Commit ·
05603d8
1
Parent(s): 0c068cb
End of training
Browse files- all_results.json +11 -11
- eval_results.json +6 -6
- runs/Feb17_04-03-35_9a9571952ad0/events.out.tfevents.1676626751.9a9571952ad0.117504.2 +3 -0
- train_results.json +6 -6
- trainer_state.json +1404 -594
all_results.json
CHANGED
|
@@ -1,13 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 24.
|
| 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": 23.
|
| 12 |
-
"train_steps_per_second": 0.
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"eval_accuracy": 0.6208920187793427,
|
| 4 |
+
"eval_loss": 0.9128149747848511,
|
| 5 |
+
"eval_runtime": 24.969,
|
| 6 |
+
"eval_samples_per_second": 68.245,
|
| 7 |
+
"eval_steps_per_second": 2.163,
|
| 8 |
+
"total_flos": 3.5642168535562445e+19,
|
| 9 |
+
"train_loss": 0.9839045113987392,
|
| 10 |
+
"train_runtime": 19975.6243,
|
| 11 |
+
"train_samples_per_second": 23.025,
|
| 12 |
+
"train_steps_per_second": 0.18
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 24.
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 2.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"eval_accuracy": 0.6208920187793427,
|
| 4 |
+
"eval_loss": 0.9128149747848511,
|
| 5 |
+
"eval_runtime": 24.969,
|
| 6 |
+
"eval_samples_per_second": 68.245,
|
| 7 |
+
"eval_steps_per_second": 2.163
|
| 8 |
}
|
runs/Feb17_04-03-35_9a9571952ad0/events.out.tfevents.1676626751.9a9571952ad0.117504.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e0a11ef24b46d8c1442845af4c604f22117a5a3d8ddda267b8922a7e46eeb08b
|
| 3 |
+
size 363
|
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": 23.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 30.0,
|
| 3 |
+
"total_flos": 3.5642168535562445e+19,
|
| 4 |
+
"train_loss": 0.9839045113987392,
|
| 5 |
+
"train_runtime": 19975.6243,
|
| 6 |
+
"train_samples_per_second": 23.025,
|
| 7 |
+
"train_steps_per_second": 0.18
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1645 +1,2455 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "large-algae-vit-wirs/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": 0.08,
|
| 12 |
-
"learning_rate":
|
| 13 |
-
"loss": 1.
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.17,
|
| 18 |
-
"learning_rate":
|
| 19 |
-
"loss": 1.
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.25,
|
| 24 |
-
"learning_rate":
|
| 25 |
-
"loss": 1.
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.33,
|
| 30 |
-
"learning_rate":
|
| 31 |
-
"loss": 1.
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.42,
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss": 1.
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.5,
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss": 1.
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.58,
|
| 48 |
-
"learning_rate":
|
| 49 |
-
"loss": 1.
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.67,
|
| 54 |
-
"learning_rate": 1.
|
| 55 |
-
"loss": 1.
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.75,
|
| 60 |
-
"learning_rate": 1.
|
| 61 |
-
"loss": 1.
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.83,
|
| 66 |
-
"learning_rate":
|
| 67 |
-
"loss": 1.
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.92,
|
| 72 |
-
"learning_rate":
|
| 73 |
-
"loss": 1.
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 1.0,
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss": 1.
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 1.0,
|
| 84 |
-
"eval_accuracy": 0.
|
| 85 |
-
"eval_loss":
|
| 86 |
-
"eval_runtime": 23.
|
| 87 |
-
"eval_samples_per_second":
|
| 88 |
-
"eval_steps_per_second": 2.
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 1.08,
|
| 93 |
-
"learning_rate":
|
| 94 |
-
"loss": 1.
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 1.17,
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"loss": 1.
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 1.25,
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss": 1.
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 1.33,
|
| 111 |
-
"learning_rate":
|
| 112 |
-
"loss": 1.
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 1.42,
|
| 117 |
-
"learning_rate":
|
| 118 |
-
"loss": 1.
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 1.5,
|
| 123 |
-
"learning_rate":
|
| 124 |
-
"loss": 1.
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 1.58,
|
| 129 |
-
"learning_rate":
|
| 130 |
-
"loss": 1.
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 1.67,
|
| 135 |
-
"learning_rate":
|
| 136 |
-
"loss": 1.
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 1.75,
|
| 141 |
-
"learning_rate":
|
| 142 |
-
"loss": 1.
|
| 143 |
"step": 210
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 1.83,
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 1.
|
| 149 |
"step": 220
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 1.92,
|
| 153 |
-
"learning_rate":
|
| 154 |
-
"loss": 1.
|
| 155 |
"step": 230
|
| 156 |
},
|
| 157 |
{
|
| 158 |
"epoch": 2.0,
|
| 159 |
-
"learning_rate":
|
| 160 |
-
"loss":
|
| 161 |
"step": 240
|
| 162 |
},
|
| 163 |
{
|
| 164 |
"epoch": 2.0,
|
| 165 |
-
"eval_accuracy": 0.
|
| 166 |
-
"eval_loss":
|
| 167 |
-
"eval_runtime": 23.
|
| 168 |
-
"eval_samples_per_second":
|
| 169 |
-
"eval_steps_per_second": 2.
|
| 170 |
"step": 240
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 2.08,
|
| 174 |
-
"learning_rate":
|
| 175 |
-
"loss": 1.
|
| 176 |
"step": 250
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 2.17,
|
| 180 |
-
"learning_rate":
|
| 181 |
-
"loss": 1.
|
| 182 |
"step": 260
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 2.25,
|
| 186 |
-
"learning_rate":
|
| 187 |
-
"loss": 1.
|
| 188 |
"step": 270
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 2.33,
|
| 192 |
-
"learning_rate":
|
| 193 |
-
"loss": 1.
|
| 194 |
"step": 280
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 2.42,
|
| 198 |
-
"learning_rate": 4.
|
| 199 |
-
"loss": 1.
|
| 200 |
"step": 290
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 2.5,
|
| 204 |
-
"learning_rate": 4.
|
| 205 |
-
"loss": 1.
|
| 206 |
"step": 300
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 2.58,
|
| 210 |
-
"learning_rate": 4.
|
| 211 |
-
"loss":
|
| 212 |
"step": 310
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 2.67,
|
| 216 |
-
"learning_rate": 4.
|
| 217 |
-
"loss": 1.
|
| 218 |
"step": 320
|
| 219 |
},
|
| 220 |
{
|
| 221 |
"epoch": 2.75,
|
| 222 |
-
"learning_rate": 4.
|
| 223 |
-
"loss": 1.
|
| 224 |
"step": 330
|
| 225 |
},
|
| 226 |
{
|
| 227 |
"epoch": 2.83,
|
| 228 |
-
"learning_rate": 4.
|
| 229 |
-
"loss": 1.
|
| 230 |
"step": 340
|
| 231 |
},
|
| 232 |
{
|
| 233 |
"epoch": 2.92,
|
| 234 |
-
"learning_rate": 4.
|
| 235 |
-
"loss":
|
| 236 |
"step": 350
|
| 237 |
},
|
| 238 |
{
|
| 239 |
"epoch": 3.0,
|
| 240 |
-
"learning_rate":
|
| 241 |
-
"loss":
|
| 242 |
"step": 360
|
| 243 |
},
|
| 244 |
{
|
| 245 |
"epoch": 3.0,
|
| 246 |
-
"eval_accuracy": 0.
|
| 247 |
-
"eval_loss":
|
| 248 |
-
"eval_runtime":
|
| 249 |
-
"eval_samples_per_second":
|
| 250 |
-
"eval_steps_per_second": 2.
|
| 251 |
"step": 360
|
| 252 |
},
|
| 253 |
{
|
| 254 |
"epoch": 3.08,
|
| 255 |
-
"learning_rate": 4.
|
| 256 |
-
"loss":
|
| 257 |
"step": 370
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 3.17,
|
| 261 |
-
"learning_rate": 4.
|
| 262 |
-
"loss": 1.
|
| 263 |
"step": 380
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 3.25,
|
| 267 |
-
"learning_rate": 4.
|
| 268 |
-
"loss":
|
| 269 |
"step": 390
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 3.33,
|
| 273 |
-
"learning_rate": 4.
|
| 274 |
-
"loss": 1.
|
| 275 |
"step": 400
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 3.42,
|
| 279 |
-
"learning_rate": 4.
|
| 280 |
-
"loss": 1.
|
| 281 |
"step": 410
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 3.5,
|
| 285 |
-
"learning_rate": 4.
|
| 286 |
-
"loss":
|
| 287 |
"step": 420
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 3.58,
|
| 291 |
-
"learning_rate": 4.
|
| 292 |
-
"loss": 1.
|
| 293 |
"step": 430
|
| 294 |
},
|
| 295 |
{
|
| 296 |
"epoch": 3.67,
|
| 297 |
-
"learning_rate": 4.
|
| 298 |
-
"loss":
|
| 299 |
"step": 440
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 3.75,
|
| 303 |
-
"learning_rate": 4.
|
| 304 |
-
"loss":
|
| 305 |
"step": 450
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 3.83,
|
| 309 |
-
"learning_rate": 4.
|
| 310 |
-
"loss":
|
| 311 |
"step": 460
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 3.92,
|
| 315 |
-
"learning_rate": 4.
|
| 316 |
-
"loss":
|
| 317 |
"step": 470
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"epoch": 4.0,
|
| 321 |
-
"learning_rate": 4.
|
| 322 |
-
"loss":
|
| 323 |
"step": 480
|
| 324 |
},
|
| 325 |
{
|
| 326 |
"epoch": 4.0,
|
| 327 |
-
"eval_accuracy": 0.
|
| 328 |
-
"eval_loss":
|
| 329 |
-
"eval_runtime":
|
| 330 |
-
"eval_samples_per_second":
|
| 331 |
-
"eval_steps_per_second": 2.
|
| 332 |
"step": 480
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 4.08,
|
| 336 |
-
"learning_rate": 4.
|
| 337 |
-
"loss":
|
| 338 |
"step": 490
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 4.17,
|
| 342 |
-
"learning_rate": 4.
|
| 343 |
-
"loss":
|
| 344 |
"step": 500
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 4.25,
|
| 348 |
-
"learning_rate": 4.
|
| 349 |
-
"loss":
|
| 350 |
"step": 510
|
| 351 |
},
|
| 352 |
{
|
| 353 |
"epoch": 4.33,
|
| 354 |
-
"learning_rate": 4.
|
| 355 |
-
"loss":
|
| 356 |
"step": 520
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 4.42,
|
| 360 |
-
"learning_rate": 4.
|
| 361 |
-
"loss":
|
| 362 |
"step": 530
|
| 363 |
},
|
| 364 |
{
|
| 365 |
"epoch": 4.5,
|
| 366 |
-
"learning_rate": 4.
|
| 367 |
-
"loss":
|
| 368 |
"step": 540
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 4.58,
|
| 372 |
-
"learning_rate": 4.
|
| 373 |
-
"loss": 1.
|
| 374 |
"step": 550
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 4.67,
|
| 378 |
-
"learning_rate": 4.
|
| 379 |
-
"loss":
|
| 380 |
"step": 560
|
| 381 |
},
|
| 382 |
{
|
| 383 |
"epoch": 4.75,
|
| 384 |
-
"learning_rate": 4.
|
| 385 |
-
"loss":
|
| 386 |
"step": 570
|
| 387 |
},
|
| 388 |
{
|
| 389 |
"epoch": 4.83,
|
| 390 |
-
"learning_rate": 4.
|
| 391 |
-
"loss":
|
| 392 |
"step": 580
|
| 393 |
},
|
| 394 |
{
|
| 395 |
"epoch": 4.92,
|
| 396 |
-
"learning_rate": 4.
|
| 397 |
-
"loss":
|
| 398 |
"step": 590
|
| 399 |
},
|
| 400 |
{
|
| 401 |
"epoch": 5.0,
|
| 402 |
-
"learning_rate": 4.
|
| 403 |
-
"loss":
|
| 404 |
"step": 600
|
| 405 |
},
|
| 406 |
{
|
| 407 |
"epoch": 5.0,
|
| 408 |
-
"eval_accuracy": 0.
|
| 409 |
-
"eval_loss":
|
| 410 |
-
"eval_runtime":
|
| 411 |
-
"eval_samples_per_second":
|
| 412 |
-
"eval_steps_per_second": 2.
|
| 413 |
"step": 600
|
| 414 |
},
|
| 415 |
{
|
| 416 |
"epoch": 5.08,
|
| 417 |
-
"learning_rate": 4.
|
| 418 |
-
"loss":
|
| 419 |
"step": 610
|
| 420 |
},
|
| 421 |
{
|
| 422 |
"epoch": 5.17,
|
| 423 |
-
"learning_rate": 4.
|
| 424 |
-
"loss":
|
| 425 |
"step": 620
|
| 426 |
},
|
| 427 |
{
|
| 428 |
"epoch": 5.25,
|
| 429 |
-
"learning_rate": 4.
|
| 430 |
-
"loss":
|
| 431 |
"step": 630
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 5.33,
|
| 435 |
-
"learning_rate": 4.
|
| 436 |
-
"loss":
|
| 437 |
"step": 640
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 5.42,
|
| 441 |
-
"learning_rate": 4.
|
| 442 |
-
"loss":
|
| 443 |
"step": 650
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"epoch": 5.5,
|
| 447 |
-
"learning_rate": 4.
|
| 448 |
-
"loss":
|
| 449 |
"step": 660
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 5.58,
|
| 453 |
-
"learning_rate": 4.
|
| 454 |
-
"loss":
|
| 455 |
"step": 670
|
| 456 |
},
|
| 457 |
{
|
| 458 |
"epoch": 5.67,
|
| 459 |
-
"learning_rate":
|
| 460 |
-
"loss":
|
| 461 |
"step": 680
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 5.75,
|
| 465 |
-
"learning_rate":
|
| 466 |
-
"loss":
|
| 467 |
"step": 690
|
| 468 |
},
|
| 469 |
{
|
| 470 |
"epoch": 5.83,
|
| 471 |
-
"learning_rate":
|
| 472 |
-
"loss": 0.
|
| 473 |
"step": 700
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 5.92,
|
| 477 |
-
"learning_rate":
|
| 478 |
-
"loss":
|
| 479 |
"step": 710
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 6.0,
|
| 483 |
-
"learning_rate":
|
| 484 |
-
"loss":
|
| 485 |
"step": 720
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"epoch": 6.0,
|
| 489 |
-
"eval_accuracy": 0.
|
| 490 |
-
"eval_loss":
|
| 491 |
-
"eval_runtime": 23.
|
| 492 |
-
"eval_samples_per_second":
|
| 493 |
-
"eval_steps_per_second": 2.
|
| 494 |
"step": 720
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 6.08,
|
| 498 |
-
"learning_rate":
|
| 499 |
-
"loss":
|
| 500 |
"step": 730
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 6.17,
|
| 504 |
-
"learning_rate":
|
| 505 |
-
"loss":
|
| 506 |
"step": 740
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 6.25,
|
| 510 |
-
"learning_rate":
|
| 511 |
-
"loss":
|
| 512 |
"step": 750
|
| 513 |
},
|
| 514 |
{
|
| 515 |
"epoch": 6.33,
|
| 516 |
-
"learning_rate":
|
| 517 |
-
"loss":
|
| 518 |
"step": 760
|
| 519 |
},
|
| 520 |
{
|
| 521 |
"epoch": 6.42,
|
| 522 |
-
"learning_rate":
|
| 523 |
-
"loss": 0.
|
| 524 |
"step": 770
|
| 525 |
},
|
| 526 |
{
|
| 527 |
"epoch": 6.5,
|
| 528 |
-
"learning_rate":
|
| 529 |
-
"loss":
|
| 530 |
"step": 780
|
| 531 |
},
|
| 532 |
{
|
| 533 |
"epoch": 6.58,
|
| 534 |
-
"learning_rate":
|
| 535 |
-
"loss": 0.
|
| 536 |
"step": 790
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 6.67,
|
| 540 |
-
"learning_rate":
|
| 541 |
-
"loss":
|
| 542 |
"step": 800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 6.75,
|
| 546 |
-
"learning_rate":
|
| 547 |
-
"loss": 0.
|
| 548 |
"step": 810
|
| 549 |
},
|
| 550 |
{
|
| 551 |
"epoch": 6.83,
|
| 552 |
-
"learning_rate":
|
| 553 |
-
"loss":
|
| 554 |
"step": 820
|
| 555 |
},
|
| 556 |
{
|
| 557 |
"epoch": 6.92,
|
| 558 |
-
"learning_rate":
|
| 559 |
-
"loss": 0.
|
| 560 |
"step": 830
|
| 561 |
},
|
| 562 |
{
|
| 563 |
"epoch": 7.0,
|
| 564 |
-
"learning_rate":
|
| 565 |
-
"loss":
|
| 566 |
"step": 840
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 7.0,
|
| 570 |
-
"eval_accuracy": 0.
|
| 571 |
-
"eval_loss":
|
| 572 |
-
"eval_runtime": 23.
|
| 573 |
-
"eval_samples_per_second":
|
| 574 |
-
"eval_steps_per_second": 2.
|
| 575 |
"step": 840
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"epoch": 7.08,
|
| 579 |
-
"learning_rate":
|
| 580 |
-
"loss":
|
| 581 |
"step": 850
|
| 582 |
},
|
| 583 |
{
|
| 584 |
"epoch": 7.17,
|
| 585 |
-
"learning_rate":
|
| 586 |
-
"loss": 0.
|
| 587 |
"step": 860
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 7.25,
|
| 591 |
-
"learning_rate":
|
| 592 |
-
"loss": 0.
