54.0922,
|
| 2046 |
-
"eval_samples_per_second": 22.517,
|
| 2047 |
-
"eval_steps_per_second": 2.829,
|
| 2048 |
-
"step": 1599
|
| 2049 |
-
},
|
| 2050 |
-
{
|
| 2051 |
-
"epoch": 14.0,
|
| 2052 |
-
"learning_rate": 6.432748538011696e-06,
|
| 2053 |
-
"loss": 0.0785,
|
| 2054 |
-
"step": 1600
|
| 2055 |
-
},
|
| 2056 |
-
{
|
| 2057 |
-
"epoch": 14.05,
|
| 2058 |
-
"learning_rate": 6.140350877192982e-06,
|
| 2059 |
-
"loss": 0.079,
|
| 2060 |
-
"step": 1605
|
| 2061 |
-
},
|
| 2062 |
-
{
|
| 2063 |
-
"epoch": 14.09,
|
| 2064 |
-
"learning_rate": 5.8479532163742686e-06,
|
| 2065 |
-
"loss": 0.1075,
|
| 2066 |
-
"step": 1610
|
| 2067 |
-
},
|
| 2068 |
-
{
|
| 2069 |
-
"epoch": 14.14,
|
| 2070 |
-
"learning_rate": 5.555555555555556e-06,
|
| 2071 |
-
"loss": 0.0769,
|
| 2072 |
-
"step": 1615
|
| 2073 |
-
},
|
| 2074 |
-
{
|
| 2075 |
-
"epoch": 14.18,
|
| 2076 |
-
"learning_rate": 5.263157894736842e-06,
|
| 2077 |
-
"loss": 0.0705,
|
| 2078 |
-
"step": 1620
|
| 2079 |
-
},
|
| 2080 |
-
{
|
| 2081 |
-
"epoch": 14.22,
|
| 2082 |
-
"learning_rate": 4.970760233918129e-06,
|
| 2083 |
-
"loss": 0.0779,
|
| 2084 |
-
"step": 1625
|
| 2085 |
-
},
|
| 2086 |
-
{
|
| 2087 |
-
"epoch": 14.27,
|
| 2088 |
-
"learning_rate": 4.678362573099415e-06,
|
| 2089 |
-
"loss": 0.0319,
|
| 2090 |
-
"step": 1630
|
| 2091 |
-
},
|
| 2092 |
-
{
|
| 2093 |
-
"epoch": 14.31,
|
| 2094 |
-
"learning_rate": 4.3859649122807014e-06,
|
| 2095 |
-
"loss": 0.0495,
|
| 2096 |
-
"step": 1635
|
| 2097 |
-
},
|
| 2098 |
-
{
|
| 2099 |
-
"epoch": 14.35,
|
| 2100 |
-
"learning_rate": 4.093567251461989e-06,
|
| 2101 |
-
"loss": 0.0851,
|
| 2102 |
-
"step": 1640
|
| 2103 |
-
},
|
| 2104 |
-
{
|
| 2105 |
-
"epoch": 14.4,
|
| 2106 |
-
"learning_rate": 3.8011695906432747e-06,
|
| 2107 |
-
"loss": 0.1291,
|
| 2108 |
-
"step": 1645
|
| 2109 |
-
},
|
| 2110 |
-
{
|
| 2111 |
-
"epoch": 14.44,
|
| 2112 |
-
"learning_rate": 3.5087719298245615e-06,
|
| 2113 |
-
"loss": 0.0778,
|
| 2114 |
-
"step": 1650
|
| 2115 |
-
},
|
| 2116 |
-
{
|
| 2117 |
-
"epoch": 14.49,
|
| 2118 |
-
"learning_rate": 3.216374269005848e-06,
|
| 2119 |
-
"loss": 0.0826,
|
| 2120 |
-
"step": 1655
|
| 2121 |
-
},
|
| 2122 |
-
{
|
| 2123 |
-
"epoch": 14.53,
|
| 2124 |
-
"learning_rate": 2.9239766081871343e-06,
|
| 2125 |
-
"loss": 0.0836,
|
| 2126 |
-
"step": 1660
|
| 2127 |
-
},
|
| 2128 |
-
{
|
| 2129 |
-
"epoch": 14.57,
|
| 2130 |
-
"learning_rate": 2.631578947368421e-06,
|
| 2131 |
-
"loss": 0.0562,
|
| 2132 |
-
"step": 1665
|
| 2133 |
-
},
|
| 2134 |
-
{
|
| 2135 |
-
"epoch": 14.62,
|
| 2136 |
-
"learning_rate": 2.3391812865497075e-06,
|
| 2137 |
-
"loss": 0.065,
|
| 2138 |
-
"step": 1670
|
| 2139 |
-
},
|
| 2140 |
-
{
|
| 2141 |
-
"epoch": 14.66,
|
| 2142 |
-
"learning_rate": 2.0467836257309943e-06,
|
| 2143 |
-
"loss": 0.0644,
|
| 2144 |
-
"step": 1675
|
| 2145 |
-
},
|
| 2146 |
-
{
|
| 2147 |
-
"epoch": 14.7,
|
| 2148 |
-
"learning_rate": 1.7543859649122807e-06,
|
| 2149 |
-
"loss": 0.056,
|
| 2150 |
-
"step": 1680
|
| 2151 |
-
},
|
| 2152 |
-
{
|
| 2153 |
-
"epoch": 14.75,
|
| 2154 |
-
"learning_rate": 1.4619883040935671e-06,
|
| 2155 |
-
"loss": 0.1165,
|
| 2156 |
-
"step": 1685
|
| 2157 |
-
},
|
| 2158 |
-
{
|
| 2159 |
-
"epoch": 14.79,
|
| 2160 |
-
"learning_rate": 1.1695906432748538e-06,
|
| 2161 |
-
"loss": 0.0744,
|
| 2162 |
-
"step": 1690
|
| 2163 |
-
},
|
| 2164 |
-
{
|
| 2165 |
-
"epoch": 14.84,
|
| 2166 |
-
"learning_rate": 8.771929824561404e-07,
|
| 2167 |
-
"loss": 0.0856,
|
| 2168 |
-
"step": 1695
|
| 2169 |
-
},
|
| 2170 |
-
{
|
| 2171 |
-
"epoch": 14.88,
|
| 2172 |
-
"learning_rate": 5.847953216374269e-07,
|
| 2173 |
-
"loss": 0.0425,
|
| 2174 |
-
"step": 1700
|
| 2175 |
-
},
|
| 2176 |
-
{
|
| 2177 |
-
"epoch": 14.92,
|
| 2178 |
-
"learning_rate": 2.9239766081871344e-07,
|
| 2179 |
-
"loss": 0.0536,
|
| 2180 |
-
"step": 1705
|
| 2181 |
},
|
| 2182 |
{
|
| 2183 |
-
"epoch":
|
| 2184 |
-
"
|
| 2185 |
-
"
|
| 2186 |
-
"
|
| 2187 |
-
|
| 2188 |
-
|
| 2189 |
-
"
|
| 2190 |
-
"eval_accuracy": 1.0,
|
| 2191 |
-
"eval_loss": 0.0007466585957445204,
|
| 2192 |
-
"eval_runtime": 54.1734,
|
| 2193 |
-
"eval_samples_per_second": 22.483,
|
| 2194 |
-
"eval_steps_per_second": 2.824,
|
| 2195 |
-
"step": 1710
|
| 2196 |
-
},
|
| 2197 |
-
{
|
| 2198 |
-
"epoch": 14.97,
|
| 2199 |
-
"step": 1710,
|
| 2200 |
-
"total_flos": 1.2584990004785971e+19,
|
| 2201 |
-
"train_loss": 0.38356898541164675,
|
| 2202 |
-
"train_runtime": 11197.9622,
|
| 2203 |
-
"train_samples_per_second": 4.891,
|
| 2204 |
-
"train_steps_per_second": 0.153
|
| 2205 |
}
|
| 2206 |
],
|
| 2207 |
-
"max_steps":
|
| 2208 |
-
"num_train_epochs":
|
| 2209 |
-
"total_flos":
|
| 2210 |
"trial_name": null,
|
| 2211 |
"trial_params": null
|
| 2212 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.9975369458128078,
|
| 3 |
+
"best_model_checkpoint": "swinv2-large-patch4-window12to16-192to256-22kto1k-ft-finetuned-Lesion-Classification-HAM10000-S/checkpoint-1140",
|
| 4 |
+
"epoch": 9.978118161925602,
|
| 5 |
+
"global_step": 1140,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.04,
|
| 12 |
+
"learning_rate": 4.385964912280702e-07,
|
| 13 |
+
"loss": 2.0503,
|
| 14 |
"step": 5
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.09,
|
| 18 |
+
"learning_rate": 8.771929824561404e-07,
|
| 19 |
+
"loss": 1.9978,
|
| 20 |
"step": 10
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.13,
|
| 24 |
+
"learning_rate": 1.3157894736842106e-06,
|
| 25 |
+
"loss": 1.9865,
|
| 26 |
"step": 15
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.18,
|
| 30 |
+
"learning_rate": 1.7543859649122807e-06,
|
| 31 |
+
"loss": 2.0066,