|
| 593 |
"step": 870
|
| 594 |
},
|
| 595 |
{
|
| 596 |
"epoch": 7.33,
|
| 597 |
-
"learning_rate":
|
| 598 |
-
"loss": 0.
|
| 599 |
"step": 880
|
| 600 |
},
|
| 601 |
{
|
| 602 |
"epoch": 7.42,
|
| 603 |
-
"learning_rate":
|
| 604 |
-
"loss":
|
| 605 |
"step": 890
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 7.5,
|
| 609 |
-
"learning_rate":
|
| 610 |
-
"loss":
|
| 611 |
"step": 900
|
| 612 |
},
|
| 613 |
{
|
| 614 |
"epoch": 7.58,
|
| 615 |
-
"learning_rate":
|
| 616 |
-
"loss": 0.
|
| 617 |
"step": 910
|
| 618 |
},
|
| 619 |
{
|
| 620 |
"epoch": 7.67,
|
| 621 |
-
"learning_rate":
|
| 622 |
-
"loss": 0.
|
| 623 |
"step": 920
|
| 624 |
},
|
| 625 |
{
|
| 626 |
"epoch": 7.75,
|
| 627 |
-
"learning_rate":
|
| 628 |
-
"loss":
|
| 629 |
"step": 930
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 7.83,
|
| 633 |
-
"learning_rate":
|
| 634 |
-
"loss": 0.
|
| 635 |
"step": 940
|
| 636 |
},
|
| 637 |
{
|
| 638 |
"epoch": 7.92,
|
| 639 |
-
"learning_rate":
|
| 640 |
-
"loss":
|
| 641 |
"step": 950
|
| 642 |
},
|
| 643 |
{
|
| 644 |
"epoch": 8.0,
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 960
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 8.0,
|
| 651 |
-
"eval_accuracy": 0.
|
| 652 |
-
"eval_loss":
|
| 653 |
-
"eval_runtime": 23.
|
| 654 |
-
"eval_samples_per_second": 72.
|
| 655 |
-
"eval_steps_per_second": 2.
|
| 656 |
"step": 960
|
| 657 |
},
|
| 658 |
{
|
| 659 |
"epoch": 8.08,
|
| 660 |
-
"learning_rate":
|
| 661 |
-
"loss": 0.
|
| 662 |
"step": 970
|
| 663 |
},
|
| 664 |
{
|
| 665 |
"epoch": 8.17,
|
| 666 |
-
"learning_rate":
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 980
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 8.25,
|
| 672 |
-
"learning_rate":
|
| 673 |
-
"loss": 0.
|
| 674 |
"step": 990
|
| 675 |
},
|
| 676 |
{
|
| 677 |
"epoch": 8.33,
|
| 678 |
-
"learning_rate":
|
| 679 |
-
"loss":
|
| 680 |
"step": 1000
|
| 681 |
},
|
| 682 |
{
|
| 683 |
"epoch": 8.42,
|
| 684 |
-
"learning_rate": 3.
|
| 685 |
-
"loss": 0.
|
| 686 |
"step": 1010
|
| 687 |
},
|
| 688 |
{
|
| 689 |
"epoch": 8.5,
|
| 690 |
-
"learning_rate": 3.
|
| 691 |
-
"loss": 0.
|
| 692 |
"step": 1020
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 8.58,
|
| 696 |
-
"learning_rate": 3.
|
| 697 |
-
"loss":
|
| 698 |
"step": 1030
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 8.67,
|
| 702 |
-
"learning_rate": 3.
|
| 703 |
-
"loss": 0.
|
| 704 |
"step": 1040
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 8.75,
|
| 708 |
-
"learning_rate": 3.
|
| 709 |
-
"loss":
|
| 710 |
"step": 1050
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 8.83,
|
| 714 |
-
"learning_rate": 3.
|
| 715 |
-
"loss":
|
| 716 |
"step": 1060
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 8.92,
|
| 720 |
-
"learning_rate": 3.
|
| 721 |
-
"loss": 0.
|
| 722 |
"step": 1070
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 9.0,
|
| 726 |
-
"learning_rate": 3.
|
| 727 |
-
"loss": 0.
|
| 728 |
"step": 1080
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 9.0,
|
| 732 |
-
"eval_accuracy": 0.
|
| 733 |
-
"eval_loss":
|
| 734 |
-
"eval_runtime": 23.
|
| 735 |
-
"eval_samples_per_second": 72.
|
| 736 |
-
"eval_steps_per_second": 2.
|
| 737 |
"step": 1080
|
| 738 |
},
|
| 739 |
{
|
| 740 |
"epoch": 9.08,
|
| 741 |
-
"learning_rate": 3.
|
| 742 |
-
"loss":
|
| 743 |
"step": 1090
|
| 744 |
},
|
| 745 |
{
|
| 746 |
"epoch": 9.17,
|
| 747 |
-
"learning_rate": 3.
|
| 748 |
-
"loss": 0.
|
| 749 |
"step": 1100
|
| 750 |
},
|
| 751 |
{
|
| 752 |
"epoch": 9.25,
|
| 753 |
-
"learning_rate":
|
| 754 |
-
"loss": 0.
|
| 755 |
"step": 1110
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 9.33,
|
| 759 |
-
"learning_rate":
|
| 760 |
-
"loss":
|
| 761 |
"step": 1120
|
| 762 |
},
|
| 763 |
{
|
| 764 |
"epoch": 9.42,
|
| 765 |
-
"learning_rate":
|
| 766 |
-
"loss": 0.
|
| 767 |
"step": 1130
|
| 768 |
},
|
| 769 |
{
|
| 770 |
"epoch": 9.5,
|
| 771 |
-
"learning_rate":
|
| 772 |
-
"loss": 0.
|
| 773 |
"step": 1140
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 9.58,
|
| 777 |
-
"learning_rate":
|
| 778 |
-
"loss":
|
| 779 |
"step": 1150
|
| 780 |
},
|
| 781 |
{
|
| 782 |
"epoch": 9.67,
|
| 783 |
-
"learning_rate":
|
| 784 |
-
"loss": 0.
|
| 785 |
"step": 1160
|
| 786 |
},
|
| 787 |
{
|
| 788 |
"epoch": 9.75,
|
| 789 |
-
"learning_rate":
|
| 790 |
-
"loss": 0.
|
| 791 |
"step": 1170
|
| 792 |
},
|
| 793 |
{
|
| 794 |
"epoch": 9.83,
|
| 795 |
-
"learning_rate":
|
| 796 |
-
"loss": 0.
|
| 797 |
"step": 1180
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 9.92,
|
| 801 |
-
"learning_rate":
|
| 802 |
-
"loss": 0.
|
| 803 |
"step": 1190
|
| 804 |
},
|
| 805 |
{
|
| 806 |
"epoch": 10.0,
|
| 807 |
-
"learning_rate":
|
| 808 |
-
"loss": 0.
|
| 809 |
"step": 1200
|
| 810 |
},
|
| 811 |
{
|
| 812 |
"epoch": 10.0,
|
| 813 |
-
"eval_accuracy": 0.
|
| 814 |
-
"eval_loss":
|
| 815 |
-
"eval_runtime": 23.
|
| 816 |
-
"eval_samples_per_second":
|
| 817 |
-
"eval_steps_per_second": 2.
|
| 818 |
"step": 1200
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 10.08,
|
| 822 |
-
"learning_rate":
|
| 823 |
-
"loss": 0.
|
| 824 |
"step": 1210
|
| 825 |
},
|
| 826 |
{
|
| 827 |
"epoch": 10.17,
|
| 828 |
-
"learning_rate":
|
| 829 |
-
"loss":
|
| 830 |
"step": 1220
|
| 831 |
},
|
| 832 |
{
|
| 833 |
"epoch": 10.25,
|
| 834 |
-
"learning_rate":
|
| 835 |
-
"loss": 0.
|
| 836 |
"step": 1230
|
| 837 |
},
|
| 838 |
{
|
| 839 |
"epoch": 10.33,
|
| 840 |
-
"learning_rate":
|
| 841 |
-
"loss": 0.
|
| 842 |
"step": 1240
|
| 843 |
},
|
| 844 |
{
|
| 845 |
"epoch": 10.42,
|
| 846 |
-
"learning_rate":
|
| 847 |
-
"loss":
|
| 848 |
"step": 1250
|
| 849 |
},
|
| 850 |
{
|
| 851 |
"epoch": 10.5,
|
| 852 |
-
"learning_rate":
|
| 853 |
-
"loss": 0.
|
| 854 |
"step": 1260
|
| 855 |
},
|
| 856 |
{
|
| 857 |
"epoch": 10.58,
|
| 858 |
-
"learning_rate":
|
| 859 |
-
"loss": 0.
|
| 860 |
"step": 1270
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 10.67,
|
| 864 |
-
"learning_rate":
|
| 865 |
-
"loss": 0.
|
| 866 |
"step": 1280
|
| 867 |
},
|
| 868 |
{
|
| 869 |
"epoch": 10.75,
|
| 870 |
-
"learning_rate":
|
| 871 |
-
"loss": 0.
|
| 872 |
"step": 1290
|
| 873 |
},
|
| 874 |
{
|
| 875 |
"epoch": 10.83,
|
| 876 |
-
"learning_rate":
|
| 877 |
-
"loss": 0.
|
| 878 |
"step": 1300
|
| 879 |
},
|
| 880 |
{
|
| 881 |
"epoch": 10.92,
|
| 882 |
-
"learning_rate":
|
| 883 |
-
"loss": 0.
|
| 884 |
"step": 1310
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 11.0,
|
| 888 |
-
"learning_rate":
|
| 889 |
-
"loss": 0.
|
| 890 |
"step": 1320
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 11.0,
|
| 894 |
-
"eval_accuracy": 0.
|
| 895 |
-
"eval_loss":
|
| 896 |
-
"eval_runtime": 23.
|
| 897 |
-
"eval_samples_per_second": 73.
|
| 898 |
-
"eval_steps_per_second": 2.
|
| 899 |
"step": 1320
|
| 900 |
},
|
| 901 |
{
|
| 902 |
"epoch": 11.08,
|
| 903 |
-
"learning_rate":
|
| 904 |
-
"loss":
|
| 905 |
"step": 1330
|
| 906 |
},
|
| 907 |
{
|
| 908 |
"epoch": 11.17,
|
| 909 |
-
"learning_rate":
|
| 910 |
-
"loss": 0.
|
| 911 |
"step": 1340
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 11.25,
|
| 915 |
-
"learning_rate":
|
| 916 |
-
"loss": 0.
|
| 917 |
"step": 1350
|
| 918 |
},
|
| 919 |
{
|
| 920 |
"epoch": 11.33,
|
| 921 |
-
"learning_rate":
|
| 922 |
-
"loss": 0.
|
| 923 |
"step": 1360
|
| 924 |
},
|
| 925 |
{
|
| 926 |
"epoch": 11.42,
|
| 927 |
-
"learning_rate":
|
| 928 |
-
"loss": 0.
|
| 929 |
"step": 1370
|
| 930 |
},
|
| 931 |
{
|
| 932 |
"epoch": 11.5,
|
| 933 |
-
"learning_rate":
|
| 934 |
-
"loss": 0.
|
| 935 |
"step": 1380
|
| 936 |
},
|
| 937 |
{
|
| 938 |
"epoch": 11.58,
|
| 939 |
-
"learning_rate":
|
| 940 |
-
"loss": 0.
|
| 941 |
"step": 1390
|
| 942 |
},
|
| 943 |
{
|
| 944 |
"epoch": 11.67,
|
| 945 |
-
"learning_rate":
|
| 946 |
-
"loss": 0.
|
| 947 |
"step": 1400
|
| 948 |
},
|
| 949 |
{
|
| 950 |
"epoch": 11.75,
|
| 951 |
-
"learning_rate":
|
| 952 |
-
"loss": 0.
|
| 953 |
"step": 1410
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 11.83,
|
| 957 |
-
"learning_rate":
|
| 958 |
-
"loss": 0.
|
| 959 |
"step": 1420
|
| 960 |
},
|
| 961 |
{
|
| 962 |
"epoch": 11.92,
|
| 963 |
-
"learning_rate":
|
| 964 |
-
"loss": 0.
|
| 965 |
"step": 1430
|
| 966 |
},
|
| 967 |
{
|
| 968 |
"epoch": 12.0,
|
| 969 |
-
"learning_rate":
|
| 970 |
-
"loss": 0.
|
| 971 |
"step": 1440
|
| 972 |
},
|
| 973 |
{
|
| 974 |
"epoch": 12.0,
|
| 975 |
-
"eval_accuracy": 0.
|
| 976 |
-
"eval_loss":
|
| 977 |
-
"eval_runtime": 23.
|
| 978 |
-
"eval_samples_per_second":
|
| 979 |
-
"eval_steps_per_second": 2.
|
| 980 |
"step": 1440
|
| 981 |
},
|
| 982 |
{
|
| 983 |
"epoch": 12.08,
|
| 984 |
-
"learning_rate":
|
| 985 |
-
"loss": 0.
|
| 986 |
"step": 1450
|
| 987 |
},
|
| 988 |
{
|
| 989 |
"epoch": 12.17,
|
| 990 |
-
"learning_rate":
|
| 991 |
-
"loss": 0.
|
| 992 |
"step": 1460
|
| 993 |
},
|
| 994 |
{
|
| 995 |
"epoch": 12.25,
|
| 996 |
-
"learning_rate":
|
| 997 |
-
"loss":
|
| 998 |
"step": 1470
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
"epoch": 12.33,
|
| 1002 |
-
"learning_rate":
|
| 1003 |
-
"loss": 0.
|
| 1004 |
"step": 1480
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 12.42,
|
| 1008 |
-
"learning_rate":
|
| 1009 |
-
"loss": 0.
|
| 1010 |
"step": 1490
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
"epoch": 12.5,
|
| 1014 |
-
"learning_rate":
|
| 1015 |
-
"loss": 0.
|
| 1016 |
"step": 1500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 12.58,
|
| 1020 |
-
"learning_rate":
|
| 1021 |
-
"loss": 0.
|
| 1022 |
"step": 1510
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
"epoch": 12.67,
|
| 1026 |
-
"learning_rate":
|
| 1027 |
-
"loss": 0.
|
| 1028 |
"step": 1520
|
| 1029 |
},
|
| 1030 |
{
|
| 1031 |
"epoch": 12.75,
|
| 1032 |
-
"learning_rate":
|
| 1033 |
-
"loss":
|
| 1034 |
"step": 1530
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
"epoch": 12.83,
|
| 1038 |
-
"learning_rate":
|
| 1039 |
-
"loss": 0.
|
| 1040 |
"step": 1540
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
"epoch": 12.92,
|
| 1044 |
-
"learning_rate":
|
| 1045 |
-
"loss": 0.
|
| 1046 |
"step": 1550
|
| 1047 |
},
|
| 1048 |
{
|
| 1049 |
"epoch": 13.0,
|
| 1050 |
-
"learning_rate":
|
| 1051 |
-
"loss": 0.
|
| 1052 |
"step": 1560
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
"epoch": 13.0,
|
| 1056 |
-
"eval_accuracy": 0.
|
| 1057 |
-
"eval_loss":
|
| 1058 |
-
"eval_runtime": 23.
|
| 1059 |
-
"eval_samples_per_second": 72.
|
| 1060 |
-
"eval_steps_per_second": 2.
|
| 1061 |
"step": 1560
|
| 1062 |
},
|
| 1063 |
{
|
| 1064 |
"epoch": 13.08,
|
| 1065 |
-
"learning_rate":
|
| 1066 |
-
"loss": 0.
|
| 1067 |
"step": 1570
|
| 1068 |
},
|
| 1069 |
{
|
| 1070 |
"epoch": 13.17,
|
| 1071 |
-
"learning_rate":
|
| 1072 |
-
"loss": 0.
|
| 1073 |
"step": 1580
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
"epoch": 13.25,
|
| 1077 |
-
"learning_rate":
|
| 1078 |
-
"loss": 0.
|
| 1079 |
"step": 1590
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 13.33,
|
| 1083 |
-
"learning_rate":
|
| 1084 |
-
"loss": 0.
|
| 1085 |
"step": 1600
|
| 1086 |
},
|
| 1087 |
{
|
| 1088 |
"epoch": 13.42,
|
| 1089 |
-
"learning_rate":
|
| 1090 |
-
"loss": 0.
|
| 1091 |
"step": 1610
|
| 1092 |
},
|
| 1093 |
{
|
| 1094 |
"epoch": 13.5,
|
| 1095 |
-
"learning_rate":
|
| 1096 |
-
"loss": 0.
|
| 1097 |
"step": 1620
|
| 1098 |
},
|
| 1099 |
{
|
| 1100 |
"epoch": 13.58,
|
| 1101 |
-
"learning_rate":
|
| 1102 |
-
"loss":
|
| 1103 |
"step": 1630
|
| 1104 |
},
|
| 1105 |
{
|
| 1106 |
"epoch": 13.67,
|
| 1107 |
-
"learning_rate":
|
| 1108 |
-
"loss":
|
| 1109 |
"step": 1640
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
"epoch": 13.75,
|
| 1113 |
-
"learning_rate":
|
| 1114 |
-
"loss": 0.
|
| 1115 |
"step": 1650
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
"epoch": 13.83,
|
| 1119 |
-
"learning_rate":
|
| 1120 |
-
"loss": 0.
|
| 1121 |
"step": 1660
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 13.92,
|
| 1125 |
-
"learning_rate":
|
| 1126 |
-
"loss": 0.
|
| 1127 |
"step": 1670
|
| 1128 |
},
|
| 1129 |
{
|
| 1130 |
"epoch": 14.0,
|
| 1131 |
-
"learning_rate":
|
| 1132 |
-
"loss": 0.
|
| 1133 |
"step": 1680
|
| 1134 |
},
|
| 1135 |
{
|
| 1136 |
"epoch": 14.0,
|
| 1137 |
-
"eval_accuracy": 0.
|
| 1138 |
-
"eval_loss":
|
| 1139 |
-
"eval_runtime": 23.
|
| 1140 |
-
"eval_samples_per_second": 71.
|
| 1141 |
-
"eval_steps_per_second": 2.
|
| 1142 |
"step": 1680
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 14.08,
|
| 1146 |
-
"learning_rate":
|
| 1147 |
-
"loss": 0.
|
| 1148 |
"step": 1690
|
| 1149 |
},
|
| 1150 |
{
|
| 1151 |
"epoch": 14.17,
|
| 1152 |
-
"learning_rate":
|
| 1153 |
-
"loss": 0.
|
| 1154 |
"step": 1700
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
"epoch": 14.25,
|
| 1158 |
-
"learning_rate":
|
| 1159 |
-
"loss": 0.
|
| 1160 |
"step": 1710
|
| 1161 |
},
|
| 1162 |
{
|
| 1163 |
"epoch": 14.33,
|
| 1164 |
-
"learning_rate":
|
| 1165 |
-
"loss": 0.
|
| 1166 |
"step": 1720
|
| 1167 |
},
|
| 1168 |
{
|
| 1169 |
"epoch": 14.42,
|
| 1170 |
-
"learning_rate":
|
| 1171 |
-
"loss": 0.
|
| 1172 |
"step": 1730
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 14.5,
|
| 1176 |
-
"learning_rate":
|
| 1177 |
-
"loss": 0.
|
| 1178 |
"step": 1740
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
"epoch": 14.58,
|
| 1182 |
-
"learning_rate":
|
| 1183 |
-
"loss": 0.
|
| 1184 |
"step": 1750
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
"epoch": 14.67,
|
| 1188 |
-
"learning_rate":
|
| 1189 |
-
"loss": 0.
|
| 1190 |
"step": 1760
|
| 1191 |
},
|
| 1192 |
{
|
| 1193 |
"epoch": 14.75,
|
| 1194 |
-
"learning_rate":
|
| 1195 |
-
"loss": 0.
|
| 1196 |
"step": 1770
|
| 1197 |
},
|
| 1198 |
{
|
| 1199 |
"epoch": 14.83,
|
| 1200 |
-
"learning_rate":
|
| 1201 |
-
"loss":
|
| 1202 |
"step": 1780
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
"epoch": 14.92,
|
| 1206 |
-
"learning_rate":
|
| 1207 |
-
"loss": 0.
|
| 1208 |
"step": 1790
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
"epoch": 15.0,
|
| 1212 |
-
"learning_rate":
|
| 1213 |
-
"loss": 0.
|
| 1214 |
"step": 1800
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 15.0,
|
| 1218 |
-
"eval_accuracy": 0.
|
| 1219 |
-
"eval_loss":
|
| 1220 |
-
"eval_runtime": 23.
|
| 1221 |
-
"eval_samples_per_second": 71.
|
| 1222 |
-
"eval_steps_per_second": 2.
|
| 1223 |
"step": 1800
|
| 1224 |
},
|
| 1225 |
{
|
| 1226 |
"epoch": 15.08,
|
| 1227 |
-
"learning_rate":
|
| 1228 |
-
"loss": 0.
|
| 1229 |
"step": 1810
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
"epoch": 15.17,
|
| 1233 |
-
"learning_rate":
|
| 1234 |
-
"loss": 0.
|
| 1235 |
"step": 1820
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 15.25,
|
| 1239 |
-
"learning_rate":
|
| 1240 |
-
"loss": 0.
|
| 1241 |
"step": 1830
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
"epoch": 15.33,
|
| 1245 |
-
"learning_rate":
|
| 1246 |
-
"loss": 0.
|
| 1247 |
"step": 1840
|
| 1248 |
},
|
| 1249 |
{
|
| 1250 |
"epoch": 15.42,
|
| 1251 |
-
"learning_rate":
|
| 1252 |
-
"loss": 0.
|
| 1253 |
"step": 1850
|
| 1254 |
},
|
| 1255 |
{
|
| 1256 |
"epoch": 15.5,
|
| 1257 |
-
"learning_rate":
|
| 1258 |
-
"loss": 0.
|
| 1259 |
"step": 1860
|
| 1260 |
},
|
| 1261 |
{
|
| 1262 |
"epoch": 15.58,
|
| 1263 |
-
"learning_rate":
|
| 1264 |
-
"loss": 0.
|
| 1265 |
"step": 1870
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
"epoch": 15.67,
|
| 1269 |
-
"learning_rate":
|
| 1270 |
-
"loss":
|
| 1271 |
"step": 1880
|
| 1272 |
},
|
| 1273 |
{
|
| 1274 |
"epoch": 15.75,
|
| 1275 |
-
"learning_rate":
|
| 1276 |
-
"loss": 0.
|
| 1277 |
"step": 1890
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 15.83,
|
| 1281 |
-
"learning_rate":
|
| 1282 |
-
"loss": 0.
|
| 1283 |
"step": 1900
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
"epoch": 15.92,
|
| 1287 |
-
"learning_rate":
|
| 1288 |
-
"loss": 0.
|
| 1289 |
"step": 1910
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
"epoch": 16.0,
|
| 1293 |
-
"learning_rate":
|
| 1294 |
-
"loss": 0.
|
| 1295 |
"step": 1920
|
| 1296 |
},
|
| 1297 |
{
|
| 1298 |
"epoch": 16.0,
|
| 1299 |
-
"eval_accuracy": 0.
|
| 1300 |
-
"eval_loss":
|
| 1301 |
-
"eval_runtime": 23.
|
| 1302 |
-
"eval_samples_per_second":
|
| 1303 |
-
"eval_steps_per_second": 2.
|
| 1304 |
"step": 1920
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
"epoch": 16.08,
|
| 1308 |
-
"learning_rate":
|
| 1309 |
-
"loss": 0.
|
| 1310 |
"step": 1930
|
| 1311 |
},
|
| 1312 |
{
|
| 1313 |
"epoch": 16.17,
|
| 1314 |
-
"learning_rate":
|
| 1315 |
-
"loss": 0.
|
| 1316 |
"step": 1940
|
| 1317 |
},
|
| 1318 |
{
|
| 1319 |
"epoch": 16.25,
|
| 1320 |
-
"learning_rate":
|
| 1321 |
-
"loss": 0.
|
| 1322 |
"step": 1950
|
| 1323 |
},
|
| 1324 |
{
|
| 1325 |
"epoch": 16.33,
|
| 1326 |
-
"learning_rate":
|
| 1327 |
-
"loss": 0.
|
| 1328 |
"step": 1960
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
"epoch": 16.42,
|
| 1332 |
-
"learning_rate":
|
| 1333 |
-
"loss": 0.
|
| 1334 |
"step": 1970
|
| 1335 |
},
|
| 1336 |
{
|
| 1337 |
"epoch": 16.5,
|
| 1338 |
-
"learning_rate":
|
| 1339 |
-
"loss": 0.
|
| 1340 |
"step": 1980
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 16.58,
|
| 1344 |
-
"learning_rate":
|
| 1345 |
-
"loss": 0.
|
| 1346 |
"step": 1990
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
"epoch": 16.67,
|
| 1350 |
-
"learning_rate":
|
| 1351 |
-
"loss": 0.
|
| 1352 |
"step": 2000
|
| 1353 |
},
|
| 1354 |
{
|
| 1355 |
"epoch": 16.75,
|
| 1356 |
-
"learning_rate":
|
| 1357 |
-
"loss": 0.
|
| 1358 |
"step": 2010
|
| 1359 |
},
|
| 1360 |
{
|
| 1361 |
"epoch": 16.83,
|
| 1362 |
-
"learning_rate":
|
| 1363 |
-
"loss": 0.
|
| 1364 |
"step": 2020
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
"epoch": 16.92,
|
| 1368 |
-
"learning_rate":
|
| 1369 |
-
"loss": 0.
|
| 1370 |
"step": 2030
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
"epoch": 17.0,
|
| 1374 |
-
"learning_rate":
|
| 1375 |
-
"loss": 0.
|
| 1376 |
"step": 2040
|
| 1377 |
},
|
| 1378 |
{
|
| 1379 |
"epoch": 17.0,
|
| 1380 |
-
"eval_accuracy": 0.
|
| 1381 |
-
"eval_loss":
|
| 1382 |
-
"eval_runtime": 23.
|
| 1383 |
-
"eval_samples_per_second": 71.
|
| 1384 |
-
"eval_steps_per_second": 2.
|
| 1385 |
"step": 2040
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
"epoch": 17.08,
|
| 1389 |
-
"learning_rate":
|
| 1390 |
-
"loss": 0.
|
| 1391 |
"step": 2050
|
| 1392 |
},
|
| 1393 |
{
|
| 1394 |
"epoch": 17.17,
|
| 1395 |
-
"learning_rate":
|
| 1396 |
-
"loss": 0.
|
| 1397 |
"step": 2060
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
"epoch": 17.25,
|
| 1401 |
-
"learning_rate":
|
| 1402 |
-
"loss": 0.
|
| 1403 |
"step": 2070
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 17.33,
|
| 1407 |
-
"learning_rate":
|
| 1408 |
-
"loss": 0.
|
| 1409 |
"step": 2080
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
"epoch": 17.42,
|
| 1413 |
-
"learning_rate":
|
| 1414 |
-
"loss": 0.
|
| 1415 |
"step": 2090
|
| 1416 |
},
|
| 1417 |
{
|
| 1418 |
"epoch": 17.5,
|
| 1419 |
-
"learning_rate":
|
| 1420 |
-
"loss": 0.
|
| 1421 |
"step": 2100
|
| 1422 |
},
|
| 1423 |
{
|
| 1424 |
"epoch": 17.58,
|
| 1425 |
-
"learning_rate":
|
| 1426 |
-
"loss": 0.
|
| 1427 |
"step": 2110
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 17.67,
|
| 1431 |
-
"learning_rate":
|
| 1432 |
-
"loss": 0.
|
| 1433 |
"step": 2120
|
| 1434 |
},
|
| 1435 |
{
|
| 1436 |
"epoch": 17.75,
|
| 1437 |
-
"learning_rate":
|
| 1438 |
-
"loss": 0.
|
| 1439 |
"step": 2130
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
"epoch": 17.83,
|
| 1443 |
-
"learning_rate":
|
| 1444 |
-
"loss": 0.
|
| 1445 |
"step": 2140
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 17.92,
|
| 1449 |
-
"learning_rate":
|
| 1450 |
-
"loss": 0.
|
| 1451 |
"step": 2150
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
"epoch": 18.0,
|
| 1455 |
-
"learning_rate":
|
| 1456 |
-
"loss": 0.
|
| 1457 |
"step": 2160
|
| 1458 |
},
|
| 1459 |
{
|
| 1460 |
"epoch": 18.0,
|
| 1461 |
-
"eval_accuracy": 0.
|
| 1462 |
-
"eval_loss":
|
| 1463 |
-
"eval_runtime":
|
| 1464 |
-
"eval_samples_per_second":
|
| 1465 |
-
"eval_steps_per_second": 2.
|
| 1466 |
"step": 2160
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
"epoch": 18.08,
|
| 1470 |
-
"learning_rate":
|
| 1471 |
-
"loss": 0.
|
| 1472 |
"step": 2170
|
| 1473 |
},
|
| 1474 |
{
|
| 1475 |
"epoch": 18.17,
|
| 1476 |
-
"learning_rate":
|
| 1477 |
-
"loss": 0.
|
| 1478 |
"step": 2180
|
| 1479 |
},
|
| 1480 |
{
|
| 1481 |
"epoch": 18.25,
|
| 1482 |
-
"learning_rate":
|
| 1483 |
-
"loss": 0.
|
| 1484 |
"step": 2190
|
| 1485 |
},
|
| 1486 |
{
|
| 1487 |
"epoch": 18.33,
|
| 1488 |
-
"learning_rate":
|
| 1489 |
-
"loss": 0.
|
| 1490 |
"step": 2200
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
"epoch": 18.42,
|
| 1494 |
-
"learning_rate":
|
| 1495 |
-
"loss":
|
| 1496 |
"step": 2210
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
"epoch": 18.5,
|
| 1500 |
-
"learning_rate":
|
| 1501 |
-
"loss": 0.
|
| 1502 |
"step": 2220
|
| 1503 |
},
|
| 1504 |
{
|
| 1505 |
"epoch": 18.58,
|
| 1506 |
-
"learning_rate":
|
| 1507 |
-
"loss": 0.
|
| 1508 |
"step": 2230
|
| 1509 |
},
|
| 1510 |
{
|
| 1511 |
"epoch": 18.67,
|
| 1512 |
-
"learning_rate":
|
| 1513 |
-
"loss": 0.
|
| 1514 |
"step": 2240
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
"epoch": 18.75,
|
| 1518 |
-
"learning_rate":
|
| 1519 |
-
"loss": 0.
|
| 1520 |
"step": 2250
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
"epoch": 18.83,
|
| 1524 |
-
"learning_rate":
|
| 1525 |
-
"loss": 0.
|
| 1526 |
"step": 2260
|
| 1527 |
},
|
| 1528 |
{
|
| 1529 |
"epoch": 18.92,
|
| 1530 |
-
"learning_rate":
|
| 1531 |
-
"loss": 0.
|
| 1532 |
"step": 2270
|
| 1533 |
},
|
| 1534 |
{
|
| 1535 |
"epoch": 19.0,
|
| 1536 |
-
"learning_rate": 2.
|
| 1537 |
-
"loss": 0.
|
| 1538 |
"step": 2280
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
"epoch": 19.0,
|
| 1542 |
-
"eval_accuracy": 0.
|
| 1543 |
-
"eval_loss":
|
| 1544 |
-
"eval_runtime": 23.
|
| 1545 |
-
"eval_samples_per_second":
|
| 1546 |
-
"eval_steps_per_second": 2.
|
| 1547 |
"step": 2280
|
| 1548 |
},
|
| 1549 |
{
|
| 1550 |
"epoch": 19.08,
|
| 1551 |
-
"learning_rate": 2.
|
| 1552 |
-
"loss": 0.
|
| 1553 |
"step": 2290
|
| 1554 |
},
|
| 1555 |
{
|
| 1556 |
"epoch": 19.17,
|
| 1557 |
-
"learning_rate": 2.
|
| 1558 |
-
"loss": 0.
|
| 1559 |
"step": 2300
|
| 1560 |
},
|
| 1561 |
{
|
| 1562 |
"epoch": 19.25,
|
| 1563 |
-
"learning_rate":
|
| 1564 |
-
"loss": 0.
|
| 1565 |
"step": 2310
|
| 1566 |
},
|
| 1567 |
{
|
| 1568 |
"epoch": 19.33,
|
| 1569 |
-
"learning_rate": 1.
|
| 1570 |
-
"loss": 0.
|
| 1571 |
"step": 2320
|
| 1572 |
},
|
| 1573 |
{
|
| 1574 |
"epoch": 19.42,
|
| 1575 |
-
"learning_rate": 1.
|
| 1576 |
-
"loss": 0.
|
| 1577 |
"step": 2330
|
| 1578 |
},
|
| 1579 |
{
|
| 1580 |
"epoch": 19.5,
|
| 1581 |
-
"learning_rate": 1.
|
| 1582 |
-
"loss": 0.
|
| 1583 |
"step": 2340
|
| 1584 |
},
|
| 1585 |
{
|
| 1586 |
"epoch": 19.58,
|
| 1587 |
-
"learning_rate": 1.
|
| 1588 |
-
"loss": 0.
|
| 1589 |
"step": 2350
|
| 1590 |
},
|
| 1591 |
{
|
| 1592 |
"epoch": 19.67,
|
| 1593 |
-
"learning_rate":
|
| 1594 |
-
"loss": 0.
|
| 1595 |
"step": 2360
|
| 1596 |
},
|
| 1597 |
{
|
| 1598 |
"epoch": 19.75,
|
| 1599 |
-
"learning_rate":
|
| 1600 |
-
"loss": 0.
|
| 1601 |
"step": 2370
|
| 1602 |
},
|
| 1603 |
{
|
| 1604 |
"epoch": 19.83,
|
| 1605 |
-
"learning_rate":
|
| 1606 |
-
"loss": 0.
|
| 1607 |
"step": 2380
|
| 1608 |
},
|
| 1609 |
{
|
| 1610 |
"epoch": 19.92,
|
| 1611 |
-
"learning_rate":
|
| 1612 |
-
"loss": 0.
|
| 1613 |
"step": 2390
|
| 1614 |
},
|
| 1615 |
{
|
| 1616 |
"epoch": 20.0,
|
| 1617 |
-
"learning_rate":
|
| 1618 |
-
"loss": 0.
|
| 1619 |
"step": 2400
|
| 1620 |
},
|
| 1621 |
{
|
| 1622 |
"epoch": 20.0,
|
| 1623 |
-
"eval_accuracy": 0.
|
| 1624 |
-
"eval_loss":
|
| 1625 |
-
"eval_runtime": 23.
|
| 1626 |
-
"eval_samples_per_second":
|
| 1627 |
-
"eval_steps_per_second": 2.
|
| 1628 |
"step": 2400
|
| 1629 |
},
|
| 1630 |
{
|
| 1631 |
-
"epoch": 20.