|
| 32 |
"step": 20
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.22,
|
| 36 |
+
"learning_rate": 2.1929824561403507e-06,
|
| 37 |
+
"loss": 1.9463,
|
| 38 |
"step": 25
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.26,
|
| 42 |
+
"learning_rate": 2.631578947368421e-06,
|
| 43 |
+
"loss": 1.94,
|
| 44 |
"step": 30
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.31,
|
| 48 |
+
"learning_rate": 3.070175438596491e-06,
|
| 49 |
+
"loss": 1.9458,
|
| 50 |
"step": 35
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.35,
|
| 54 |
+
"learning_rate": 3.5087719298245615e-06,
|
| 55 |
+
"loss": 1.8351,
|
| 56 |
"step": 40
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.39,
|
| 60 |
+
"learning_rate": 3.9473684210526315e-06,
|
| 61 |
+
"loss": 1.8686,
|
| 62 |
"step": 45
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.44,
|
| 66 |
+
"learning_rate": 4.3859649122807014e-06,
|
| 67 |
+
"loss": 1.8466,
|
| 68 |
"step": 50
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.48,
|
| 72 |
+
"learning_rate": 4.824561403508772e-06,
|
| 73 |
+
"loss": 1.8131,
|
| 74 |
"step": 55
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 0.53,
|
| 78 |
+
"learning_rate": 5.263157894736842e-06,
|
| 79 |
+
"loss": 1.7527,
|
| 80 |
"step": 60
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.57,
|
| 84 |
+
"learning_rate": 5.701754385964912e-06,
|
| 85 |
+
"loss": 1.7692,
|
| 86 |
"step": 65
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.61,
|
| 90 |
+
"learning_rate": 6.140350877192982e-06,
|
| 91 |
+
"loss": 1.6492,
|
| 92 |
"step": 70
|
| 93 |
},
|
| 94 |
{
|
| 95 |
"epoch": 0.66,
|
| 96 |
+
"learning_rate": 6.578947368421053e-06,
|
| 97 |
+
"loss": 1.5949,
|
| 98 |
"step": 75
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.7,
|
| 102 |
+
"learning_rate": 7.017543859649123e-06,
|
| 103 |
+
"loss": 1.5051,
|
| 104 |
"step": 80
|
| 105 |
},
|
| 106 |
{
|
| 107 |
"epoch": 0.74,
|
| 108 |
+
"learning_rate": 7.456140350877193e-06,
|
| 109 |
+
"loss": 1.5077,
|
| 110 |
"step": 85
|
| 111 |
},
|
| 112 |
{
|
| 113 |
"epoch": 0.79,
|
| 114 |
+
"learning_rate": 7.894736842105263e-06,
|
| 115 |
+
"loss": 1.3687,
|
| 116 |
"step": 90
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.83,
|
| 120 |
+
"learning_rate": 8.333333333333334e-06,
|
| 121 |
+
"loss": 1.3231,
|
| 122 |
"step": 95
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.88,
|
| 126 |
+
"learning_rate": 8.771929824561403e-06,
|
| 127 |
+
"loss": 1.3126,
|
| 128 |
"step": 100
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.92,
|
| 132 |
+
"learning_rate": 9.210526315789474e-06,
|
| 133 |
+
"loss": 1.1667,
|
| 134 |
"step": 105
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 0.96,
|
| 138 |
+
"learning_rate": 9.649122807017545e-06,
|
| 139 |
+
"loss": 1.2378,
|
| 140 |
"step": 110
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 1.0,
|
| 144 |
+
"eval_accuracy": 0.5935960591133005,
|
| 145 |
+
"eval_loss": 0.997550368309021,
|
| 146 |
+
"eval_runtime": 959.756,
|
| 147 |
+
"eval_samples_per_second": 1.269,
|
| 148 |
+
"eval_steps_per_second": 0.159,
|
| 149 |
"step": 114
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 1.01,
|
| 153 |
+
"learning_rate": 1.0087719298245614e-05,
|
| 154 |
+
"loss": 1.0408,
|
| 155 |
"step": 115
|
| 156 |
},
|
| 157 |
{
|
| 158 |
"epoch": 1.05,
|
| 159 |
+
"learning_rate": 1.0526315789473684e-05,
|
| 160 |
+
"loss": 1.0499,
|
| 161 |
"step": 120
|
| 162 |
},
|
| 163 |
{
|
| 164 |
"epoch": 1.09,
|
| 165 |
+
"learning_rate": 1.0964912280701754e-05,
|
| 166 |
+
"loss": 0.9867,
|
| 167 |
"step": 125
|
| 168 |
},
|
| 169 |
{
|
| 170 |
"epoch": 1.14,
|
| 171 |
+
"learning_rate": 1.1403508771929824e-05,
|
| 172 |
+
"loss": 0.9692,
|
| 173 |
"step": 130
|
| 174 |
},
|
| 175 |
{
|
| 176 |
"epoch": 1.18,
|
| 177 |
+
"learning_rate": 1.1842105263157895e-05,
|
| 178 |
+
"loss": 1.0219,
|
| 179 |
"step": 135
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 1.23,
|
| 183 |
+
"learning_rate": 1.2280701754385964e-05,
|
| 184 |
+
"loss": 0.9686,
|
| 185 |
"step": 140
|
| 186 |
},
|
| 187 |
{
|
| 188 |
"epoch": 1.27,
|
| 189 |
+
"learning_rate": 1.2719298245614037e-05,
|
| 190 |
+
"loss": 0.942,
|
| 191 |
"step": 145
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 1.31,
|
| 195 |
+
"learning_rate": 1.3157894736842106e-05,
|
| 196 |
+
"loss": 0.9243,
|
| 197 |
"step": 150
|
| 198 |
},
|
| 199 |
{
|
| 200 |
"epoch": 1.36,
|
| 201 |
+
"learning_rate": 1.3596491228070177e-05,
|
| 202 |
+
"loss": 0.791,
|
| 203 |
"step": 155
|
| 204 |
},
|
| 205 |
{
|
| 206 |
"epoch": 1.4,
|
| 207 |
+
"learning_rate": 1.4035087719298246e-05,
|
| 208 |
+
"loss": 0.8906,
|
| 209 |
"step": 160
|
| 210 |
},
|
| 211 |
{
|
| 212 |
"epoch": 1.44,
|
| 213 |
+
"learning_rate": 1.4473684210526317e-05,
|
| 214 |
+
"loss": 0.7858,
|
| 215 |
"step": 165
|
| 216 |
},
|
| 217 |
{
|
| 218 |
"epoch": 1.49,
|
| 219 |
+
"learning_rate": 1.4912280701754386e-05,
|
| 220 |
+
"loss": 0.7899,
|
| 221 |
"step": 170
|
| 222 |
},
|
| 223 |
{
|
| 224 |
"epoch": 1.53,
|
| 225 |
+
"learning_rate": 1.5350877192982457e-05,
|
| 226 |
+
"loss": 0.7818,
|
| 227 |
"step": 175
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 1.58,
|
| 231 |
+
"learning_rate": 1.5789473684210526e-05,
|
| 232 |
+
"loss": 0.7952,
|
| 233 |
"step": 180
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 1.62,
|
| 237 |
+
"learning_rate": 1.62280701754386e-05,
|
| 238 |
+
"loss": 0.7489,
|
| 239 |
"step": 185
|
| 240 |
},
|
| 241 |
{
|
| 242 |
"epoch": 1.66,
|
| 243 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 244 |
+
"loss": 0.7705,