|
| 1632 |
-
"
|
| 1633 |
-
"
|
| 1634 |
-
"
|
| 1635 |
-
|
| 1636 |
-
|
| 1637 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1638 |
}
|
| 1639 |
],
|
| 1640 |
-
"max_steps":
|
| 1641 |
-
"num_train_epochs":
|
| 1642 |
-
"total_flos":
|
| 1643 |
"trial_name": null,
|
| 1644 |
"trial_params": null
|
| 1645 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.6208920187793427,
|
| 3 |
+
"best_model_checkpoint": "large-algae-vit-wirs/checkpoint-120",
|
| 4 |
+
"epoch": 30.0,
|
| 5 |
+
"global_step": 3600,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.08,
|
| 12 |
+
"learning_rate": 1.388888888888889e-06,
|
| 13 |
+
"loss": 1.3612,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.17,
|
| 18 |
+
"learning_rate": 2.777777777777778e-06,
|
| 19 |
+
"loss": 1.3039,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.25,
|
| 24 |
+
"learning_rate": 4.166666666666667e-06,
|
| 25 |
+
"loss": 1.3076,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.33,
|
| 30 |
+
"learning_rate": 5.555555555555556e-06,
|
| 31 |
+
"loss": 1.2573,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.42,
|
| 36 |
+
"learning_rate": 6.944444444444445e-06,
|
| 37 |
+
"loss": 1.2478,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.5,
|
| 42 |
+
"learning_rate": 8.333333333333334e-06,
|
| 43 |
+
"loss": 1.2011,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.58,
|
| 48 |
+
"learning_rate": 9.722222222222223e-06,
|
| 49 |
+
"loss": 1.1795,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.67,
|
| 54 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 55 |
+
"loss": 1.1731,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.75,
|
| 60 |
+
"learning_rate": 1.25e-05,
|
| 61 |
+
"loss": 1.2165,
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.83,
|
| 66 |
+
"learning_rate": 1.388888888888889e-05,
|
| 67 |
+
"loss": 1.2024,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.92,
|
| 72 |
+
"learning_rate": 1.527777777777778e-05,
|
| 73 |
+
"loss": 1.1816,
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 1.0,
|
| 78 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 79 |
+
"loss": 1.1662,
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 1.0,
|
| 84 |
+
"eval_accuracy": 0.6208920187793427,
|
| 85 |
+
"eval_loss": 0.9128149747848511,
|
| 86 |
+
"eval_runtime": 23.9101,
|
| 87 |
+
"eval_samples_per_second": 71.267,
|
| 88 |
+
"eval_steps_per_second": 2.258,
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 1.08,
|
| 93 |
+
"learning_rate": 1.8055555555555555e-05,
|
| 94 |
+
"loss": 1.1644,
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 1.17,
|
| 99 |
+
"learning_rate": 1.9444444444444445e-05,
|
| 100 |
+
"loss": 1.1601,
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 1.25,
|
| 105 |
+
"learning_rate": 2.0833333333333336e-05,
|
| 106 |
+
"loss": 1.117,
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 1.33,
|
| 111 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 112 |
+
"loss": 1.1592,
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 1.42,
|
| 117 |
+
"learning_rate": 2.361111111111111e-05,
|
| 118 |
+
"loss": 1.1402,
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 1.5,
|
| 123 |
+
"learning_rate": 2.5e-05,
|
| 124 |
+
"loss": 1.1438,
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 1.58,
|
| 129 |
+
"learning_rate": 2.6388888888888892e-05,
|
| 130 |
+
"loss": 1.1215,
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 1.67,
|
| 135 |
+
"learning_rate": 2.777777777777778e-05,
|
| 136 |
+
"loss": 1.1412,
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 1.75,
|
| 141 |
+
"learning_rate": 2.916666666666667e-05,
|
| 142 |
+
"loss": 1.1488,
|
| 143 |
"step": 210
|
| 144 |
},
|
| 145 |
{
|
| 146 |
"epoch": 1.83,
|
| 147 |
+
"learning_rate": 3.055555555555556e-05,
|
| 148 |
+
"loss": 1.1669,
|
| 149 |
"step": 220
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 1.92,
|
| 153 |
+
"learning_rate": 3.194444444444444e-05,
|
| 154 |
+
"loss": 1.12,
|
| 155 |
"step": 230
|
| 156 |
},
|
| 157 |
{
|
| 158 |
"epoch": 2.0,
|
| 159 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 160 |
+
"loss": 1.0885,
|
| 161 |
"step": 240
|
| 162 |
},
|
| 163 |
{
|
| 164 |
"epoch": 2.0,
|
| 165 |
+
"eval_accuracy": 0.613849765258216,
|
| 166 |
+
"eval_loss": 0.9469314217567444,
|
| 167 |
+
"eval_runtime": 23.792,
|
| 168 |
+
"eval_samples_per_second": 71.621,
|
| 169 |
+
"eval_steps_per_second": 2.27,
|
| 170 |
"step": 240
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 2.08,
|
| 174 |
+
"learning_rate": 3.472222222222222e-05,
|
| 175 |
+
"loss": 1.0651,
|
| 176 |
"step": 250
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 2.17,
|
| 180 |
+
"learning_rate": 3.611111111111111e-05,
|
| 181 |
+
"loss": 1.108,
|
| 182 |
"step": 260
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 2.25,
|
| 186 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 187 |
+
"loss": 1.1074,
|
| 188 |
"step": 270
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 2.33,
|
| 192 |
+
"learning_rate": 3.888888888888889e-05,
|
| 193 |
+
"loss": 1.0908,
|
| 194 |
"step": 280
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 2.42,
|
| 198 |
+
"learning_rate": 4.027777777777778e-05,
|
| 199 |
+
"loss": 1.0659,
|
| 200 |
"step": 290
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 2.5,
|
| 204 |
+
"learning_rate": 4.166666666666667e-05,
|
| 205 |
+
"loss": 1.1009,
|
| 206 |
"step": 300
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 2.58,
|
| 210 |
+
"learning_rate": 4.305555555555556e-05,
|
| 211 |
+
"loss": 1.0968,
|
| 212 |
"step": 310
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 2.67,
|
| 216 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 217 |
+
"loss": 1.0859,
|
| 218 |
"step": 320
|
| 219 |
},
|
| 220 |
{
|
| 221 |
"epoch": 2.75,
|
| 222 |
+
"learning_rate": 4.5833333333333334e-05,
|
| 223 |
+
"loss": 1.1342,
|
| 224 |
"step": 330
|
| 225 |
},
|
| 226 |
{
|
| 227 |
"epoch": 2.83,
|
| 228 |
+
"learning_rate": 4.722222222222222e-05,
|
| 229 |
+
"loss": 1.0747,
|
| 230 |
"step": 340
|
| 231 |
},
|
| 232 |
{
|
| 233 |
"epoch": 2.92,
|
| 234 |
+
"learning_rate": 4.8611111111111115e-05,
|
| 235 |
+
"loss": 1.0587,
|
| 236 |
"step": 350
|
| 237 |
},
|
| 238 |
{
|
| 239 |
"epoch": 3.0,
|
| 240 |
+
"learning_rate": 5e-05,
|
| 241 |
+
"loss": 1.1315,
|
| 242 |
"step": 360
|
| 243 |
},
|
| 244 |
{
|
| 245 |
"epoch": 3.0,
|
| 246 |
+
"eval_accuracy": 0.5757042253521126,
|
| 247 |
+
"eval_loss": 1.0919097661972046,
|
| 248 |
+
"eval_runtime": 23.8029,
|
| 249 |
+
"eval_samples_per_second": 71.588,
|
| 250 |
+
"eval_steps_per_second": 2.269,
|
| 251 |
"step": 360
|
| 252 |
},
|
| 253 |
{
|
| 254 |
"epoch": 3.08,
|
| 255 |
+
"learning_rate": 4.984567901234568e-05,
|
| 256 |
+
"loss": 1.0413,
|
| 257 |
"step": 370
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 3.17,
|
| 261 |
+
"learning_rate": 4.969135802469136e-05,
|
| 262 |
+
"loss": 1.0521,
|
| 263 |
"step": 380
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 3.25,
|
| 267 |
+
"learning_rate": 4.9537037037037035e-05,
|
| 268 |
+
"loss": 1.0694,
|
| 269 |
"step": 390
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 3.33,
|
| 273 |
+
"learning_rate": 4.938271604938271e-05,
|
| 274 |
+
"loss": 1.0888,
|
| 275 |
"step": 400
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 3.42,
|
| 279 |
+
"learning_rate": 4.92283950617284e-05,
|
| 280 |
+
"loss": 1.0717,
|
| 281 |
"step": 410
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 3.5,
|
| 285 |
+
"learning_rate": 4.9074074074074075e-05,
|
| 286 |
+
"loss": 1.0258,
|
| 287 |
"step": 420
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 3.58,
|
| 291 |
+
"learning_rate": 4.891975308641975e-05,
|
| 292 |
+
"loss": 1.078,
|
| 293 |
"step": 430
|
| 294 |
},
|
| 295 |
{
|
| 296 |
"epoch": 3.67,
|
| 297 |
+
"learning_rate": 4.876543209876544e-05,
|
| 298 |
+
"loss": 1.0596,
|
| 299 |
"step": 440
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 3.75,
|
| 303 |
+
"learning_rate": 4.8611111111111115e-05,
|
| 304 |
+
"loss": 1.059,
|
| 305 |
"step": 450
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 3.83,
|
| 309 |
+
"learning_rate": 4.845679012345679e-05,
|
| 310 |
+
"loss": 1.0505,
|
| 311 |
"step": 460
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 3.92,
|
| 315 |
+
"learning_rate": 4.830246913580247e-05,
|
| 316 |
+
"loss": 1.0184,
|
| 317 |
"step": 470
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"epoch": 4.0,
|
| 321 |
+
"learning_rate": 4.814814814814815e-05,
|
| 322 |
+
"loss": 1.0542,
|
| 323 |
"step": 480
|
| 324 |
},
|
| 325 |
{
|
| 326 |
"epoch": 4.0,
|
| 327 |
+
"eval_accuracy": 0.5598591549295775,
|
| 328 |
+
"eval_loss": 1.229072093963623,
|
| 329 |
+
"eval_runtime": 24.2826,
|
| 330 |
+
"eval_samples_per_second": 70.174,
|
| 331 |
+
"eval_steps_per_second": 2.224,
|
| 332 |
"step": 480
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 4.08,
|
| 336 |
+
"learning_rate": 4.799382716049383e-05,
|
| 337 |
+
"loss": 1.0394,
|
| 338 |
"step": 490
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 4.17,
|
| 342 |
+
"learning_rate": 4.783950617283951e-05,
|
| 343 |
+
"loss": 1.0383,
|
| 344 |
"step": 500
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 4.25,
|
| 348 |
+
"learning_rate": 4.768518518518519e-05,
|
| 349 |
+
"loss": 1.0385,
|
| 350 |
"step": 510
|
| 351 |
},
|
| 352 |
{
|
| 353 |
"epoch": 4.33,
|
| 354 |
+
"learning_rate": 4.7530864197530866e-05,
|
| 355 |
+
"loss": 1.0761,
|
| 356 |
"step": 520
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 4.42,
|
| 360 |
+
"learning_rate": 4.7376543209876543e-05,
|
| 361 |
+
"loss": 1.0378,
|
| 362 |
"step": 530
|
| 363 |
},
|
| 364 |
{
|
| 365 |
"epoch": 4.5,
|
| 366 |
+
"learning_rate": 4.722222222222222e-05,
|
| 367 |
+
"loss": 1.0378,
|
| 368 |
"step": 540
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 4.58,
|
| 372 |
+
"learning_rate": 4.70679012345679e-05,
|
| 373 |
+
"loss": 1.0333,
|
| 374 |
"step": 550
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 4.67,
|
| 378 |
+
"learning_rate": 4.691358024691358e-05,
|
| 379 |
+
"loss": 1.0729,
|
| 380 |
"step": 560
|
| 381 |
},
|
| 382 |
{
|
| 383 |
"epoch": 4.75,
|
| 384 |
+
"learning_rate": 4.675925925925926e-05,
|
| 385 |
+
"loss": 1.0342,
|
| 386 |
"step": 570
|
| 387 |
},
|
| 388 |
{
|
| 389 |
"epoch": 4.83,
|
| 390 |
+
"learning_rate": 4.6604938271604945e-05,
|
| 391 |
+
"loss": 1.0316,
|
| 392 |
"step": 580
|
| 393 |
},
|
| 394 |
{
|
| 395 |
"epoch": 4.92,
|
| 396 |
+
"learning_rate": 4.645061728395062e-05,
|
| 397 |
+
"loss": 1.0213,
|
| 398 |
"step": 590
|
| 399 |
},
|
| 400 |
{
|
| 401 |
"epoch": 5.0,
|
| 402 |
+
"learning_rate": 4.62962962962963e-05,
|
| 403 |
+
"loss": 1.028,
|
| 404 |
"step": 600
|
| 405 |
},
|
| 406 |
{
|
| 407 |
"epoch": 5.0,
|
| 408 |
+
"eval_accuracy": 0.5598591549295775,
|
| 409 |
+
"eval_loss": 1.1931029558181763,
|
| 410 |
+
"eval_runtime": 24.0021,
|
| 411 |
+
"eval_samples_per_second": 70.994,
|
| 412 |
+
"eval_steps_per_second": 2.25,
|
| 413 |
"step": 600
|
| 414 |
},
|
| 415 |
{
|
| 416 |
"epoch": 5.08,
|
| 417 |
+
"learning_rate": 4.614197530864198e-05,
|
| 418 |
+
"loss": 1.0657,
|
| 419 |
"step": 610
|
| 420 |
},
|
| 421 |
{
|
| 422 |
"epoch": 5.17,
|
| 423 |
+
"learning_rate": 4.5987654320987656e-05,
|
| 424 |
+
"loss": 1.0504,
|
| 425 |
"step": 620
|
| 426 |
},
|
| 427 |
{
|
| 428 |
"epoch": 5.25,
|
| 429 |
+
"learning_rate": 4.5833333333333334e-05,
|
| 430 |
+
"loss": 1.0345,
|
| 431 |
"step": 630
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 5.33,
|
| 435 |
+
"learning_rate": 4.567901234567901e-05,
|
| 436 |
+
"loss": 1.0113,
|
| 437 |
"step": 640
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 5.42,
|
| 441 |
+
"learning_rate": 4.5524691358024696e-05,
|
| 442 |
+
"loss": 1.0316,
|
| 443 |
"step": 650
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"epoch": 5.5,
|
| 447 |
+
"learning_rate": 4.5370370370370374e-05,
|
| 448 |
+
"loss": 1.0287,
|
| 449 |
"step": 660
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 5.58,
|
| 453 |
+
"learning_rate": 4.521604938271605e-05,
|
| 454 |
+
"loss": 1.064,
|
| 455 |
"step": 670
|
| 456 |
},
|
| 457 |
{
|
| 458 |
"epoch": 5.67,
|
| 459 |
+
"learning_rate": 4.506172839506173e-05,
|
| 460 |
+
"loss": 1.0079,
|
| 461 |
"step": 680
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 5.75,
|
| 465 |
+
"learning_rate": 4.490740740740741e-05,
|
| 466 |
+
"loss": 1.007,
|
| 467 |
"step": 690
|
| 468 |
},
|
| 469 |
{
|
| 470 |
"epoch": 5.83,
|
| 471 |
+
"learning_rate": 4.4753086419753084e-05,
|
| 472 |
+
"loss": 0.9965,
|
| 473 |
"step": 700
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 5.92,
|
| 477 |
+
"learning_rate": 4.459876543209877e-05,
|
| 478 |
+
"loss": 1.035,
|
| 479 |
"step": 710
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 6.0,
|
| 483 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 484 |
+
"loss": 1.0023,
|
| 485 |
"step": 720
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"epoch": 6.0,
|
| 489 |
+
"eval_accuracy": 0.5674882629107981,
|
| 490 |
+
"eval_loss": 1.1548198461532593,
|
| 491 |
+
"eval_runtime": 23.6784,
|
| 492 |
+
"eval_samples_per_second": 71.964,
|
| 493 |
+
"eval_steps_per_second": 2.281,
|
| 494 |
"step": 720
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 6.08,
|
| 498 |
+
"learning_rate": 4.429012345679013e-05,
|
| 499 |
+
"loss": 1.033,
|
| 500 |
"step": 730
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 6.17,
|
| 504 |
+
"learning_rate": 4.413580246913581e-05,
|
| 505 |
+
"loss": 1.0003,
|
| 506 |
"step": 740
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 6.25,
|
| 510 |
+
"learning_rate": 4.3981481481481486e-05,
|
| 511 |
+
"loss": 1.0196,
|
| 512 |
"step": 750
|
| 513 |