|
| 245 |
"step": 190
|
| 246 |
},
|
| 247 |
{
|
| 248 |
"epoch": 1.71,
|
| 249 |
+
"learning_rate": 1.7105263157894737e-05,
|
| 250 |
+
"loss": 0.917,
|
| 251 |
"step": 195
|
| 252 |
},
|
| 253 |
{
|
| 254 |
"epoch": 1.75,
|
| 255 |
+
"learning_rate": 1.7543859649122806e-05,
|
| 256 |
+
"loss": 0.8671,
|
| 257 |
"step": 200
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 1.79,
|
| 261 |
+
"learning_rate": 1.7982456140350878e-05,
|
| 262 |
+
"loss": 0.8763,
|
| 263 |
"step": 205
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 1.84,
|
| 267 |
+
"learning_rate": 1.8421052631578947e-05,
|
| 268 |
+
"loss": 0.6994,
|
| 269 |
"step": 210
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 1.88,
|
| 273 |
+
"learning_rate": 1.885964912280702e-05,
|
| 274 |
+
"loss": 0.6047,
|
| 275 |
"step": 215
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 1.93,
|
| 279 |
+
"learning_rate": 1.929824561403509e-05,
|
| 280 |
+
"loss": 0.6536,
|
| 281 |
"step": 220
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 1.97,
|
| 285 |
+
"learning_rate": 1.9736842105263158e-05,
|
| 286 |
+
"loss": 0.7272,
|
| 287 |
"step": 225
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 2.0,
|
| 291 |
+
"eval_accuracy": 0.8308702791461412,
|
| 292 |
+
"eval_loss": 0.4748782813549042,
|
| 293 |
+
"eval_runtime": 51.2882,
|
| 294 |
+
"eval_samples_per_second": 23.748,
|
| 295 |
+
"eval_steps_per_second": 2.983,
|
| 296 |
"step": 228
|
| 297 |
},
|
| 298 |
{
|
| 299 |
"epoch": 2.01,
|
| 300 |
+
"learning_rate": 2.0175438596491227e-05,
|
| 301 |
+
"loss": 0.6975,
|
| 302 |
"step": 230
|
| 303 |
},
|
| 304 |
{
|
| 305 |
"epoch": 2.06,
|
| 306 |
+
"learning_rate": 2.06140350877193e-05,
|
| 307 |
+
"loss": 0.5554,
|
| 308 |
"step": 235
|
| 309 |
},
|
| 310 |
{
|
| 311 |
"epoch": 2.1,
|
| 312 |
+
"learning_rate": 2.105263157894737e-05,
|
| 313 |
+
"loss": 0.5656,
|
| 314 |
"step": 240
|
| 315 |
},
|
| 316 |
{
|
| 317 |
"epoch": 2.14,
|
| 318 |
+
"learning_rate": 2.149122807017544e-05,
|
| 319 |
+
"loss": 0.6029,
|
| 320 |
"step": 245
|
| 321 |
},
|
| 322 |
{
|
| 323 |
"epoch": 2.19,
|
| 324 |
+
"learning_rate": 2.1929824561403507e-05,
|
| 325 |
+
"loss": 0.5438,
|
| 326 |
"step": 250
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 2.23,
|
| 330 |
+
"learning_rate": 2.236842105263158e-05,
|
| 331 |
+
"loss": 0.6093,
|
| 332 |
"step": 255
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 2.28,
|
| 336 |
+
"learning_rate": 2.280701754385965e-05,
|
| 337 |
+
"loss": 0.6688,
|
| 338 |
"step": 260
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 2.32,
|
| 342 |
+
"learning_rate": 2.324561403508772e-05,
|
| 343 |
+
"loss": 0.6388,
|
| 344 |
"step": 265
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 2.36,
|
| 348 |
+
"learning_rate": 2.368421052631579e-05,
|
| 349 |
+
"loss": 0.5758,
|
| 350 |
"step": 270
|
| 351 |
},
|
| 352 |
{
|
| 353 |
"epoch": 2.41,
|
| 354 |
+
"learning_rate": 2.412280701754386e-05,
|
| 355 |
+
"loss": 0.6343,
|
| 356 |
"step": 275
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 2.45,
|
| 360 |
+
"learning_rate": 2.456140350877193e-05,
|
| 361 |
+
"loss": 0.7233,
|
| 362 |
"step": 280
|
| 363 |
},
|
| 364 |
{
|
| 365 |
"epoch": 2.49,
|
| 366 |
+
"learning_rate": 2.5e-05,
|
| 367 |
+
"loss": 0.4706,
|
| 368 |
"step": 285
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 2.54,
|
| 372 |
+
"learning_rate": 2.5438596491228074e-05,
|
| 373 |
+
"loss": 0.5836,
|
| 374 |
"step": 290
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 2.58,
|
| 378 |
+
"learning_rate": 2.5877192982456143e-05,
|
| 379 |
+
"loss": 0.5001,
|
| 380 |
"step": 295
|
| 381 |
},
|
| 382 |
{
|
| 383 |
"epoch": 2.63,
|
| 384 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 385 |
+
"loss": 0.6911,
|
| 386 |
"step": 300
|
| 387 |
},
|
| 388 |
{
|
| 389 |
"epoch": 2.67,
|
| 390 |
+
"learning_rate": 2.675438596491228e-05,
|
| 391 |
+
"loss": 0.7413,
|
| 392 |
"step": 305
|
| 393 |
},
|
| 394 |
{
|
| 395 |
"epoch": 2.71,
|
| 396 |
+
"learning_rate": 2.7192982456140354e-05,
|
| 397 |
+
"loss": 0.5479,
|
| 398 |
"step": 310
|
| 399 |
},
|
| 400 |
{
|
| 401 |
"epoch": 2.76,
|
| 402 |
+
"learning_rate": 2.7631578947368426e-05,
|
| 403 |
+
"loss": 0.4444,
|
| 404 |
"step": 315
|
| 405 |
},
|
| 406 |
{
|
| 407 |
"epoch": 2.8,
|
| 408 |
+
"learning_rate": 2.8070175438596492e-05,
|
| 409 |
+
"loss": 0.5584,
|
| 410 |
"step": 320
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 2.84,
|
| 414 |
+
"learning_rate": 2.850877192982456e-05,
|
| 415 |
+
"loss": 0.3468,
|
| 416 |
"step": 325
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 2.89,
|
| 420 |
+
"learning_rate": 2.8947368421052634e-05,
|
| 421 |
+
"loss": 0.6219,
|
| 422 |
"step": 330
|
| 423 |
},
|
| 424 |
{
|
| 425 |
"epoch": 2.93,
|
| 426 |
+
"learning_rate": 2.9385964912280706e-05,
|
| 427 |
+
"loss": 0.5142,
|
| 428 |
"step": 335
|
| 429 |
},
|
| 430 |
{
|
| 431 |
"epoch": 2.98,
|
| 432 |
+
"learning_rate": 2.9824561403508772e-05,
|
| 433 |
+
"loss": 0.4335,
|
| 434 |
"step": 340
|
| 435 |
},
|
| 436 |
{
|
| 437 |
"epoch": 2.99,
|
| 438 |
+
"eval_accuracy": 0.9195402298850575,
|
| 439 |
+
"eval_loss": 0.24883201718330383,
|
| 440 |
+
"eval_runtime": 51.2735,
|
| 441 |
+
"eval_samples_per_second": 23.755,
|
| 442 |
+
"eval_steps_per_second": 2.984,
|
| 443 |
"step": 342
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"epoch": 3.02,
|
| 447 |
+
"learning_rate": 3.0263157894736844e-05,
|
| 448 |
+
"loss": 0.4105,
|
| 449 |
"step": 345
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 3.06,
|
| 453 |
+
"learning_rate": 3.0701754385964913e-05,