},
|
| 514 |
{
|
| 515 |
"epoch": 6.33,
|
| 516 |
+
"learning_rate": 4.3827160493827164e-05,
|
| 517 |
+
"loss": 1.0211,
|
| 518 |
"step": 760
|
| 519 |
},
|
| 520 |
{
|
| 521 |
"epoch": 6.42,
|
| 522 |
+
"learning_rate": 4.367283950617284e-05,
|
| 523 |
+
"loss": 0.9964,
|
| 524 |
"step": 770
|
| 525 |
},
|
| 526 |
{
|
| 527 |
"epoch": 6.5,
|
| 528 |
+
"learning_rate": 4.351851851851852e-05,
|
| 529 |
+
"loss": 1.0339,
|
| 530 |
"step": 780
|
| 531 |
},
|
| 532 |
{
|
| 533 |
"epoch": 6.58,
|
| 534 |
+
"learning_rate": 4.33641975308642e-05,
|
| 535 |
+
"loss": 0.9788,
|
| 536 |
"step": 790
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 6.67,
|
| 540 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 541 |
+
"loss": 1.0567,
|
| 542 |
"step": 800
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 6.75,
|
| 546 |
+
"learning_rate": 4.305555555555556e-05,
|
| 547 |
+
"loss": 0.994,
|
| 548 |
"step": 810
|
| 549 |
},
|
| 550 |
{
|
| 551 |
"epoch": 6.83,
|
| 552 |
+
"learning_rate": 4.290123456790124e-05,
|
| 553 |
+
"loss": 1.0389,
|
| 554 |
"step": 820
|
| 555 |
},
|
| 556 |
{
|
| 557 |
"epoch": 6.92,
|
| 558 |
+
"learning_rate": 4.2746913580246915e-05,
|
| 559 |
+
"loss": 0.9833,
|
| 560 |
"step": 830
|
| 561 |
},
|
| 562 |
{
|
| 563 |
"epoch": 7.0,
|
| 564 |
+
"learning_rate": 4.259259259259259e-05,
|
| 565 |
+
"loss": 1.0176,
|
| 566 |
"step": 840
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 7.0,
|
| 570 |
+
"eval_accuracy": 0.5757042253521126,
|
| 571 |
+
"eval_loss": 1.0932233333587646,
|
| 572 |
+
"eval_runtime": 23.5291,
|
| 573 |
+
"eval_samples_per_second": 72.421,
|
| 574 |
+
"eval_steps_per_second": 2.295,
|
| 575 |
"step": 840
|
| 576 |
},
|
| 577 |
{
|
| 578 |
"epoch": 7.08,
|
| 579 |
+
"learning_rate": 4.243827160493827e-05,
|
| 580 |
+
"loss": 1.021,
|
| 581 |
"step": 850
|
| 582 |
},
|
| 583 |
{
|
| 584 |
"epoch": 7.17,
|
| 585 |
+
"learning_rate": 4.2283950617283955e-05,
|
| 586 |
+
"loss": 0.9636,
|
| 587 |
"step": 860
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 7.25,
|
| 591 |
+
"learning_rate": 4.212962962962963e-05,
|
| 592 |
+
"loss": 0.9965,
|
| 593 |
"step": 870
|
| 594 |
},
|
| 595 |
{
|
| 596 |
"epoch": 7.33,
|
| 597 |
+
"learning_rate": 4.197530864197531e-05,
|
| 598 |
+
"loss": 0.9829,
|
| 599 |
"step": 880
|
| 600 |
},
|
| 601 |
{
|
| 602 |
"epoch": 7.42,
|
| 603 |
+
"learning_rate": 4.1820987654320994e-05,
|
| 604 |
+
"loss": 1.0139,
|
| 605 |
"step": 890
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"epoch": 7.5,
|
| 609 |
+
"learning_rate": 4.166666666666667e-05,
|
| 610 |
+
"loss": 1.0446,
|
| 611 |
"step": 900
|
| 612 |
},
|
| 613 |
{
|
| 614 |
"epoch": 7.58,
|
| 615 |
+
"learning_rate": 4.151234567901235e-05,
|
| 616 |
+
"loss": 0.9917,
|
| 617 |
"step": 910
|
| 618 |
},
|
| 619 |
{
|
| 620 |
"epoch": 7.67,
|
| 621 |
+
"learning_rate": 4.135802469135803e-05,
|
| 622 |
+
"loss": 0.9904,
|
| 623 |
"step": 920
|
| 624 |
},
|
| 625 |
{
|
| 626 |
"epoch": 7.75,
|
| 627 |
+
"learning_rate": 4.1203703703703705e-05,
|
| 628 |
+
"loss": 1.0068,
|
| 629 |
"step": 930
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 7.83,
|
| 633 |
+
"learning_rate": 4.104938271604938e-05,
|
| 634 |
+
"loss": 0.974,
|
| 635 |
"step": 940
|
| 636 |
},
|
| 637 |
{
|
| 638 |
"epoch": 7.92,
|
| 639 |
+
"learning_rate": 4.089506172839506e-05,
|
| 640 |
+
"loss": 1.0291,
|
| 641 |
"step": 950
|
| 642 |
},
|
| 643 |
{
|
| 644 |
"epoch": 8.0,
|
| 645 |
+
"learning_rate": 4.074074074074074e-05,
|
| 646 |
+
"loss": 0.992,
|
| 647 |
"step": 960
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 8.0,
|
| 651 |
+
"eval_accuracy": 0.5751173708920188,
|
| 652 |
+
"eval_loss": 1.1386847496032715,
|
| 653 |
+
"eval_runtime": 23.4444,
|
| 654 |
+
"eval_samples_per_second": 72.683,
|
| 655 |
+
"eval_steps_per_second": 2.303,
|
| 656 |
"step": 960
|
| 657 |
},
|
| 658 |
{
|
| 659 |
"epoch": 8.08,
|
| 660 |
+
"learning_rate": 4.058641975308642e-05,
|
| 661 |
+
"loss": 0.9983,
|
| 662 |
"step": 970
|
| 663 |
},
|
| 664 |
{
|
| 665 |
"epoch": 8.17,
|
| 666 |
+
"learning_rate": 4.04320987654321e-05,
|
| 667 |
+
"loss": 0.964,
|
| 668 |
"step": 980
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 8.25,
|
| 672 |
+
"learning_rate": 4.027777777777778e-05,
|
| 673 |
+
"loss": 0.9514,
|
| 674 |
"step": 990
|
| 675 |
},
|
| 676 |
{
|
| 677 |
"epoch": 8.33,
|
| 678 |
+
"learning_rate": 4.012345679012346e-05,
|
| 679 |
+
"loss": 1.003,
|
| 680 |
"step": 1000
|
| 681 |
},
|
| 682 |
{
|
| 683 |
"epoch": 8.42,
|
| 684 |
+
"learning_rate": 3.996913580246914e-05,
|
| 685 |
+
"loss": 0.9575,
|
| 686 |
"step": 1010
|
| 687 |
},
|
| 688 |
{
|
| 689 |
"epoch": 8.5,
|
| 690 |
+
"learning_rate": 3.981481481481482e-05,
|
| 691 |
+
"loss": 0.9965,
|
| 692 |
"step": 1020
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 8.58,
|
| 696 |
+
"learning_rate": 3.9660493827160496e-05,
|
| 697 |
+
"loss": 1.0445,
|
| 698 |
"step": 1030
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 8.67,
|
| 702 |
+
"learning_rate": 3.950617283950617e-05,
|
| 703 |
+
"loss": 0.9638,
|
| 704 |
"step": 1040
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 8.75,
|
| 708 |
+
"learning_rate": 3.935185185185186e-05,
|
| 709 |
+
"loss": 1.024,
|
| 710 |
"step": 1050
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 8.83,
|
| 714 |
+
"learning_rate": 3.9197530864197535e-05,
|
| 715 |
+
"loss": 1.0393,
|
| 716 |
"step": 1060
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 8.92,
|
| 720 |
+
"learning_rate": 3.904320987654321e-05,
|
| 721 |
+
"loss": 0.9897,
|
| 722 |
"step": 1070
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 9.0,
|
| 726 |
+
"learning_rate": 3.888888888888889e-05,
|
| 727 |
+
"loss": 0.9891,
|
| 728 |
"step": 1080
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 9.0,
|
| 732 |
+
"eval_accuracy": 0.5463615023474179,
|
| 733 |
+
"eval_loss": 1.238718867301941,
|
| 734 |
+
"eval_runtime": 23.4716,
|
| 735 |
+
"eval_samples_per_second": 72.598,
|
| 736 |
+
"eval_steps_per_second": 2.301,
|
| 737 |
"step": 1080
|
| 738 |
},
|
| 739 |
{
|
| 740 |
"epoch": 9.08,
|
| 741 |
+
"learning_rate": 3.873456790123457e-05,
|
| 742 |
+
"loss": 1.0242,
|
| 743 |
"step": 1090
|
| 744 |
},
|
| 745 |
{
|
| 746 |
"epoch": 9.17,
|
| 747 |
+
"learning_rate": 3.8580246913580246e-05,
|
| 748 |
+
"loss": 0.9982,
|
| 749 |
"step": 1100
|
| 750 |
},
|
| 751 |
{
|
| 752 |
"epoch": 9.25,
|
| 753 |
+
"learning_rate": 3.8425925925925924e-05,
|
| 754 |
+
"loss": 0.9985,
|
| 755 |
"step": 1110
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 9.33,
|
| 759 |
+
"learning_rate": 3.82716049382716e-05,
|
| 760 |
+
"loss": 1.0509,
|
| 761 |
"step": 1120
|
| 762 |
},
|
| 763 |
{
|
| 764 |
"epoch": 9.42,
|
| 765 |
+
"learning_rate": 3.8117283950617286e-05,
|
| 766 |
+
"loss": 0.9643,
|
| 767 |
"step": 1130
|
| 768 |
},
|
| 769 |
{
|
| 770 |
"epoch": 9.5,
|
| 771 |
+
"learning_rate": 3.7962962962962964e-05,
|
| 772 |
+
"loss": 0.9595,
|
| 773 |
"step": 1140
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"epoch": 9.58,
|
| 777 |
+
"learning_rate": 3.780864197530865e-05,
|
| 778 |
+
"loss": 1.0199,
|
| 779 |
"step": 1150
|
| 780 |
},
|
| 781 |
{
|
| 782 |
"epoch": 9.67,
|
| 783 |
+
"learning_rate": 3.7654320987654326e-05,
|
| 784 |
+
"loss": 0.948,
|
| 785 |
"step": 1160
|
| 786 |
},
|
| 787 |
{
|
| 788 |
"epoch": 9.75,
|
| 789 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 790 |
+
"loss": 0.9953,
|
| 791 |
"step": 1170
|
| 792 |
},
|
| 793 |
{
|
| 794 |
"epoch": 9.83,
|
| 795 |
+
"learning_rate": 3.734567901234568e-05,
|
| 796 |
+
"loss": 0.9848,
|
| 797 |
"step": 1180
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 9.92,
|
| 801 |
+
"learning_rate": 3.719135802469136e-05,
|
| 802 |
+
"loss": 0.9249,
|
| 803 |
"step": 1190
|
| 804 |
},
|
| 805 |
{
|
| 806 |
"epoch": 10.0,
|
| 807 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 808 |
+
"loss": 0.9635,
|
| 809 |
"step": 1200
|
| 810 |
},
|
| 811 |
{
|
| 812 |
"epoch": 10.0,
|
| 813 |
+
"eval_accuracy": 0.5428403755868545,
|
| 814 |
+
"eval_loss": 1.3771986961364746,
|
| 815 |
+
"eval_runtime": 23.4919,
|
| 816 |
+
"eval_samples_per_second": 72.536,
|
| 817 |
+
"eval_steps_per_second": 2.299,
|
| 818 |
"step": 1200
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 10.08,
|
| 822 |
+
"learning_rate": 3.6882716049382714e-05,
|
| 823 |
+
"loss": 0.9894,
|
| 824 |
"step": 1210
|
| 825 |
},
|
| 826 |
{
|
| 827 |
"epoch": 10.17,
|
| 828 |
+
"learning_rate": 3.67283950617284e-05,
|
| 829 |
+
"loss": 1.0025,
|
| 830 |
"step": 1220
|
| 831 |
},
|
| 832 |
{
|
| 833 |
"epoch": 10.25,
|
| 834 |
+
"learning_rate": 3.6574074074074076e-05,
|
| 835 |
+
"loss": 0.9747,
|
| 836 |
"step": 1230
|
| 837 |
},
|
| 838 |
{
|
| 839 |
"epoch": 10.33,
|
| 840 |
+
"learning_rate": 3.6419753086419754e-05,
|
| 841 |
+
"loss": 0.9813,
|
| 842 |
"step": 1240
|
| 843 |
},
|
| 844 |
{
|
| 845 |
"epoch": 10.42,
|
| 846 |
+
"learning_rate": 3.626543209876543e-05,
|
| 847 |
+
"loss": 1.0371,
|
| 848 |
"step": 1250
|
| 849 |
},
|
| 850 |
{
|
| 851 |
"epoch": 10.5,
|
| 852 |
+
"learning_rate": 3.611111111111111e-05,
|
| 853 |
+
"loss": 0.9673,
|
| 854 |
"step": 1260
|
| 855 |
},
|
| 856 |
{
|
| 857 |
"epoch": 10.58,
|
| 858 |
+
"learning_rate": 3.5956790123456794e-05,
|
| 859 |
+
"loss": 0.9378,
|
| 860 |
"step": 1270
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 10.67,
|
| 864 |
+
"learning_rate": 3.580246913580247e-05,
|
| 865 |
+
"loss": 0.9659,
|
| 866 |
"step": 1280
|
| 867 |
},
|
| 868 |
{
|
| 869 |
"epoch": 10.75,
|
| 870 |
+
"learning_rate": 3.564814814814815e-05,
|
| 871 |
+
"loss": 0.9527,
|
| 872 |
"step": 1290
|
| 873 |
},
|
| 874 |
{
|
| 875 |
"epoch": 10.83,
|
| 876 |
+
"learning_rate": 3.5493827160493834e-05,
|
| 877 |
+
"loss": 0.9805,
|
| 878 |
"step": 1300
|
| 879 |
},
|
| 880 |
{
|
| 881 |
"epoch": 10.92,
|
| 882 |
+
"learning_rate": 3.533950617283951e-05,
|
| 883 |
+
"loss": 0.9974,
|
| 884 |
"step": 1310
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 11.0,
|
| 888 |
+
"learning_rate": 3.518518518518519e-05,
|
| 889 |
+
"loss": 0.9764,
|
| 890 |
"step": 1320
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 11.0,
|
| 894 |
+
"eval_accuracy": 0.5258215962441315,
|
| 895 |
+
"eval_loss": 1.4329094886779785,
|
| 896 |
+
"eval_runtime": 23.2917,
|
| 897 |
+
"eval_samples_per_second": 73.159,
|
| 898 |
+
"eval_steps_per_second": 2.318,
|
| 899 |
"step": 1320
|
| 900 |
},
|
| 901 |
{
|
| 902 |
"epoch": 11.08,
|
| 903 |
+
"learning_rate": 3.503086419753087e-05,
|
| 904 |
+
"loss": 1.0079,
|
| 905 |
"step": 1330
|
| 906 |
},
|
| 907 |
{
|
| 908 |
"epoch": 11.17,
|
| 909 |
+
"learning_rate": 3.4876543209876545e-05,
|
| 910 |
+
"loss": 0.99,
|
| 911 |
"step": 1340
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 11.25,
|
| 915 |
+
"learning_rate": 3.472222222222222e-05,
|
| 916 |
+
"loss": 0.9836,
|
| 917 |
"step": 1350
|
| 918 |
},
|
| 919 |
{
|
| 920 |
"epoch": 11.33,
|
| 921 |
+
"learning_rate": 3.45679012345679e-05,
|
| 922 |
+
"loss": 0.9996,
|
| 923 |
"step": 1360
|
| 924 |
},
|
| 925 |
{
|
| 926 |
"epoch": 11.42,
|
| 927 |
+
"learning_rate": 3.441358024691358e-05,
|
| 928 |
+
"loss": 0.9794,
|
| 929 |
"step": 1370
|
| 930 |
},
|
| 931 |
{
|
| 932 |
"epoch": 11.5,
|
| 933 |
+
"learning_rate": 3.425925925925926e-05,
|
| 934 |
+
"loss": 0.9639,
|
| 935 |
"step": 1380
|
| 936 |
},
|
| 937 |
{
|
| 938 |
"epoch": 11.58,
|
| 939 |
+
"learning_rate": 3.410493827160494e-05,
|
| 940 |
+
"loss": 0.9945,
|
| 941 |
"step": 1390
|
| 942 |
},
|
| 943 |
{
|
| 944 |
"epoch": 11.67,
|
| 945 |
+
"learning_rate": 3.395061728395062e-05,
|
| 946 |
+
"loss": 0.9832,
|
| 947 |
"step": 1400
|
| 948 |
},
|
| 949 |
{
|
| 950 |
"epoch": 11.75,
|
| 951 |
+
"learning_rate": 3.3796296296296295e-05,
|
| 952 |
+
"loss": 0.9606,
|
| 953 |
"step": 1410
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 11.83,
|
| 957 |
+
"learning_rate": 3.364197530864198e-05,
|
| 958 |
+
"loss": 0.9581,
|
| 959 |
"step": 1420
|
| 960 |
},
|
| 961 |
{
|
| 962 |
"epoch": 11.92,
|
| 963 |
+
"learning_rate": 3.348765432098766e-05,
|
| 964 |
+
"loss": 0.9784,
|
| 965 |
"step": 1430
|
| 966 |
},
|
| 967 |
{
|
| 968 |
"epoch": 12.0,
|
| 969 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 970 |
+
"loss": 0.9375,
|
| 971 |
"step": 1440
|
| 972 |
},
|
| 973 |
{
|
| 974 |
"epoch": 12.0,
|
| 975 |
+
"eval_accuracy": 0.5522300469483568,
|
| 976 |
+
"eval_loss": 1.2830231189727783,
|
| 977 |
+
"eval_runtime": 23.496,
|
| 978 |
+
"eval_samples_per_second": 72.523,
|
| 979 |
+
"eval_steps_per_second": 2.298,
|
| 980 |
"step": 1440
|
| 981 |
},
|
| 982 |
{
|
| 983 |
"epoch": 12.08,
|
| 984 |
+
"learning_rate": 3.317901234567901e-05,
|
| 985 |
+
"loss": 0.9912,
|
| 986 |
"step": 1450
|
| 987 |
},
|
| 988 |
{
|
| 989 |
"epoch": 12.17,
|
| 990 |
+
"learning_rate": 3.30246913580247e-05,
|
| 991 |
+
"loss": 0.9722,
|
| 992 |
"step": 1460
|
| 993 |
},
|
| 994 |
{
|
| 995 |
"epoch": 12.25,
|
| 996 |
+
"learning_rate": 3.2870370370370375e-05,
|
| 997 |
+
"loss": 1.0089,
|
| 998 |
"step": 1470
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
"epoch": 12.33,
|
| 1002 |
+
"learning_rate": 3.271604938271605e-05,
|
| 1003 |
+
"loss": 0.954,
|
| 1004 |
"step": 1480
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 12.42,
|
| 1008 |
+
"learning_rate": 3.256172839506173e-05,
|
| 1009 |
+
"loss": 0.9823,
|
| 1010 |
"step": 1490
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
"epoch": 12.5,
|
| 1014 |
+
"learning_rate": 3.240740740740741e-05,
|
| 1015 |
+
"loss": 0.9813,
|
| 1016 |
"step": 1500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 12.58,
|
| 1020 |
+
"learning_rate": 3.2253086419753086e-05,
|
| 1021 |
+
"loss": 0.9827,
|
| 1022 |
"step": 1510