|
| 454 |
+
"loss": 0.4176,
|
| 455 |
"step": 350
|
| 456 |
},
|
| 457 |
{
|
| 458 |
"epoch": 3.11,
|
| 459 |
+
"learning_rate": 3.1140350877192986e-05,
|
| 460 |
+
"loss": 0.5252,
|
| 461 |
"step": 355
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 3.15,
|
| 465 |
+
"learning_rate": 3.157894736842105e-05,
|
| 466 |
+
"loss": 0.4416,
|
| 467 |
"step": 360
|
| 468 |
},
|
| 469 |
{
|
| 470 |
"epoch": 3.19,
|
| 471 |
+
"learning_rate": 3.2017543859649124e-05,
|
| 472 |
+
"loss": 0.4243,
|
| 473 |
"step": 365
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 3.24,
|
| 477 |
+
"learning_rate": 3.24561403508772e-05,
|
| 478 |
+
"loss": 0.3625,
|
| 479 |
"step": 370
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 3.28,
|
| 483 |
+
"learning_rate": 3.289473684210527e-05,
|
| 484 |
+
"loss": 0.3327,
|
| 485 |
"step": 375
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"epoch": 3.33,
|
| 489 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 490 |
+
"loss": 0.3239,
|
| 491 |
"step": 380
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 3.37,
|
| 495 |
+
"learning_rate": 3.377192982456141e-05,
|
| 496 |
+
"loss": 0.4551,
|
| 497 |
"step": 385
|
| 498 |
},
|
| 499 |
{
|
| 500 |
"epoch": 3.41,
|
| 501 |
+
"learning_rate": 3.421052631578947e-05,
|
| 502 |
+
"loss": 0.4704,
|
| 503 |
"step": 390
|
| 504 |
},
|
| 505 |
{
|
| 506 |
"epoch": 3.46,
|
| 507 |
+
"learning_rate": 3.4649122807017546e-05,
|
| 508 |
+
"loss": 0.442,
|
| 509 |
"step": 395
|
| 510 |
},
|
| 511 |
{
|
| 512 |
"epoch": 3.5,
|
| 513 |
+
"learning_rate": 3.508771929824561e-05,
|
| 514 |
+
"loss": 0.3715,
|
| 515 |
"step": 400
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 3.54,
|
| 519 |
+
"learning_rate": 3.5526315789473684e-05,
|
| 520 |
+
"loss": 0.3742,
|
| 521 |
"step": 405
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 3.59,
|
| 525 |
+
"learning_rate": 3.5964912280701756e-05,
|
| 526 |
+
"loss": 0.3209,
|
| 527 |
"step": 410
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"epoch": 3.63,
|
| 531 |
+
"learning_rate": 3.640350877192983e-05,
|
| 532 |
+
"loss": 0.3335,
|
| 533 |
"step": 415
|
| 534 |
},
|
| 535 |
{
|
| 536 |
"epoch": 3.68,
|
| 537 |
+
"learning_rate": 3.6842105263157895e-05,
|
| 538 |
+
"loss": 0.3519,
|
| 539 |
"step": 420
|
| 540 |
},
|
| 541 |
{
|
| 542 |
"epoch": 3.72,
|
| 543 |
+
"learning_rate": 3.728070175438597e-05,
|
| 544 |
+
"loss": 0.3011,
|
| 545 |
"step": 425
|
| 546 |
},
|
| 547 |
{
|
| 548 |
"epoch": 3.76,
|
| 549 |
+
"learning_rate": 3.771929824561404e-05,
|
| 550 |
+
"loss": 0.4543,
|
| 551 |
"step": 430
|
| 552 |
},
|
| 553 |
{
|
| 554 |
"epoch": 3.81,
|
| 555 |
+
"learning_rate": 3.815789473684211e-05,
|
| 556 |
+
"loss": 0.3693,
|
| 557 |
"step": 435
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 3.85,
|
| 561 |
+
"learning_rate": 3.859649122807018e-05,
|
| 562 |
+
"loss": 0.3256,
|
| 563 |
"step": 440
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"epoch": 3.89,
|
| 567 |
+
"learning_rate": 3.9035087719298244e-05,
|
| 568 |
+
"loss": 0.3463,
|
| 569 |
"step": 445
|
| 570 |
},
|
| 571 |
{
|
| 572 |
"epoch": 3.94,
|
| 573 |
+
"learning_rate": 3.9473684210526316e-05,
|
| 574 |
+
"loss": 0.3209,
|
| 575 |
"step": 450
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"epoch": 3.98,
|
| 579 |
+
"learning_rate": 3.991228070175439e-05,
|
| 580 |
+
"loss": 0.3298,
|
| 581 |
"step": 455
|
| 582 |
},
|
| 583 |
{
|
| 584 |
"epoch": 4.0,
|
| 585 |
+
"eval_accuracy": 0.9310344827586207,
|
| 586 |
+
"eval_loss": 0.17004548013210297,
|
| 587 |
+
"eval_runtime": 51.2681,
|
| 588 |
+
"eval_samples_per_second": 23.757,
|
| 589 |
+
"eval_steps_per_second": 2.984,
|
| 590 |
"step": 457
|
| 591 |
},
|
| 592 |
{
|
| 593 |
"epoch": 4.03,
|
| 594 |
+
"learning_rate": 4.0350877192982455e-05,
|
| 595 |
+
"loss": 0.3795,
|
| 596 |
"step": 460
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 4.07,
|
| 600 |
+
"learning_rate": 4.078947368421053e-05,
|
| 601 |
+
"loss": 0.2746,
|
| 602 |
"step": 465
|
| 603 |
},
|
| 604 |
{
|
| 605 |
"epoch": 4.11,
|
| 606 |
+
"learning_rate": 4.12280701754386e-05,
|
| 607 |
+
"loss": 0.2629,
|
| 608 |
"step": 470
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 4.16,
|
| 612 |
+
"learning_rate": 4.166666666666667e-05,
|
| 613 |
+
"loss": 0.3851,
|
| 614 |
"step": 475
|
| 615 |
},
|
| 616 |
{
|
| 617 |
"epoch": 4.2,
|
| 618 |
+
"learning_rate": 4.210526315789474e-05,
|
| 619 |
"loss": 0.3722,
|
| 620 |
"step": 480
|
| 621 |
},
|
| 622 |
{
|
| 623 |
"epoch": 4.25,
|
| 624 |
+
"learning_rate": 4.254385964912281e-05,
|
| 625 |
+
"loss": 0.2845,
|
| 626 |
"step": 485
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 4.29,
|
| 630 |
+
"learning_rate": 4.298245614035088e-05,
|
| 631 |
+
"loss": 0.313,
|
| 632 |
"step": 490
|
| 633 |
},
|
| 634 |
{
|
| 635 |
"epoch": 4.33,
|
| 636 |
+
"learning_rate": 4.342105263157895e-05,
|
| 637 |
+
"loss": 0.2468,
|
| 638 |
"step": 495
|
| 639 |
},
|
| 640 |
{
|
| 641 |
"epoch": 4.38,
|
| 642 |
+
"learning_rate": 4.3859649122807014e-05,
|
| 643 |
+
"loss": 0.3358,
|
| 644 |
"step": 500
|
| 645 |
},
|
| 646 |
{
|
| 647 |
"epoch": 4.42,
|
| 648 |
+
"learning_rate": 4.429824561403509e-05,
|
| 649 |
+
"loss": 0.4219,
|
| 650 |
"step": 505
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 4.46,
|
| 654 |
+
"learning_rate": 4.473684210526316e-05,
|
| 655 |
+
"loss": 0.3461,
|
| 656 |
"step": 510
|
| 657 |
},
|
| 658 |
{
|
| 659 |
"epoch": 4.51,
|
| 660 |
+
"learning_rate": 4.517543859649123e-05,
|
| 661 |
+
"loss": 0.3383,
|
| 662 |
"step": 515
|
| 663 |
},
|
| 664 |
{
|
| 665 |
"epoch": 4.55,