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
"epoch": 12.67,
|
| 1026 |
+
"learning_rate": 3.209876543209876e-05,
|
| 1027 |
+
"loss": 0.9573,
|
| 1028 |
"step": 1520
|
| 1029 |
},
|
| 1030 |
{
|
| 1031 |
"epoch": 12.75,
|
| 1032 |
+
"learning_rate": 3.194444444444444e-05,
|
| 1033 |
+
"loss": 1.002,
|
| 1034 |
"step": 1530
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
"epoch": 12.83,
|
| 1038 |
+
"learning_rate": 3.1790123456790125e-05,
|
| 1039 |
+
"loss": 0.9474,
|
| 1040 |
"step": 1540
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
"epoch": 12.92,
|
| 1044 |
+
"learning_rate": 3.16358024691358e-05,
|
| 1045 |
+
"loss": 0.9435,
|
| 1046 |
"step": 1550
|
| 1047 |
},
|
| 1048 |
{
|
| 1049 |
"epoch": 13.0,
|
| 1050 |
+
"learning_rate": 3.148148148148148e-05,
|
| 1051 |
+
"loss": 0.9574,
|
| 1052 |
"step": 1560
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
"epoch": 13.0,
|
| 1056 |
+
"eval_accuracy": 0.522887323943662,
|
| 1057 |
+
"eval_loss": 1.400294542312622,
|
| 1058 |
+
"eval_runtime": 23.5276,
|
| 1059 |
+
"eval_samples_per_second": 72.426,
|
| 1060 |
+
"eval_steps_per_second": 2.295,
|
| 1061 |
"step": 1560
|
| 1062 |
},
|
| 1063 |
{
|
| 1064 |
"epoch": 13.08,
|
| 1065 |
+
"learning_rate": 3.1327160493827165e-05,
|
| 1066 |
+
"loss": 0.9727,
|
| 1067 |
"step": 1570
|
| 1068 |
},
|
| 1069 |
{
|
| 1070 |
"epoch": 13.17,
|
| 1071 |
+
"learning_rate": 3.117283950617284e-05,
|
| 1072 |
+
"loss": 0.9709,
|
| 1073 |
"step": 1580
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
"epoch": 13.25,
|
| 1077 |
+
"learning_rate": 3.101851851851852e-05,
|
| 1078 |
+
"loss": 0.9753,
|
| 1079 |
"step": 1590
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 13.33,
|
| 1083 |
+
"learning_rate": 3.08641975308642e-05,
|
| 1084 |
+
"loss": 0.9267,
|
| 1085 |
"step": 1600
|
| 1086 |
},
|
| 1087 |
{
|
| 1088 |
"epoch": 13.42,
|
| 1089 |
+
"learning_rate": 3.0709876543209876e-05,
|
| 1090 |
+
"loss": 0.9905,
|
| 1091 |
"step": 1610
|
| 1092 |
},
|
| 1093 |
{
|
| 1094 |
"epoch": 13.5,
|
| 1095 |
+
"learning_rate": 3.055555555555556e-05,
|
| 1096 |
+
"loss": 0.9499,
|
| 1097 |
"step": 1620
|
| 1098 |
},
|
| 1099 |
{
|
| 1100 |
"epoch": 13.58,
|
| 1101 |
+
"learning_rate": 3.0401234567901238e-05,
|
| 1102 |
+
"loss": 1.0304,
|
| 1103 |
"step": 1630
|
| 1104 |
},
|
| 1105 |
{
|
| 1106 |
"epoch": 13.67,
|
| 1107 |
+
"learning_rate": 3.0246913580246916e-05,
|
| 1108 |
+
"loss": 1.0021,
|
| 1109 |
"step": 1640
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
"epoch": 13.75,
|
| 1113 |
+
"learning_rate": 3.0092592592592593e-05,
|
| 1114 |
+
"loss": 0.9771,
|
| 1115 |
"step": 1650
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
"epoch": 13.83,
|
| 1119 |
+
"learning_rate": 2.993827160493827e-05,
|
| 1120 |
+
"loss": 0.9278,
|
| 1121 |
"step": 1660
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 13.92,
|
| 1125 |
+
"learning_rate": 2.9783950617283952e-05,
|
| 1126 |
+
"loss": 0.9555,
|
| 1127 |
"step": 1670
|
| 1128 |
},
|
| 1129 |
{
|
| 1130 |
"epoch": 14.0,
|
| 1131 |
+
"learning_rate": 2.962962962962963e-05,
|
| 1132 |
+
"loss": 0.9907,
|
| 1133 |
"step": 1680
|
| 1134 |
},
|
| 1135 |
{
|
| 1136 |
"epoch": 14.0,
|
| 1137 |
+
"eval_accuracy": 0.5422535211267606,
|
| 1138 |
+
"eval_loss": 1.344726324081421,
|
| 1139 |
+
"eval_runtime": 23.9181,
|
| 1140 |
+
"eval_samples_per_second": 71.243,
|
| 1141 |
+
"eval_steps_per_second": 2.258,
|
| 1142 |
"step": 1680
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 14.08,
|
| 1146 |
+
"learning_rate": 2.9475308641975308e-05,
|
| 1147 |
+
"loss": 0.9437,
|
| 1148 |
"step": 1690
|
| 1149 |
},
|
| 1150 |
{
|
| 1151 |
"epoch": 14.17,
|
| 1152 |
+
"learning_rate": 2.9320987654320992e-05,
|
| 1153 |
+
"loss": 0.9556,
|
| 1154 |
"step": 1700
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
"epoch": 14.25,
|
| 1158 |
+
"learning_rate": 2.916666666666667e-05,
|
| 1159 |
+
"loss": 0.9587,
|
| 1160 |
"step": 1710
|
| 1161 |
},
|
| 1162 |
{
|
| 1163 |
"epoch": 14.33,
|
| 1164 |
+
"learning_rate": 2.9012345679012347e-05,
|
| 1165 |
+
"loss": 0.9754,
|
| 1166 |
"step": 1720
|
| 1167 |
},
|
| 1168 |
{
|
| 1169 |
"epoch": 14.42,
|
| 1170 |
+
"learning_rate": 2.8858024691358025e-05,
|
| 1171 |
+
"loss": 0.9724,
|
| 1172 |
"step": 1730
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 14.5,
|
| 1176 |
+
"learning_rate": 2.8703703703703706e-05,
|
| 1177 |
+
"loss": 0.9733,
|
| 1178 |
"step": 1740
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
"epoch": 14.58,
|
| 1182 |
+
"learning_rate": 2.8549382716049384e-05,
|
| 1183 |
+
"loss": 0.9725,
|
| 1184 |
"step": 1750
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
"epoch": 14.67,
|
| 1188 |
+
"learning_rate": 2.839506172839506e-05,
|
| 1189 |
+
"loss": 0.9776,
|
| 1190 |
"step": 1760
|
| 1191 |
},
|
| 1192 |
{
|
| 1193 |
"epoch": 14.75,
|
| 1194 |
+
"learning_rate": 2.824074074074074e-05,
|
| 1195 |
+
"loss": 0.95,
|
| 1196 |
"step": 1770
|
| 1197 |
},
|
| 1198 |
{
|
| 1199 |
"epoch": 14.83,
|
| 1200 |
+
"learning_rate": 2.8086419753086424e-05,
|
| 1201 |
+
"loss": 1.006,
|
| 1202 |
"step": 1780
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
"epoch": 14.92,
|
| 1206 |
+
"learning_rate": 2.79320987654321e-05,
|
| 1207 |
+
"loss": 0.9395,
|
| 1208 |
"step": 1790
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
"epoch": 15.0,
|
| 1212 |
+
"learning_rate": 2.777777777777778e-05,
|
| 1213 |
+
"loss": 0.9507,
|
| 1214 |
"step": 1800
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 15.0,
|
| 1218 |
+
"eval_accuracy": 0.5604460093896714,
|
| 1219 |
+
"eval_loss": 1.290692925453186,
|
| 1220 |
+
"eval_runtime": 23.8671,
|
| 1221 |
+
"eval_samples_per_second": 71.395,
|
| 1222 |
+
"eval_steps_per_second": 2.263,
|
| 1223 |
"step": 1800
|
| 1224 |
},
|
| 1225 |
{
|
| 1226 |
"epoch": 15.08,
|
| 1227 |
+
"learning_rate": 2.762345679012346e-05,
|
| 1228 |
+
"loss": 0.9754,
|
| 1229 |
"step": 1810
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
"epoch": 15.17,
|
| 1233 |
+
"learning_rate": 2.7469135802469138e-05,
|
| 1234 |
+
"loss": 0.9558,
|
| 1235 |
"step": 1820
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 15.25,
|
| 1239 |
+
"learning_rate": 2.7314814814814816e-05,
|
| 1240 |
+
"loss": 0.9572,
|
| 1241 |
"step": 1830
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
"epoch": 15.33,
|
| 1245 |
+
"learning_rate": 2.7160493827160493e-05,
|
| 1246 |
+
"loss": 0.9987,
|
| 1247 |
"step": 1840
|
| 1248 |
},
|
| 1249 |
{
|
| 1250 |
"epoch": 15.42,
|
| 1251 |
+
"learning_rate": 2.700617283950617e-05,
|
| 1252 |
+
"loss": 0.9429,
|
| 1253 |
"step": 1850
|
| 1254 |
},
|
| 1255 |
{
|
| 1256 |
"epoch": 15.5,
|
| 1257 |
+
"learning_rate": 2.6851851851851855e-05,
|
| 1258 |
+
"loss": 0.9161,
|
| 1259 |
"step": 1860
|
| 1260 |
},
|
| 1261 |
{
|
| 1262 |
"epoch": 15.58,
|
| 1263 |
+
"learning_rate": 2.6697530864197533e-05,
|
| 1264 |
+
"loss": 0.9153,
|
| 1265 |
"step": 1870
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
"epoch": 15.67,
|
| 1269 |
+
"learning_rate": 2.654320987654321e-05,
|
| 1270 |
+
"loss": 1.0111,
|
| 1271 |
"step": 1880
|
| 1272 |
},
|
| 1273 |
{
|
| 1274 |
"epoch": 15.75,
|
| 1275 |
+
"learning_rate": 2.6388888888888892e-05,
|
| 1276 |
+
"loss": 0.9889,
|
| 1277 |
"step": 1890
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 15.83,
|
| 1281 |
+
"learning_rate": 2.623456790123457e-05,
|
| 1282 |
+
"loss": 0.9616,
|
| 1283 |
"step": 1900
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
"epoch": 15.92,
|
| 1287 |
+
"learning_rate": 2.6080246913580247e-05,
|
| 1288 |
+
"loss": 0.9414,
|
| 1289 |
"step": 1910
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
"epoch": 16.0,
|
| 1293 |
+
"learning_rate": 2.5925925925925925e-05,
|
| 1294 |
+
"loss": 0.9866,
|
| 1295 |
"step": 1920
|
| 1296 |
},
|
| 1297 |
{
|
| 1298 |
"epoch": 16.0,
|
| 1299 |
+
"eval_accuracy": 0.5393192488262911,
|
| 1300 |
+
"eval_loss": 1.4577730894088745,
|
| 1301 |
+
"eval_runtime": 23.9597,
|
| 1302 |
+
"eval_samples_per_second": 71.119,
|
| 1303 |
+
"eval_steps_per_second": 2.254,
|
| 1304 |
"step": 1920
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
"epoch": 16.08,
|
| 1308 |
+
"learning_rate": 2.5771604938271603e-05,
|
| 1309 |
+
"loss": 0.9638,
|
| 1310 |
"step": 1930
|
| 1311 |
},
|
| 1312 |
{
|
| 1313 |
"epoch": 16.17,
|
| 1314 |
+
"learning_rate": 2.5617283950617287e-05,
|
| 1315 |
+
"loss": 0.9562,
|
| 1316 |
"step": 1940
|
| 1317 |
},
|
| 1318 |
{
|
| 1319 |
"epoch": 16.25,
|
| 1320 |
+
"learning_rate": 2.5462962962962965e-05,
|
| 1321 |
+
"loss": 0.9464,
|
| 1322 |
"step": 1950
|
| 1323 |
},
|
| 1324 |
{
|
| 1325 |
"epoch": 16.33,
|
| 1326 |
+
"learning_rate": 2.5308641975308646e-05,
|
| 1327 |
+
"loss": 0.9533,
|
| 1328 |
"step": 1960
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
"epoch": 16.42,
|
| 1332 |
+
"learning_rate": 2.5154320987654324e-05,
|
| 1333 |
+
"loss": 0.9763,
|
| 1334 |
"step": 1970
|
| 1335 |
},
|
| 1336 |
{
|
| 1337 |
"epoch": 16.5,
|
| 1338 |
+
"learning_rate": 2.5e-05,
|
| 1339 |
+
"loss": 0.9795,
|
| 1340 |
"step": 1980
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 16.58,
|
| 1344 |
+
"learning_rate": 2.484567901234568e-05,
|
| 1345 |
+
"loss": 0.9827,
|
| 1346 |
"step": 1990
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
"epoch": 16.67,
|
| 1350 |
+
"learning_rate": 2.4691358024691357e-05,
|
| 1351 |
+
"loss": 0.9269,
|
| 1352 |
"step": 2000
|
| 1353 |
},
|
| 1354 |
{
|
| 1355 |
"epoch": 16.75,
|
| 1356 |
+
"learning_rate": 2.4537037037037038e-05,
|
| 1357 |
+
"loss": 0.9397,
|
| 1358 |
"step": 2010
|
| 1359 |
},
|
| 1360 |
{
|
| 1361 |
"epoch": 16.83,
|
| 1362 |
+
"learning_rate": 2.438271604938272e-05,
|
| 1363 |
+
"loss": 0.9639,
|
| 1364 |
"step": 2020
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
"epoch": 16.92,
|
| 1368 |
+
"learning_rate": 2.4228395061728396e-05,
|
| 1369 |
+
"loss": 0.9626,
|
| 1370 |
"step": 2030
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
"epoch": 17.0,
|
| 1374 |
+
"learning_rate": 2.4074074074074074e-05,
|
| 1375 |
+
"loss": 0.9297,
|
| 1376 |
"step": 2040
|
| 1377 |
},
|
| 1378 |
{
|
| 1379 |
"epoch": 17.0,
|
| 1380 |
+
"eval_accuracy": 0.528169014084507,
|
| 1381 |
+
"eval_loss": 1.477900743484497,
|
| 1382 |
+
"eval_runtime": 23.9974,
|
| 1383 |
+
"eval_samples_per_second": 71.008,
|
| 1384 |
+
"eval_steps_per_second": 2.25,
|
| 1385 |
"step": 2040
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
"epoch": 17.08,
|
| 1389 |
+
"learning_rate": 2.3919753086419755e-05,
|
| 1390 |
+
"loss": 0.9454,
|
| 1391 |
"step": 2050
|
| 1392 |
},
|
| 1393 |
{
|
| 1394 |
"epoch": 17.17,
|
| 1395 |
+
"learning_rate": 2.3765432098765433e-05,
|
| 1396 |
+
"loss": 0.9679,
|
| 1397 |
"step": 2060
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
"epoch": 17.25,
|
| 1401 |
+
"learning_rate": 2.361111111111111e-05,
|
| 1402 |
+
"loss": 0.9748,
|
| 1403 |
"step": 2070
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 17.33,
|
| 1407 |
+
"learning_rate": 2.345679012345679e-05,
|
| 1408 |
+
"loss": 0.9272,
|
| 1409 |
"step": 2080
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
"epoch": 17.42,
|
| 1413 |
+
"learning_rate": 2.3302469135802473e-05,
|
| 1414 |
+
"loss": 0.9831,
|
| 1415 |
"step": 2090
|
| 1416 |
},
|
| 1417 |
{
|
| 1418 |
"epoch": 17.5,
|
| 1419 |
+
"learning_rate": 2.314814814814815e-05,
|
| 1420 |
+
"loss": 0.9593,
|
| 1421 |
"step": 2100
|
| 1422 |
},
|
| 1423 |
{
|
| 1424 |
"epoch": 17.58,
|
| 1425 |
+
"learning_rate": 2.2993827160493828e-05,
|
| 1426 |
+
"loss": 0.9239,
|
| 1427 |
"step": 2110
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
"epoch": 17.67,
|
| 1431 |
+
"learning_rate": 2.2839506172839506e-05,
|
| 1432 |
+
"loss": 0.9255,
|
| 1433 |
"step": 2120
|
| 1434 |
},
|
| 1435 |
{
|
| 1436 |
"epoch": 17.75,
|
| 1437 |
+
"learning_rate": 2.2685185185185187e-05,
|
| 1438 |
+
"loss": 0.9606,
|
| 1439 |
"step": 2130
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
"epoch": 17.83,
|
| 1443 |
+
"learning_rate": 2.2530864197530865e-05,
|
| 1444 |
+
"loss": 0.9407,
|
| 1445 |
"step": 2140
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 17.92,
|
| 1449 |
+
"learning_rate": 2.2376543209876542e-05,
|
| 1450 |
+
"loss": 0.9578,
|
| 1451 |
"step": 2150
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
"epoch": 18.0,
|
| 1455 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 1456 |
+
"loss": 0.9385,
|
| 1457 |
"step": 2160
|
| 1458 |
},
|
| 1459 |
{
|
| 1460 |
"epoch": 18.0,
|
| 1461 |
+
"eval_accuracy": 0.5469483568075117,
|
| 1462 |
+
"eval_loss": 1.3874293565750122,
|
| 1463 |
+
"eval_runtime": 24.2732,
|
| 1464 |
+
"eval_samples_per_second": 70.201,
|
| 1465 |
+
"eval_steps_per_second": 2.225,
|
| 1466 |
"step": 2160
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
"epoch": 18.08,
|
| 1470 |
+
"learning_rate": 2.2067901234567904e-05,
|
| 1471 |
+
"loss": 0.9642,
|
| 1472 |
"step": 2170
|
| 1473 |
},
|
| 1474 |
{
|
| 1475 |
"epoch": 18.17,
|
| 1476 |
+
"learning_rate": 2.1913580246913582e-05,
|
| 1477 |
+
"loss": 0.9263,
|
| 1478 |
"step": 2180
|
| 1479 |
},
|
| 1480 |
{
|
| 1481 |
"epoch": 18.25,
|
| 1482 |
+
"learning_rate": 2.175925925925926e-05,
|
| 1483 |
+
"loss": 0.9209,
|
| 1484 |
"step": 2190
|
| 1485 |
},
|
| 1486 |
{
|
| 1487 |
"epoch": 18.33,
|
| 1488 |
+
"learning_rate": 2.1604938271604937e-05,
|
| 1489 |
+
"loss": 0.9415,
|
| 1490 |
"step": 2200
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
"epoch": 18.42,
|
| 1494 |
+
"learning_rate": 2.145061728395062e-05,
|
| 1495 |
+
"loss": 1.0009,
|
| 1496 |
"step": 2210
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
"epoch": 18.5,
|
| 1500 |
+
"learning_rate": 2.1296296296296296e-05,
|
| 1501 |
+
"loss": 0.9318,
|
| 1502 |
"step": 2220
|