|
| 666 |
+
"learning_rate": 4.56140350877193e-05,
|
| 667 |
+
"loss": 0.3851,
|
| 668 |
"step": 520
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 4.6,
|
| 672 |
+
"learning_rate": 4.605263157894737e-05,
|
| 673 |
+
"loss": 0.3833,
|
| 674 |
"step": 525
|
| 675 |
},
|
| 676 |
{
|
| 677 |
"epoch": 4.64,
|
| 678 |
+
"learning_rate": 4.649122807017544e-05,
|
| 679 |
+
"loss": 0.4285,
|
| 680 |
"step": 530
|
| 681 |
},
|
| 682 |
{
|
| 683 |
"epoch": 4.68,
|
| 684 |
+
"learning_rate": 4.6929824561403515e-05,
|
| 685 |
+
"loss": 0.5201,
|
| 686 |
"step": 535
|
| 687 |
},
|
| 688 |
{
|
| 689 |
"epoch": 4.73,
|
| 690 |
+
"learning_rate": 4.736842105263158e-05,
|
| 691 |
+
"loss": 0.3688,
|
| 692 |
"step": 540
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 4.77,
|
| 696 |
+
"learning_rate": 4.780701754385965e-05,
|
| 697 |
+
"loss": 0.2883,
|
| 698 |
"step": 545
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 4.81,
|
| 702 |
+
"learning_rate": 4.824561403508772e-05,
|
| 703 |
+
"loss": 0.2787,
|
| 704 |
"step": 550
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 4.86,
|
| 708 |
+
"learning_rate": 4.868421052631579e-05,
|
| 709 |
+
"loss": 0.1929,
|
| 710 |
"step": 555
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 4.9,
|
| 714 |
+
"learning_rate": 4.912280701754386e-05,
|
| 715 |
+
"loss": 0.2865,
|
| 716 |
"step": 560
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 4.95,
|
| 720 |
+
"learning_rate": 4.956140350877193e-05,
|
| 721 |
+
"loss": 0.2997,
|
| 722 |
"step": 565
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 4.99,
|
| 726 |
+
"learning_rate": 5e-05,
|
| 727 |
+
"loss": 0.177,
|
| 728 |
"step": 570
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 5.0,
|
| 732 |
+
"eval_accuracy": 0.9261083743842364,
|
| 733 |
+
"eval_loss": 0.21159406006336212,
|
| 734 |
+
"eval_runtime": 51.2395,
|
| 735 |
+
"eval_samples_per_second": 23.771,
|
| 736 |
+
"eval_steps_per_second": 2.986,
|
| 737 |
"step": 571
|
| 738 |
},
|
| 739 |
{
|
| 740 |
"epoch": 5.03,
|
| 741 |
+
"learning_rate": 4.956140350877193e-05,
|
| 742 |
+
"loss": 0.3122,
|
| 743 |
"step": 575
|
| 744 |
},
|
| 745 |
{
|
| 746 |
"epoch": 5.08,
|
| 747 |
+
"learning_rate": 4.912280701754386e-05,
|
| 748 |
+
"loss": 0.3343,
|
| 749 |
"step": 580
|
| 750 |
},
|
| 751 |
{
|
| 752 |
"epoch": 5.12,
|
| 753 |
+
"learning_rate": 4.868421052631579e-05,
|
| 754 |
+
"loss": 0.5452,
|
| 755 |
"step": 585
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 5.16,
|
| 759 |
+
"learning_rate": 4.824561403508772e-05,
|
| 760 |
+
"loss": 0.2825,
|
| 761 |
"step": 590
|
| 762 |
},
|
| 763 |
{
|
| 764 |
"epoch": 5.21,
|
| 765 |
+
"learning_rate": 4.780701754385965e-05,
|
| 766 |
+
"loss": 0.345,
|
| 767 |
"step": 595
|
| 768 |
},
|
| 769 |
{
|
| 770 |
"epoch": 5.25,
|
| 771 |
+
"learning_rate": 4.736842105263158e-05,
|
| 772 |
+
"loss": 0.3114,
|
| 773 |
"step": 600
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 5.3,
|
| 777 |
+
"learning_rate": 4.6929824561403515e-05,
|
| 778 |
+
"loss": 0.236,
|
| 779 |
"step": 605
|
| 780 |
},
|
| 781 |
{
|
| 782 |
"epoch": 5.34,
|
| 783 |
+
"learning_rate": 4.649122807017544e-05,
|
| 784 |
+
"loss": 0.1891,
|
| 785 |
"step": 610
|
| 786 |
},
|
| 787 |
{
|
| 788 |
"epoch": 5.38,
|
| 789 |
+
"learning_rate": 4.605263157894737e-05,
|
| 790 |
+
"loss": 0.2101,
|
| 791 |
"step": 615
|
| 792 |
},
|
| 793 |
{
|
| 794 |
"epoch": 5.43,
|
| 795 |
+
"learning_rate": 4.56140350877193e-05,
|
| 796 |
+
"loss": 0.3216,
|
| 797 |
"step": 620
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 5.47,
|
| 801 |
+
"learning_rate": 4.517543859649123e-05,
|
| 802 |
+
"loss": 0.2823,
|
| 803 |
"step": 625
|
| 804 |
},
|
| 805 |
{
|
| 806 |
"epoch": 5.51,
|
| 807 |
+
"learning_rate": 4.473684210526316e-05,
|
| 808 |
+
"loss": 0.3153,
|
| 809 |
"step": 630
|
| 810 |
},
|
| 811 |
{
|
| 812 |
"epoch": 5.56,
|
| 813 |
+
"learning_rate": 4.429824561403509e-05,
|
| 814 |
+
"loss": 0.3139,
|
| 815 |
"step": 635
|
| 816 |
},
|
| 817 |
{
|
| 818 |
"epoch": 5.6,
|
| 819 |
+
"learning_rate": 4.3859649122807014e-05,
|
| 820 |
+
"loss": 0.3497,
|
| 821 |
"step": 640
|
| 822 |
},
|
| 823 |
{
|
| 824 |
"epoch": 5.65,
|
| 825 |
+
"learning_rate": 4.342105263157895e-05,
|
| 826 |
+
"loss": 0.2596,
|
| 827 |
"step": 645
|
| 828 |
},
|
| 829 |
{
|
| 830 |
"epoch": 5.69,
|
| 831 |
+
"learning_rate": 4.298245614035088e-05,
|
| 832 |
+
"loss": 0.3244,
|
| 833 |
"step": 650
|
| 834 |
},
|
| 835 |
{
|
| 836 |
"epoch": 5.73,
|
| 837 |
+
"learning_rate": 4.254385964912281e-05,
|
| 838 |
+
"loss": 0.2692,
|
| 839 |
"step": 655
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 5.78,
|
| 843 |
+
"learning_rate": 4.210526315789474e-05,
|
| 844 |
+
"loss": 0.2558,
|
| 845 |
"step": 660
|
| 846 |
},
|
| 847 |
{
|
| 848 |
"epoch": 5.82,
|
| 849 |
+
"learning_rate": 4.166666666666667e-05,
|
| 850 |
+
"loss": 0.234,
|
| 851 |
"step": 665
|
| 852 |
},
|
| 853 |
{
|
| 854 |
"epoch": 5.86,
|
| 855 |
+
"learning_rate": 4.12280701754386e-05,
|
| 856 |
+
"loss": 0.3071,
|
| 857 |
"step": 670
|
| 858 |
},
|
| 859 |
{
|
| 860 |
"epoch": 5.91,
|
| 861 |
+
"learning_rate": 4.078947368421053e-05,
|
| 862 |
+
"loss": 0.1587,
|
| 863 |
"step": 675
|
| 864 |
},
|
| 865 |
{
|
| 866 |
"epoch": 5.95,
|
| 867 |
+
"learning_rate": 4.0350877192982455e-05,
|
| 868 |
+
"loss": 0.2624,
|
| 869 |
"step": 680
|
| 870 |
},
|
| 871 |
{
|
| 872 |
"epoch": 6.0,
|
| 873 |
+
"learning_rate": 3.991228070175439e-05,
|
| 874 |
+
"loss": 0.2299,
|
| 875 |
"step": 685
|
| 876 |
},
|
| 877 |
{
|
| 878 |
"epoch": 6.0,
|
| 879 |
+
"eval_accuracy": 0.9753694581280788,
|
| 880 |
+
"eval_loss": 0.09334532916545868,
|
| 881 |
+
"eval_runtime": 51.1765,
|
| 882 |
+
"eval_samples_per_second": 23.8,
|
| 883 |
+
"eval_steps_per_second": 2.99,
|
| 884 |
"step": 685
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 6.04,
|
| 888 |
+
"learning_rate": 3.9473684210526316e-05,
|
| 889 |
+
"loss": 0.1728,
|
| 890 |
"step": 690
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 6.08,
|
| 894 |
+
"learning_rate": 3.9035087719298244e-05,
|
| 895 |
+
"loss": 0.1751,
|
| 896 |
"step": 695
|
| 897 |
},
|
| 898 |
{
|
| 899 |
"epoch": 6.13,
|
| 900 |
+
"learning_rate": 3.859649122807018e-05,
|
| 901 |
+
"loss": 0.2653,
|
| 902 |
"step": 700
|
| 903 |
},
|
| 904 |
{
|
| 905 |
"epoch": 6.17,
|
| 906 |
+
"learning_rate": 3.815789473684211e-05,
|
| 907 |
+
"loss": 0.2508,
|
| 908 |
"step": 705
|
| 909 |
},
|
| 910 |
{
|
| 911 |
"epoch": 6.21,
|
| 912 |
+
"learning_rate": 3.771929824561404e-05,
|
| 913 |
+
"loss": 0.1909,
|
| 914 |
"step": 710
|
| 915 |
},
|
| 916 |
{
|
| 917 |
"epoch": 6.26,
|
| 918 |
+
"learning_rate": 3.728070175438597e-05,
|
| 919 |
+
"loss": 0.2278,
|
| 920 |
"step": 715
|
| 921 |
},
|
| 922 |
{
|
| 923 |
"epoch": 6.3,
|
| 924 |
+
"learning_rate": 3.6842105263157895e-05,
|
| 925 |
+
"loss": 0.2178,
|
| 926 |
"step": 720
|
| 927 |
},
|
| 928 |
{
|
| 929 |
"epoch": 6.35,
|
| 930 |
+
"learning_rate": 3.640350877192983e-05,
|
| 931 |
+
"loss": 0.2482,
|
| 932 |
"step": 725
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 6.39,
|
| 936 |
+
"learning_rate": 3.5964912280701756e-05,
|
| 937 |
+
"loss": 0.2605,
|
| 938 |
"step": 730
|
| 939 |
},
|
| 940 |
{
|
| 941 |
"epoch": 6.43,
|
| 942 |
+
"learning_rate": 3.5526315789473684e-05,
|
| 943 |
+
"loss": 0.3173,
|
| 944 |
"step": 735
|
| 945 |
},
|
| 946 |
{
|
| 947 |
"epoch": 6.48,
|
| 948 |
+
"learning_rate": 3.508771929824561e-05,
|
| 949 |
+
"loss": 0.1419,
|
| 950 |
"step": 740
|
| 951 |
},
|
| 952 |
{
|
| 953 |
"epoch": 6.52,
|
| 954 |
+
"learning_rate": 3.4649122807017546e-05,
|
| 955 |
+
"loss": 0.2116,
|
| 956 |
"step": 745
|
| 957 |
},
|
| 958 |
{
|
| 959 |
"epoch": 6.56,
|
| 960 |
+
"learning_rate": 3.421052631578947e-05,
|
| 961 |
+
"loss": 0.1438,
|
| 962 |
"step": 750
|
| 963 |
},
|
| 964 |
{
|
| 965 |
"epoch": 6.61,
|
| 966 |
+
"learning_rate": 3.377192982456141e-05,
|
| 967 |
+
"loss": 0.1522,
|
| 968 |
"step": 755
|
| 969 |
},
|
| 970 |
{
|
| 971 |
"epoch": 6.65,
|
| 972 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 973 |
+
"loss": 0.1583,
|
| 974 |
"step": 760
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 6.7,
|
| 978 |
+
"learning_rate": 3.289473684210527e-05,
|
| 979 |
+
"loss": 0.1541,
|
| 980 |
"step": 765
|
| 981 |
},
|
| 982 |
{
|
| 983 |
"epoch": 6.74,
|
| 984 |
+
"learning_rate": 3.24561403508772e-05,
|
| 985 |
+
"loss": 0.2615,
|
| 986 |
"step": 770
|
| 987 |
},
|
| 988 |
{
|
| 989 |
"epoch": 6.78,
|
| 990 |
+
"learning_rate": 3.2017543859649124e-05,
|
| 991 |
+
"loss": 0.1603,
|
| 992 |
"step": 775
|
| 993 |
},
|
| 994 |
{
|
| 995 |
"epoch": 6.83,
|
| 996 |
+
"learning_rate": 3.157894736842105e-05,
|
| 997 |
+
"loss": 0.1657,
|
| 998 |
"step": 780
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
"epoch": 6.87,
|
| 1002 |
+
"learning_rate": 3.1140350877192986e-05,
|
| 1003 |
+
"loss": 0.1074,
|
| 1004 |
"step": 785
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 6.91,
|
| 1008 |
+
"learning_rate": 3.0701754385964913e-05,
|
| 1009 |
+
"loss": 0.1414,
|
| 1010 |
"step": 790
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
"epoch": 6.96,
|
| 1014 |
+
"learning_rate": 3.0263157894736844e-05,
|
| 1015 |
+
"loss": 0.2586,
|
| 1016 |
"step": 795
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 6.99,
|
| 1020 |
+
"eval_accuracy": 0.986863711001642,
|
| 1021 |
+
"eval_loss": 0.031560178846120834,
|
| 1022 |
+
"eval_runtime": 51.3582,
|
| 1023 |
+
"eval_samples_per_second": 23.716,
|
| 1024 |
+
"eval_steps_per_second": 2.979,
|
| 1025 |
"step": 799
|
| 1026 |
},
|
| 1027 |
{
|
| 1028 |
"epoch": 7.0,
|
| 1029 |
+
"learning_rate": 2.9824561403508772e-05,
|
| 1030 |
+
"loss": 0.1753,
|
| 1031 |
"step": 800
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
"epoch": 7.05,
|
| 1035 |
+
"learning_rate": 2.9385964912280706e-05,
|
| 1036 |
+
"loss": 0.1409,
|
| 1037 |
"step": 805
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 7.09,
|
| 1041 |
+
"learning_rate": 2.8947368421052634e-05,
|
| 1042 |
+
"loss": 0.0889,
|
| 1043 |
"step": 810
|
| 1044 |
},
|
| 1045 |
{
|
| 1046 |
"epoch": 7.13,
|
| 1047 |
+
"learning_rate": 2.850877192982456e-05,
|
| 1048 |
+
"loss": 0.1245,
|
| 1049 |
"step": 815
|
| 1050 |
},
|
| 1051 |
{
|
| 1052 |
"epoch": 7.18,
|
| 1053 |
+
"learning_rate": 2.8070175438596492e-05,
|
| 1054 |
+
"loss": 0.116,
|
| 1055 |
"step": 820
|
| 1056 |
},
|
| 1057 |
{
|
| 1058 |
"epoch": 7.22,
|
| 1059 |
+
"learning_rate": 2.7631578947368426e-05,
|
| 1060 |
+
"loss": 0.196,
|
| 1061 |
"step": 825
|
| 1062 |
},
|
| 1063 |
{
|
| 1064 |
"epoch": 7.26,
|
| 1065 |
+
"learning_rate": 2.7192982456140354e-05,
|
| 1066 |
+
"loss": 0.0874,
|
| 1067 |
"step": 830
|
| 1068 |
},
|
| 1069 |
{
|
| 1070 |
"epoch": 7.31,
|
| 1071 |
+
"learning_rate": 2.675438596491228e-05,
|
| 1072 |
+
"loss": 0.2338,
|
| 1073 |
"step": 835
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
"epoch": 7.35,
|
| 1077 |
+
"learning_rate": 2.6315789473684212e-05,
|
| 1078 |
+
"loss": 0.247,
|
| 1079 |
"step": 840
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 7.4,
|
| 1083 |
+
"learning_rate": 2.5877192982456143e-05,
|
| 1084 |
+
"loss": 0.2131,
|
| 1085 |
"step": 845
|
| 1086 |
},
|