| 1503 |
},
|
| 1504 |
{
|
| 1505 |
"epoch": 18.58,
|
| 1506 |
+
"learning_rate": 2.1141975308641977e-05,
|
| 1507 |
+
"loss": 0.9256,
|
| 1508 |
"step": 2230
|
| 1509 |
},
|
| 1510 |
{
|
| 1511 |
"epoch": 18.67,
|
| 1512 |
+
"learning_rate": 2.0987654320987655e-05,
|
| 1513 |
+
"loss": 0.9422,
|
| 1514 |
"step": 2240
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
"epoch": 18.75,
|
| 1518 |
+
"learning_rate": 2.0833333333333336e-05,
|
| 1519 |
+
"loss": 0.9357,
|
| 1520 |
"step": 2250
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
"epoch": 18.83,
|
| 1524 |
+
"learning_rate": 2.0679012345679014e-05,
|
| 1525 |
+
"loss": 0.9546,
|
| 1526 |
"step": 2260
|
| 1527 |
},
|
| 1528 |
{
|
| 1529 |
"epoch": 18.92,
|
| 1530 |
+
"learning_rate": 2.052469135802469e-05,
|
| 1531 |
+
"loss": 0.9144,
|
| 1532 |
"step": 2270
|
| 1533 |
},
|
| 1534 |
{
|
| 1535 |
"epoch": 19.0,
|
| 1536 |
+
"learning_rate": 2.037037037037037e-05,
|
| 1537 |
+
"loss": 0.9951,
|
| 1538 |
"step": 2280
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
"epoch": 19.0,
|
| 1542 |
+
"eval_accuracy": 0.5586854460093896,
|
| 1543 |
+
"eval_loss": 1.2975717782974243,
|
| 1544 |
+
"eval_runtime": 23.895,
|
| 1545 |
+
"eval_samples_per_second": 71.312,
|
| 1546 |
+
"eval_steps_per_second": 2.26,
|
| 1547 |
"step": 2280
|
| 1548 |
},
|
| 1549 |
{
|
| 1550 |
"epoch": 19.08,
|
| 1551 |
+
"learning_rate": 2.021604938271605e-05,
|
| 1552 |
+
"loss": 0.9519,
|
| 1553 |
"step": 2290
|
| 1554 |
},
|
| 1555 |
{
|
| 1556 |
"epoch": 19.17,
|
| 1557 |
+
"learning_rate": 2.006172839506173e-05,
|
| 1558 |
+
"loss": 0.9362,
|
| 1559 |
"step": 2300
|
| 1560 |
},
|
| 1561 |
{
|
| 1562 |
"epoch": 19.25,
|
| 1563 |
+
"learning_rate": 1.990740740740741e-05,
|
| 1564 |
+
"loss": 0.9226,
|
| 1565 |
"step": 2310
|
| 1566 |
},
|
| 1567 |
{
|
| 1568 |
"epoch": 19.33,
|
| 1569 |
+
"learning_rate": 1.9753086419753087e-05,
|
| 1570 |
+
"loss": 0.9176,
|
| 1571 |
"step": 2320
|
| 1572 |
},
|
| 1573 |
{
|
| 1574 |
"epoch": 19.42,
|
| 1575 |
+
"learning_rate": 1.9598765432098768e-05,
|
| 1576 |
+
"loss": 0.9464,
|
| 1577 |
"step": 2330
|
| 1578 |
},
|
| 1579 |
{
|
| 1580 |
"epoch": 19.5,
|
| 1581 |
+
"learning_rate": 1.9444444444444445e-05,
|
| 1582 |
+
"loss": 0.9369,
|
| 1583 |
"step": 2340
|
| 1584 |
},
|
| 1585 |
{
|
| 1586 |
"epoch": 19.58,
|
| 1587 |
+
"learning_rate": 1.9290123456790123e-05,
|
| 1588 |
+
"loss": 0.9732,
|
| 1589 |
"step": 2350
|
| 1590 |
},
|
| 1591 |
{
|
| 1592 |
"epoch": 19.67,
|
| 1593 |
+
"learning_rate": 1.91358024691358e-05,
|
| 1594 |
+
"loss": 0.9639,
|
| 1595 |
"step": 2360
|
| 1596 |
},
|
| 1597 |
{
|
| 1598 |
"epoch": 19.75,
|
| 1599 |
+
"learning_rate": 1.8981481481481482e-05,
|
| 1600 |
+
"loss": 0.9821,
|
| 1601 |
"step": 2370
|
| 1602 |
},
|
| 1603 |
{
|
| 1604 |
"epoch": 19.83,
|
| 1605 |
+
"learning_rate": 1.8827160493827163e-05,
|
| 1606 |
+
"loss": 0.9472,
|
| 1607 |
"step": 2380
|
| 1608 |
},
|
| 1609 |
{
|
| 1610 |
"epoch": 19.92,
|
| 1611 |
+
"learning_rate": 1.867283950617284e-05,
|
| 1612 |
+
"loss": 0.9674,
|
| 1613 |
"step": 2390
|
| 1614 |
},
|
| 1615 |
{
|
| 1616 |
"epoch": 20.0,
|
| 1617 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 1618 |
+
"loss": 0.9794,
|
| 1619 |
"step": 2400
|
| 1620 |
},
|
| 1621 |
{
|
| 1622 |
"epoch": 20.0,
|
| 1623 |
+
"eval_accuracy": 0.556924882629108,
|
| 1624 |
+
"eval_loss": 1.3109557628631592,
|
| 1625 |
+
"eval_runtime": 23.3371,
|
| 1626 |
+
"eval_samples_per_second": 73.017,
|
| 1627 |
+
"eval_steps_per_second": 2.314,
|
| 1628 |
"step": 2400
|
| 1629 |
},
|
| 1630 |
{
|
| 1631 |
+
"epoch": 20.08,
|
| 1632 |
+
"learning_rate": 1.83641975308642e-05,
|
| 1633 |
+
"loss": 0.959,
|
| 1634 |
+
"step": 2410
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 20.17,
|
| 1638 |
+
"learning_rate": 1.8209876543209877e-05,
|
| 1639 |
+
"loss": 0.9208,
|
| 1640 |
+
"step": 2420
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 20.25,
|
| 1644 |
+
"learning_rate": 1.8055555555555555e-05,
|
| 1645 |
+
"loss": 0.8842,
|
| 1646 |
+
"step": 2430
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 20.33,
|
| 1650 |
+
"learning_rate": 1.7901234567901236e-05,
|
| 1651 |
+
"loss": 0.9423,
|
| 1652 |
+
"step": 2440
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 20.42,
|
| 1656 |
+
"learning_rate": 1.7746913580246917e-05,
|
| 1657 |
+
"loss": 0.909,
|
| 1658 |
+
"step": 2450
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 20.5,
|
| 1662 |
+
"learning_rate": 1.7592592592592595e-05,
|
| 1663 |
+
"loss": 0.93,
|
| 1664 |
+
"step": 2460
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 20.58,
|
| 1668 |
+
"learning_rate": 1.7438271604938272e-05,
|
| 1669 |
+
"loss": 0.94,
|
| 1670 |
+
"step": 2470
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 20.67,
|
| 1674 |
+
"learning_rate": 1.728395061728395e-05,
|
| 1675 |
+
"loss": 0.9199,
|
| 1676 |
+
"step": 2480
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 20.75,
|
| 1680 |
+
"learning_rate": 1.712962962962963e-05,
|
| 1681 |
+
"loss": 0.9436,
|
| 1682 |
+
"step": 2490
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 20.83,
|
| 1686 |
+
"learning_rate": 1.697530864197531e-05,
|
| 1687 |
+
"loss": 0.9872,
|
| 1688 |
+
"step": 2500
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 20.92,
|
| 1692 |
+
"learning_rate": 1.682098765432099e-05,
|
| 1693 |
+
"loss": 0.9622,
|
| 1694 |
+
"step": 2510
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 21.0,
|
| 1698 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 1699 |
+
"loss": 0.9974,
|
| 1700 |
+
"step": 2520
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 21.0,
|
| 1704 |
+
"eval_accuracy": 0.5275821596244131,
|
| 1705 |
+
"eval_loss": 1.3648896217346191,
|
| 1706 |
+
"eval_runtime": 23.4473,
|
| 1707 |
+
"eval_samples_per_second": 72.674,
|
| 1708 |
+
"eval_steps_per_second": 2.303,
|
| 1709 |
+
"step": 2520
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 21.08,
|
| 1713 |
+
"learning_rate": 1.651234567901235e-05,
|
| 1714 |
+
"loss": 0.9337,
|
| 1715 |
+
"step": 2530
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 21.17,
|
| 1719 |
+
"learning_rate": 1.6358024691358026e-05,
|
| 1720 |
+
"loss": 0.9923,
|
| 1721 |
+
"step": 2540
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 21.25,
|
| 1725 |
+
"learning_rate": 1.6203703703703704e-05,
|
| 1726 |
+
"loss": 0.9443,
|
| 1727 |
+
"step": 2550
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 21.33,
|
| 1731 |
+
"learning_rate": 1.604938271604938e-05,
|
| 1732 |
+
"loss": 0.9586,
|
| 1733 |
+
"step": 2560
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 21.42,
|
| 1737 |
+
"learning_rate": 1.5895061728395063e-05,
|
| 1738 |
+
"loss": 0.9377,
|
| 1739 |
+
"step": 2570
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 21.5,
|
| 1743 |
+
"learning_rate": 1.574074074074074e-05,
|
| 1744 |
+
"loss": 0.9484,
|
| 1745 |
+
"step": 2580
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 21.58,
|
| 1749 |
+
"learning_rate": 1.558641975308642e-05,
|
| 1750 |
+
"loss": 0.926,
|
| 1751 |
+
"step": 2590
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 21.67,
|
| 1755 |
+
"learning_rate": 1.54320987654321e-05,
|
| 1756 |
+
"loss": 0.9426,
|
| 1757 |
+
"step": 2600
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 21.75,
|
| 1761 |
+
"learning_rate": 1.527777777777778e-05,
|
| 1762 |
+
"loss": 0.9252,
|
| 1763 |
+
"step": 2610
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 21.83,
|
| 1767 |
+
"learning_rate": 1.5123456790123458e-05,
|
| 1768 |
+
"loss": 0.9228,
|
| 1769 |
+
"step": 2620
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 21.92,
|
| 1773 |
+
"learning_rate": 1.4969135802469136e-05,
|
| 1774 |
+
"loss": 0.9403,
|
| 1775 |
+
"step": 2630
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 22.0,
|
| 1779 |
+
"learning_rate": 1.4814814814814815e-05,
|
| 1780 |
+
"loss": 0.9284,
|
| 1781 |
+
"step": 2640
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 22.0,
|
| 1785 |
+
"eval_accuracy": 0.5363849765258216,
|
| 1786 |
+
"eval_loss": 1.3712799549102783,
|
| 1787 |
+
"eval_runtime": 23.3686,
|
| 1788 |
+
"eval_samples_per_second": 72.918,
|
| 1789 |
+
"eval_steps_per_second": 2.311,
|
| 1790 |
+
"step": 2640
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 22.08,
|
| 1794 |
+
"learning_rate": 1.4660493827160496e-05,
|
| 1795 |
+
"loss": 0.9516,
|
| 1796 |
+
"step": 2650
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 22.17,
|
| 1800 |
+
"learning_rate": 1.4506172839506174e-05,
|
| 1801 |
+
"loss": 0.932,
|
| 1802 |
+
"step": 2660
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 22.25,
|
| 1806 |
+
"learning_rate": 1.4351851851851853e-05,
|
| 1807 |
+
"loss": 0.939,
|
| 1808 |
+
"step": 2670
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 22.33,
|
| 1812 |
+
"learning_rate": 1.419753086419753e-05,
|
| 1813 |
+
"loss": 0.9343,
|
| 1814 |
+
"step": 2680
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 22.42,
|
| 1818 |
+
"learning_rate": 1.4043209876543212e-05,
|
| 1819 |
+
"loss": 0.968,
|
| 1820 |
+
"step": 2690
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 22.5,
|
| 1824 |
+
"learning_rate": 1.388888888888889e-05,
|
| 1825 |
+
"loss": 0.9169,
|
| 1826 |
+
"step": 2700
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 22.58,
|
| 1830 |
+
"learning_rate": 1.3734567901234569e-05,
|
| 1831 |
+
"loss": 0.9549,
|
| 1832 |
+
"step": 2710
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 22.67,
|
| 1836 |
+
"learning_rate": 1.3580246913580247e-05,
|
| 1837 |
+
"loss": 0.9106,
|
| 1838 |
+
"step": 2720
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 22.75,
|
| 1842 |
+
"learning_rate": 1.3425925925925928e-05,
|
| 1843 |
+
"loss": 0.9678,
|
| 1844 |
+
"step": 2730
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 22.83,
|
| 1848 |
+
"learning_rate": 1.3271604938271605e-05,
|
| 1849 |
+
"loss": 0.9296,
|
| 1850 |
+
"step": 2740
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 22.92,
|
| 1854 |
+
"learning_rate": 1.3117283950617285e-05,
|
| 1855 |
+
"loss": 0.9827,
|
| 1856 |
+
"step": 2750
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 23.0,
|
| 1860 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 1861 |
+
"loss": 0.9144,
|
| 1862 |
+
"step": 2760
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 23.0,
|
| 1866 |
+
"eval_accuracy": 0.534037558685446,
|
| 1867 |
+
"eval_loss": 1.4117422103881836,
|
| 1868 |
+
"eval_runtime": 23.4898,
|
| 1869 |
+
"eval_samples_per_second": 72.542,
|
| 1870 |
+
"eval_steps_per_second": 2.299,
|
| 1871 |
+
"step": 2760
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 23.08,
|
| 1875 |
+
"learning_rate": 1.2808641975308644e-05,
|
| 1876 |
+
"loss": 0.946,
|
| 1877 |
+
"step": 2770
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 23.17,
|
| 1881 |
+
"learning_rate": 1.2654320987654323e-05,
|
| 1882 |
+
"loss": 0.9184,
|
| 1883 |
+
"step": 2780
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 23.25,
|
| 1887 |
+
"learning_rate": 1.25e-05,
|
| 1888 |
+
"loss": 0.8966,
|
| 1889 |
+
"step": 2790
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 23.33,
|
| 1893 |
+
"learning_rate": 1.2345679012345678e-05,
|
| 1894 |
+
"loss": 0.9174,
|
| 1895 |
+
"step": 2800
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 23.42,
|
| 1899 |
+
"learning_rate": 1.219135802469136e-05,
|
| 1900 |
+
"loss": 0.9383,
|
| 1901 |
+
"step": 2810
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 23.5,
|
| 1905 |
+
"learning_rate": 1.2037037037037037e-05,
|
| 1906 |
+
"loss": 0.942,
|
| 1907 |
+
"step": 2820
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 23.58,
|
| 1911 |
+
"learning_rate": 1.1882716049382716e-05,
|
| 1912 |
+
"loss": 0.9137,
|
| 1913 |
+
"step": 2830
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 23.67,
|
| 1917 |
+
"learning_rate": 1.1728395061728396e-05,
|
| 1918 |
+
"loss": 0.933,
|
| 1919 |
+
"step": 2840
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 23.75,
|
| 1923 |
+
"learning_rate": 1.1574074074074075e-05,
|
| 1924 |
+
"loss": 0.9348,
|
| 1925 |
+
"step": 2850
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 23.83,
|
| 1929 |
+
"learning_rate": 1.1419753086419753e-05,
|
| 1930 |
+
"loss": 0.8839,
|
| 1931 |
+
"step": 2860
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 23.92,
|
| 1935 |
+
"learning_rate": 1.1265432098765432e-05,
|
| 1936 |
+
"loss": 0.9548,
|
| 1937 |
+
"step": 2870
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 24.0,
|
| 1941 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 1942 |
+
"loss": 0.9771,
|
| 1943 |
+
"step": 2880
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 24.0,
|
| 1947 |
+
"eval_accuracy": 0.5357981220657277,
|
| 1948 |
+
"eval_loss": 1.383649230003357,
|
| 1949 |
+
"eval_runtime": 23.4971,
|
| 1950 |
+
"eval_samples_per_second": 72.52,
|
| 1951 |
+
"eval_steps_per_second": 2.298,
|
| 1952 |
+
"step": 2880
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 24.08,
|
| 1956 |
+
"learning_rate": 1.0956790123456791e-05,
|
| 1957 |
+
"loss": 0.9594,
|
| 1958 |
+
"step": 2890
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 24.17,
|
| 1962 |
+
"learning_rate": 1.0802469135802469e-05,
|
| 1963 |
+
"loss": 0.9391,
|
| 1964 |
+
"step": 2900
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 24.25,
|
| 1968 |
+
"learning_rate": 1.0648148148148148e-05,
|
| 1969 |
+
"loss": 0.9376,
|
| 1970 |
+
"step": 2910
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 24.33,
|
| 1974 |
+
"learning_rate": 1.0493827160493827e-05,
|
| 1975 |
+
"loss": 0.9284,
|
| 1976 |
+
"step": 2920
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 24.42,
|
| 1980 |
+
"learning_rate": 1.0339506172839507e-05,
|
| 1981 |
+
"loss": 0.9186,
|
| 1982 |
+