| 1087 |
{
|
| 1088 |
"epoch": 7.44,
|
| 1089 |
+
"learning_rate": 2.5438596491228074e-05,
|
| 1090 |
+
"loss": 0.2523,
|
| 1091 |
"step": 850
|
| 1092 |
},
|
| 1093 |
{
|
| 1094 |
"epoch": 7.48,
|
| 1095 |
+
"learning_rate": 2.5e-05,
|
| 1096 |
+
"loss": 0.1632,
|
| 1097 |
"step": 855
|
| 1098 |
},
|
| 1099 |
{
|
| 1100 |
"epoch": 7.53,
|
| 1101 |
+
"learning_rate": 2.456140350877193e-05,
|
| 1102 |
+
"loss": 0.1257,
|
| 1103 |
"step": 860
|
| 1104 |
},
|
| 1105 |
{
|
| 1106 |
"epoch": 7.57,
|
| 1107 |
+
"learning_rate": 2.412280701754386e-05,
|
| 1108 |
+
"loss": 0.1589,
|
| 1109 |
"step": 865
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
"epoch": 7.61,
|
| 1113 |
+
"learning_rate": 2.368421052631579e-05,
|
| 1114 |
+
"loss": 0.1895,
|
| 1115 |
"step": 870
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
"epoch": 7.66,
|
| 1119 |
+
"learning_rate": 2.324561403508772e-05,
|
| 1120 |
+
"loss": 0.1422,
|
| 1121 |
"step": 875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 7.7,
|
| 1125 |
+
"learning_rate": 2.280701754385965e-05,
|
| 1126 |
+
"loss": 0.1701,
|
| 1127 |
"step": 880
|
| 1128 |
},
|
| 1129 |
{
|
| 1130 |
"epoch": 7.75,
|
| 1131 |
+
"learning_rate": 2.236842105263158e-05,
|
| 1132 |
+
"loss": 0.0639,
|
| 1133 |
"step": 885
|
| 1134 |
},
|
| 1135 |
{
|
| 1136 |
"epoch": 7.79,
|
| 1137 |
+
"learning_rate": 2.1929824561403507e-05,
|
| 1138 |
+
"loss": 0.2418,
|
| 1139 |
"step": 890
|
| 1140 |
},
|
| 1141 |
{
|
| 1142 |
"epoch": 7.83,
|
| 1143 |
+
"learning_rate": 2.149122807017544e-05,
|
| 1144 |
+
"loss": 0.0928,
|
| 1145 |
"step": 895
|
| 1146 |
},
|
| 1147 |
{
|
| 1148 |
"epoch": 7.88,
|
| 1149 |
+
"learning_rate": 2.105263157894737e-05,
|
| 1150 |
+
"loss": 0.1306,
|
| 1151 |
"step": 900
|
| 1152 |
},
|
| 1153 |
{
|
| 1154 |
"epoch": 7.92,
|
| 1155 |
+
"learning_rate": 2.06140350877193e-05,
|
| 1156 |
+
"loss": 0.1759,
|
| 1157 |
"step": 905
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
"epoch": 7.96,
|
| 1161 |
+
"learning_rate": 2.0175438596491227e-05,
|
| 1162 |
+
"loss": 0.1053,
|
| 1163 |
"step": 910
|
| 1164 |
},
|
| 1165 |
{
|
| 1166 |
"epoch": 8.0,
|
| 1167 |
+
"eval_accuracy": 0.9909688013136289,
|
| 1168 |
+
"eval_loss": 0.025560511276125908,
|
| 1169 |
+
"eval_runtime": 51.3274,
|
| 1170 |
+
"eval_samples_per_second": 23.73,
|
| 1171 |
+
"eval_steps_per_second": 2.981,
|
| 1172 |
"step": 914
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 8.01,
|
| 1176 |
+
"learning_rate": 1.9736842105263158e-05,
|
| 1177 |
+
"loss": 0.1086,
|
| 1178 |
"step": 915
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
"epoch": 8.05,
|
| 1182 |
+
"learning_rate": 1.929824561403509e-05,
|
| 1183 |
+
"loss": 0.1473,
|
| 1184 |
"step": 920
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
"epoch": 8.1,
|
| 1188 |
+
"learning_rate": 1.885964912280702e-05,
|
| 1189 |
+
"loss": 0.047,
|
| 1190 |
"step": 925
|
| 1191 |
},
|
| 1192 |
{
|
| 1193 |
"epoch": 8.14,
|
| 1194 |
+
"learning_rate": 1.8421052631578947e-05,
|
| 1195 |
+
"loss": 0.1855,
|
| 1196 |
"step": 930
|
| 1197 |
},
|
| 1198 |
{
|
| 1199 |
"epoch": 8.18,
|
| 1200 |
+
"learning_rate": 1.7982456140350878e-05,
|
| 1201 |
+
"loss": 0.1216,
|
| 1202 |
"step": 935
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
"epoch": 8.23,
|
| 1206 |
+
"learning_rate": 1.7543859649122806e-05,
|
| 1207 |
+
"loss": 0.1677,
|
| 1208 |
"step": 940
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
"epoch": 8.27,
|
| 1212 |
+
"learning_rate": 1.7105263157894737e-05,
|
| 1213 |
+
"loss": 0.1112,
|
| 1214 |
"step": 945
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 8.32,
|
| 1218 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 1219 |
+
"loss": 0.0748,
|
| 1220 |
"step": 950
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
"epoch": 8.36,
|
| 1224 |
+
"learning_rate": 1.62280701754386e-05,
|
| 1225 |
+
"loss": 0.124,
|
| 1226 |
"step": 955
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
"epoch": 8.4,
|
| 1230 |
+
"learning_rate": 1.5789473684210526e-05,
|
| 1231 |
+
"loss": 0.1319,
|
| 1232 |
"step": 960
|
| 1233 |
},
|
| 1234 |
{
|
| 1235 |
"epoch": 8.45,
|
| 1236 |
+
"learning_rate": 1.5350877192982457e-05,
|
| 1237 |
+
"loss": 0.1867,
|
| 1238 |
"step": 965
|
| 1239 |
},
|
| 1240 |
{
|
| 1241 |
"epoch": 8.49,
|
| 1242 |
+
"learning_rate": 1.4912280701754386e-05,
|
| 1243 |
+
"loss": 0.1458,
|
| 1244 |
"step": 970
|
| 1245 |
},
|
| 1246 |
{
|
| 1247 |
"epoch": 8.53,
|
| 1248 |
+
"learning_rate": 1.4473684210526317e-05,
|
| 1249 |
+
"loss": 0.1482,
|
| 1250 |
"step": 975
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
"epoch": 8.58,
|
| 1254 |
+
"learning_rate": 1.4035087719298246e-05,
|
| 1255 |
+
"loss": 0.122,
|
| 1256 |
"step": 980
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 8.62,
|
| 1260 |
+
"learning_rate": 1.3596491228070177e-05,
|
| 1261 |
+
"loss": 0.0777,
|
| 1262 |
"step": 985
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
"epoch": 8.67,
|
| 1266 |
+
"learning_rate": 1.3157894736842106e-05,
|
| 1267 |
+
"loss": 0.1007,
|
| 1268 |
"step": 990
|
| 1269 |
},
|
| 1270 |
{
|
| 1271 |
"epoch": 8.71,
|
| 1272 |
+
"learning_rate": 1.2719298245614037e-05,
|
| 1273 |
+
"loss": 0.1103,
|
| 1274 |
"step": 995
|
| 1275 |
},
|
| 1276 |
{
|
| 1277 |
"epoch": 8.75,
|
| 1278 |
+
"learning_rate": 1.2280701754385964e-05,
|
| 1279 |
+
"loss": 0.0859,
|
| 1280 |
"step": 1000
|
| 1281 |
},
|
| 1282 |
{
|
| 1283 |
"epoch": 8.8,
|
| 1284 |
+
"learning_rate": 1.1842105263157895e-05,
|
| 1285 |
+
"loss": 0.0928,
|
| 1286 |
"step": 1005
|
| 1287 |
},
|
| 1288 |
{
|
| 1289 |
"epoch": 8.84,
|
| 1290 |
+
"learning_rate": 1.1403508771929824e-05,
|
| 1291 |
+
"loss": 0.0475,
|
| 1292 |
"step": 1010
|
| 1293 |
},
|
| 1294 |
{
|
| 1295 |
"epoch": 8.88,
|
| 1296 |
+
"learning_rate": 1.0964912280701754e-05,
|
| 1297 |
+
"loss": 0.0863,
|
| 1298 |
"step": 1015
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 8.93,
|
| 1302 |
+
"learning_rate": 1.0526315789473684e-05,
|
| 1303 |
+
"loss": 0.1315,
|
| 1304 |
"step": 1020
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
"epoch": 8.97,
|
| 1308 |
+
"learning_rate": 1.0087719298245614e-05,
|
| 1309 |
+
"loss": 0.2159,
|
| 1310 |
"step": 1025
|
| 1311 |
},
|
| 1312 |
{
|
| 1313 |
"epoch": 9.0,
|
| 1314 |
+
"eval_accuracy": 0.9958949096880131,
|
| 1315 |
+
"eval_loss": 0.014652877114713192,
|
| 1316 |
+
"eval_runtime": 51.3705,
|
| 1317 |
+
"eval_samples_per_second": 23.71,
|
| 1318 |
+
"eval_steps_per_second": 2.978,
|
| 1319 |
"step": 1028
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 9.02,
|
| 1323 |
+
"learning_rate": 9.649122807017545e-06,
|
| 1324 |
+
"loss": 0.1224,
|
| 1325 |
"step": 1030
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
"epoch": 9.06,
|
| 1329 |
+
"learning_rate": 9.210526315789474e-06,
|
| 1330 |
+
"loss": 0.0728,
|
| 1331 |
"step": 1035
|
| 1332 |
},
|
| 1333 |
{
|
| 1334 |
"epoch": 9.1,
|
| 1335 |
+
"learning_rate": 8.771929824561403e-06,
|
| 1336 |
+
"loss": 0.0981,
|
| 1337 |
"step": 1040
|
| 1338 |
},
|
| 1339 |
{
|
| 1340 |
"epoch": 9.15,
|
| 1341 |
+
"learning_rate": 8.333333333333334e-06,
|
| 1342 |
+
"loss": 0.1195,
|
| 1343 |
"step": 1045
|
| 1344 |
},
|
| 1345 |
{
|
| 1346 |
"epoch": 9.19,
|
| 1347 |
+
"learning_rate": 7.894736842105263e-06,
|
| 1348 |
+
"loss": 0.1001,
|
| 1349 |
"step": 1050
|
| 1350 |
},
|
| 1351 |
{
|
| 1352 |
"epoch": 9.23,
|
| 1353 |
+
"learning_rate": 7.456140350877193e-06,
|
| 1354 |
+
"loss": 0.0816,
|
| 1355 |
"step": 1055
|
| 1356 |
},
|
| 1357 |
{
|
| 1358 |
"epoch": 9.28,
|
| 1359 |
+
"learning_rate": 7.017543859649123e-06,
|
| 1360 |
+
"loss": 0.1013,
|
| 1361 |
"step": 1060
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 9.32,
|
| 1365 |
+
"learning_rate": 6.578947368421053e-06,
|
| 1366 |
+
"loss": 0.0797,
|
| 1367 |
"step": 1065
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
"epoch": 9.37,
|
| 1371 |
+
"learning_rate": 6.140350877192982e-06,
|
| 1372 |
+
"loss": 0.0894,
|
| 1373 |
"step": 1070
|
| 1374 |
},
|
| 1375 |
{
|
| 1376 |
"epoch": 9.41,
|
| 1377 |
+
"learning_rate": 5.701754385964912e-06,
|
| 1378 |
+
"loss": 0.0353,
|
| 1379 |
"step": 1075
|
| 1380 |
},
|
| 1381 |
{
|
| 1382 |
"epoch": 9.45,
|
| 1383 |
+
"learning_rate": 5.263157894736842e-06,
|
| 1384 |
+
"loss": 0.0845,
|
| 1385 |
"step": 1080
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
"epoch": 9.5,
|
| 1389 |
+
"learning_rate": 4.824561403508772e-06,
|
| 1390 |
+
"loss": 0.0855,
|
| 1391 |
"step": 1085
|
| 1392 |
},
|
| 1393 |
{
|
| 1394 |
"epoch": 9.54,
|
| 1395 |
+
"learning_rate": 4.3859649122807014e-06,
|
| 1396 |
+
"loss": 0.0886,
|
| 1397 |
"step": 1090
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
"epoch": 9.58,
|
| 1401 |
+
"learning_rate": 3.9473684210526315e-06,
|
| 1402 |
+
"loss": 0.0394,
|
| 1403 |
"step": 1095
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 9.63,
|
| 1407 |
+
"learning_rate": 3.5087719298245615e-06,
|
| 1408 |
+
"loss": 0.0745,
|
| 1409 |
"step": 1100
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
"epoch": 9.67,
|
| 1413 |
+
"learning_rate": 3.070175438596491e-06,
|
| 1414 |
+
"loss": 0.073,
|
| 1415 |
"step": 1105
|
| 1416 |
},
|
| 1417 |
{
|
| 1418 |
"epoch": 9.72,
|
| 1419 |
+
"learning_rate": 2.631578947368421e-06,
|
| 1420 |
+
"loss": 0.0753,
|
| 1421 |
"step": 1110
|
| 1422 |
},
|
| 1423 |
{
|
| 1424 |
"epoch": 9.76,
|
| 1425 |
+
"learning_rate": 2.1929824561403507e-06,
|
| 1426 |
+
"loss": 0.0744,
|
| 1427 |
"step": 1115
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 9.8,
|
| 1431 |
+
"learning_rate": 1.7543859649122807e-06,
|
| 1432 |
+
"loss": 0.1127,
|
| 1433 |
"step": 1120
|
| 1434 |
},
|
| 1435 |
{
|
| 1436 |
"epoch": 9.85,
|
| 1437 |
+
"learning_rate": 1.3157894736842106e-06,
|
| 1438 |
+
"loss": 0.123,
|
| 1439 |
"step": 1125
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
"epoch": 9.89,
|
| 1443 |
+
"learning_rate": 8.771929824561404e-07,
|
| 1444 |
+
"loss": 0.0915,
|
| 1445 |
"step": 1130
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 9.93,
|
| 1449 |
+
"learning_rate": 4.385964912280702e-07,
|
| 1450 |
+
"loss": 0.0422,
|
| 1451 |
"step": 1135
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
"epoch": 9.98,
|
| 1455 |
+
"learning_rate": 0.0,
|
| 1456 |
+
"loss": 0.0607,
|
| 1457 |
"step": 1140
|
| 1458 |
},
|
| 1459 |
{
|
| 1460 |
+
"epoch": 9.98,
|
| 1461 |
+
"eval_accuracy": 0.9975369458128078,
|
| 1462 |
+
"eval_loss": 0.006894771475344896,
|
| 1463 |
+
"eval_runtime": 51.4226,
|
| 1464 |
+
"eval_samples_per_second": 23.686,
|
| 1465 |
+
"eval_steps_per_second": 2.975,
|
| 1466 |
+
"step": 1140
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
+
"epoch": 9.98,
|
| 1470 |
+
"step": 1140,
|
| 1471 |
+
"total_flos": 8.390377142257582e+18,
|
| 1472 |
+
"train_loss": 0.465066796070651,
|
| 1473 |
+
"train_runtime": 8763.5167,
|
| 1474 |
+
"train_samples_per_second": 4.166,
|
| 1475 |
+
"train_steps_per_second": 0.13
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1476 |
}
|
| 1477 |
],
|
| 1478 |
+
"max_steps": 1140,
|
| 1479 |
+
"num_train_epochs": 10,
|
| 1480 |
+
"total_flos": 8.390377142257582e+18,
|
| 1481 |
"trial_name": null,
|
| 1482 |
"trial_params": null
|
| 1483 |
}
|