"step": 2930
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 24.5,
|
| 1986 |
+
"learning_rate": 1.0185185185185185e-05,
|
| 1987 |
+
"loss": 0.8916,
|
| 1988 |
+
"step": 2940
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 24.58,
|
| 1992 |
+
"learning_rate": 1.0030864197530866e-05,
|
| 1993 |
+
"loss": 0.9302,
|
| 1994 |
+
"step": 2950
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 24.67,
|
| 1998 |
+
"learning_rate": 9.876543209876543e-06,
|
| 1999 |
+
"loss": 0.9572,
|
| 2000 |
+
"step": 2960
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 24.75,
|
| 2004 |
+
"learning_rate": 9.722222222222223e-06,
|
| 2005 |
+
"loss": 0.8967,
|
| 2006 |
+
"step": 2970
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 24.83,
|
| 2010 |
+
"learning_rate": 9.5679012345679e-06,
|
| 2011 |
+
"loss": 0.9554,
|
| 2012 |
+
"step": 2980
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 24.92,
|
| 2016 |
+
"learning_rate": 9.413580246913581e-06,
|
| 2017 |
+
"loss": 0.9231,
|
| 2018 |
+
"step": 2990
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 25.0,
|
| 2022 |
+
"learning_rate": 9.259259259259259e-06,
|
| 2023 |
+
"loss": 0.8994,
|
| 2024 |
+
"step": 3000
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 25.0,
|
| 2028 |
+
"eval_accuracy": 0.528169014084507,
|
| 2029 |
+
"eval_loss": 1.5077193975448608,
|
| 2030 |
+
"eval_runtime": 23.4692,
|
| 2031 |
+
"eval_samples_per_second": 72.606,
|
| 2032 |
+
"eval_steps_per_second": 2.301,
|
| 2033 |
+
"step": 3000
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 25.08,
|
| 2037 |
+
"learning_rate": 9.104938271604939e-06,
|
| 2038 |
+
"loss": 0.9479,
|
| 2039 |
+
"step": 3010
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 25.17,
|
| 2043 |
+
"learning_rate": 8.950617283950618e-06,
|
| 2044 |
+
"loss": 0.9191,
|
| 2045 |
+
"step": 3020
|
| 2046 |
+
},
|
| 2047 |
+
{
|
| 2048 |
+
"epoch": 25.25,
|
| 2049 |
+
"learning_rate": 8.796296296296297e-06,
|
| 2050 |
+
"loss": 0.9563,
|
| 2051 |
+
"step": 3030
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 25.33,
|
| 2055 |
+
"learning_rate": 8.641975308641975e-06,
|
| 2056 |
+
"loss": 0.9932,
|
| 2057 |
+
"step": 3040
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 25.42,
|
| 2061 |
+
"learning_rate": 8.487654320987654e-06,
|
| 2062 |
+
"loss": 0.9327,
|
| 2063 |
+
"step": 3050
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 25.5,
|
| 2067 |
+
"learning_rate": 8.333333333333334e-06,
|
| 2068 |
+
"loss": 0.9056,
|
| 2069 |
+
"step": 3060
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 25.58,
|
| 2073 |
+
"learning_rate": 8.179012345679013e-06,
|
| 2074 |
+
"loss": 0.9228,
|
| 2075 |
+
"step": 3070
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 25.67,
|
| 2079 |
+
"learning_rate": 8.02469135802469e-06,
|
| 2080 |
+
"loss": 0.9059,
|
| 2081 |
+
"step": 3080
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 25.75,
|
| 2085 |
+
"learning_rate": 7.87037037037037e-06,
|
| 2086 |
+
"loss": 0.9224,
|
| 2087 |
+
"step": 3090
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 25.83,
|
| 2091 |
+
"learning_rate": 7.71604938271605e-06,
|
| 2092 |
+
"loss": 0.8891,
|
| 2093 |
+
"step": 3100
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 25.92,
|
| 2097 |
+
"learning_rate": 7.561728395061729e-06,
|
| 2098 |
+
"loss": 0.9142,
|
| 2099 |
+
"step": 3110
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 26.0,
|
| 2103 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 2104 |
+
"loss": 0.9061,
|
| 2105 |
+
"step": 3120
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 26.0,
|
| 2109 |
+
"eval_accuracy": 0.5328638497652582,
|
| 2110 |
+
"eval_loss": 1.4622243642807007,
|
| 2111 |
+
"eval_runtime": 23.8116,
|
| 2112 |
+
"eval_samples_per_second": 71.562,
|
| 2113 |
+
"eval_steps_per_second": 2.268,
|
| 2114 |
+
"step": 3120
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 26.08,
|
| 2118 |
+
"learning_rate": 7.253086419753087e-06,
|
| 2119 |
+
"loss": 0.9487,
|
| 2120 |
+
"step": 3130
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 26.17,
|
| 2124 |
+
"learning_rate": 7.098765432098765e-06,
|
| 2125 |
+
"loss": 0.9639,
|
| 2126 |
+
"step": 3140
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 26.25,
|
| 2130 |
+
"learning_rate": 6.944444444444445e-06,
|
| 2131 |
+
"loss": 0.9248,
|
| 2132 |
+
"step": 3150
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 26.33,
|
| 2136 |
+
"learning_rate": 6.790123456790123e-06,
|
| 2137 |
+
"loss": 0.901,
|
| 2138 |
+
"step": 3160
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 26.42,
|
| 2142 |
+
"learning_rate": 6.635802469135803e-06,
|
| 2143 |
+
"loss": 0.9158,
|
| 2144 |
+
"step": 3170
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 26.5,
|
| 2148 |
+
"learning_rate": 6.481481481481481e-06,
|
| 2149 |
+
"loss": 0.9161,
|
| 2150 |
+
"step": 3180
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 26.58,
|
| 2154 |
+
"learning_rate": 6.3271604938271615e-06,
|
| 2155 |
+
"loss": 0.9121,
|
| 2156 |
+
"step": 3190
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 26.67,
|
| 2160 |
+
"learning_rate": 6.172839506172839e-06,
|
| 2161 |
+
"loss": 0.9275,
|
| 2162 |
+
"step": 3200
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 26.75,
|
| 2166 |
+
"learning_rate": 6.0185185185185185e-06,
|
| 2167 |
+
"loss": 0.9336,
|
| 2168 |
+
"step": 3210
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 26.83,
|
| 2172 |
+
"learning_rate": 5.864197530864198e-06,
|
| 2173 |
+
"loss": 0.9263,
|
| 2174 |
+
"step": 3220
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 26.92,
|
| 2178 |
+
"learning_rate": 5.7098765432098764e-06,
|
| 2179 |
+
"loss": 0.932,
|
| 2180 |
+
"step": 3230
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 27.0,
|
| 2184 |
+
"learning_rate": 5.555555555555556e-06,
|
| 2185 |
+
"loss": 0.9071,
|
| 2186 |
+
"step": 3240
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 27.0,
|
| 2190 |
+
"eval_accuracy": 0.5393192488262911,
|
| 2191 |
+
"eval_loss": 1.430330753326416,
|
| 2192 |
+
"eval_runtime": 23.403,
|
| 2193 |
+
"eval_samples_per_second": 72.811,
|
| 2194 |
+
"eval_steps_per_second": 2.307,
|
| 2195 |
+
"step": 3240
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 27.08,
|
| 2199 |
+
"learning_rate": 5.401234567901234e-06,
|
| 2200 |
+
"loss": 0.9078,
|
| 2201 |
+
"step": 3250
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 27.17,
|
| 2205 |
+
"learning_rate": 5.246913580246914e-06,
|
| 2206 |
+
"loss": 0.9895,
|
| 2207 |
+
"step": 3260
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 27.25,
|
| 2211 |
+
"learning_rate": 5.092592592592592e-06,
|
| 2212 |
+
"loss": 0.8999,
|
| 2213 |
+
"step": 3270
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 27.33,
|
| 2217 |
+
"learning_rate": 4.938271604938272e-06,
|
| 2218 |
+
"loss": 0.9041,
|
| 2219 |
+
"step": 3280
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 27.42,
|
| 2223 |
+
"learning_rate": 4.78395061728395e-06,
|
| 2224 |
+
"loss": 0.958,
|
| 2225 |
+
"step": 3290
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 27.5,
|
| 2229 |
+
"learning_rate": 4.6296296296296296e-06,
|
| 2230 |
+
"loss": 0.9329,
|
| 2231 |
+
"step": 3300
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 27.58,
|
| 2235 |
+
"learning_rate": 4.475308641975309e-06,
|
| 2236 |
+
"loss": 0.902,
|
| 2237 |
+
"step": 3310
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 27.67,
|
| 2241 |
+
"learning_rate": 4.3209876543209875e-06,
|
| 2242 |
+
"loss": 0.9415,
|
| 2243 |
+
"step": 3320
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 27.75,
|
| 2247 |
+
"learning_rate": 4.166666666666667e-06,
|
| 2248 |
+
"loss": 0.8901,
|
| 2249 |
+
"step": 3330
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 27.83,
|
| 2253 |
+
"learning_rate": 4.012345679012345e-06,
|
| 2254 |
+
"loss": 0.9495,
|
| 2255 |
+
"step": 3340
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 27.92,
|
| 2259 |
+
"learning_rate": 3.858024691358025e-06,
|
| 2260 |
+
"loss": 0.9019,
|
| 2261 |
+
"step": 3350
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 28.0,
|
| 2265 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 2266 |
+
"loss": 0.9288,
|
| 2267 |
+
"step": 3360
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 28.0,
|
| 2271 |
+
"eval_accuracy": 0.5328638497652582,
|
| 2272 |
+
"eval_loss": 1.4555658102035522,
|
| 2273 |
+
"eval_runtime": 23.4988,
|
| 2274 |
+
"eval_samples_per_second": 72.514,
|
| 2275 |
+
"eval_steps_per_second": 2.298,
|
| 2276 |
+
"step": 3360
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 28.08,
|
| 2280 |
+
"learning_rate": 3.5493827160493827e-06,
|
| 2281 |
+
"loss": 0.9258,
|
| 2282 |
+
"step": 3370
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 28.17,
|
| 2286 |
+
"learning_rate": 3.3950617283950617e-06,
|
| 2287 |
+
"loss": 0.9017,
|
| 2288 |
+
"step": 3380
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 28.25,
|
| 2292 |
+
"learning_rate": 3.2407407407407406e-06,
|
| 2293 |
+
"loss": 0.8841,
|
| 2294 |
+
"step": 3390
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 28.33,
|
| 2298 |
+
"learning_rate": 3.0864197530864196e-06,
|
| 2299 |
+
"loss": 0.8903,
|
| 2300 |
+
"step": 3400
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 28.42,
|
| 2304 |
+
"learning_rate": 2.932098765432099e-06,
|
| 2305 |
+
"loss": 0.9242,
|
| 2306 |
+
"step": 3410
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 28.5,
|
| 2310 |
+
"learning_rate": 2.777777777777778e-06,
|
| 2311 |
+
"loss": 0.9272,
|
| 2312 |
+
"step": 3420
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 28.58,
|
| 2316 |
+
"learning_rate": 2.623456790123457e-06,
|
| 2317 |
+
"loss": 0.8952,
|
| 2318 |
+
"step": 3430
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 28.67,
|
| 2322 |
+
"learning_rate": 2.469135802469136e-06,
|
| 2323 |
+
"loss": 0.9263,
|
| 2324 |
+
"step": 3440
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 28.75,
|
| 2328 |
+
"learning_rate": 2.3148148148148148e-06,
|
| 2329 |
+
"loss": 0.9586,
|
| 2330 |
+
"step": 3450
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 28.83,
|
| 2334 |
+
"learning_rate": 2.1604938271604937e-06,
|
| 2335 |
+
"loss": 0.929,
|
| 2336 |
+
"step": 3460
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 28.92,
|
| 2340 |
+
"learning_rate": 2.0061728395061727e-06,
|
| 2341 |
+
"loss": 0.9409,
|
| 2342 |
+
"step": 3470
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 29.0,
|
| 2346 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 2347 |
+
"loss": 0.9285,
|
| 2348 |
+
"step": 3480
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 29.0,
|
| 2352 |
+
"eval_accuracy": 0.5446009389671361,
|
| 2353 |
+
"eval_loss": 1.3899754285812378,
|
| 2354 |
+
"eval_runtime": 23.3569,
|
| 2355 |
+
"eval_samples_per_second": 72.955,
|
| 2356 |
+
"eval_steps_per_second": 2.312,
|
| 2357 |
+
"step": 3480
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 29.08,
|
| 2361 |
+
"learning_rate": 1.6975308641975308e-06,
|
| 2362 |
+
"loss": 0.9286,
|
| 2363 |
+
"step": 3490
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 29.17,
|
| 2367 |
+
"learning_rate": 1.5432098765432098e-06,
|
| 2368 |
+
"loss": 0.9462,
|
| 2369 |
+
"step": 3500
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 29.25,
|
| 2373 |
+
"learning_rate": 1.388888888888889e-06,
|
| 2374 |
+
"loss": 0.9352,
|
| 2375 |
+
"step": 3510
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 29.33,
|
| 2379 |
+
"learning_rate": 1.234567901234568e-06,
|
| 2380 |
+
"loss": 0.9116,
|
| 2381 |
+
"step": 3520
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 29.42,
|
| 2385 |
+
"learning_rate": 1.0802469135802469e-06,
|
| 2386 |
+
"loss": 0.8827,
|
| 2387 |
+
"step": 3530
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 29.5,
|
| 2391 |
+
"learning_rate": 9.259259259259259e-07,
|
| 2392 |
+
"loss": 0.9091,
|
| 2393 |
+
"step": 3540
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 29.58,
|
| 2397 |
+
"learning_rate": 7.716049382716049e-07,
|
| 2398 |
+
"loss": 0.9104,
|
| 2399 |
+
"step": 3550
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 29.67,
|
| 2403 |
+
"learning_rate": 6.17283950617284e-07,
|
| 2404 |
+
"loss": 0.9446,
|
| 2405 |
+
"step": 3560
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 29.75,
|
| 2409 |
+
"learning_rate": 4.6296296296296297e-07,
|
| 2410 |
+
"loss": 0.9111,
|
| 2411 |
+
"step": 3570
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 29.83,
|
| 2415 |
+
"learning_rate": 3.08641975308642e-07,
|
| 2416 |
+
"loss": 0.9165,
|
| 2417 |
+
"step": 3580
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 29.92,
|
| 2421 |
+
"learning_rate": 1.54320987654321e-07,
|
| 2422 |
+
"loss": 0.9068,
|
| 2423 |
+
"step": 3590
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 30.0,
|
| 2427 |
+
"learning_rate": 0.0,
|
| 2428 |
+
"loss": 0.8955,
|
| 2429 |
+
"step": 3600
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 30.0,
|
| 2433 |
+
"eval_accuracy": 0.5387323943661971,
|
| 2434 |
+
"eval_loss": 1.4081873893737793,
|
| 2435 |
+
"eval_runtime": 24.0466,
|
| 2436 |
+
"eval_samples_per_second": 70.862,
|
| 2437 |
+
"eval_steps_per_second": 2.246,
|
| 2438 |
+
"step": 3600
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 30.0,
|
| 2442 |
+
"step": 3600,
|
| 2443 |
+
"total_flos": 3.5642168535562445e+19,
|
| 2444 |
+
"train_loss": 0.9839045113987392,
|
| 2445 |
+
"train_runtime": 19975.6243,
|
| 2446 |
+
"train_samples_per_second": 23.025,
|
| 2447 |
+
"train_steps_per_second": 0.18
|
| 2448 |
}
|
| 2449 |
],
|
| 2450 |
+
"max_steps": 3600,
|
| 2451 |
+
"num_train_epochs": 30,
|
| 2452 |
+
"total_flos": 3.5642168535562445e+19,
|
| 2453 |
"trial_name": null,
|
| 2454 |
"trial_params": null
|
| 2455 |
}
|