Upload 9 files
Browse files- config.json +1 -1
- model.safetensors +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1132 -1132
- training_args.bin +1 -1
config.json
CHANGED
|
@@ -41,6 +41,6 @@
|
|
| 41 |
"sep_token_id": 50282,
|
| 42 |
"sparse_pred_ignore_index": -100,
|
| 43 |
"sparse_prediction": false,
|
| 44 |
-
"transformers_version": "4.56.
|
| 45 |
"vocab_size": 50368
|
| 46 |
}
|
|
|
|
| 41 |
"sep_token_id": 50282,
|
| 42 |
"sparse_pred_ignore_index": -100,
|
| 43 |
"sparse_prediction": false,
|
| 44 |
+
"transformers_version": "4.56.1",
|
| 45 |
"vocab_size": 50368
|
| 46 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1583544840
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4c513c87136b7061f89a0058cf57e10feabc8eaa6dc84ac77ff0f5a223c2f19c
|
| 3 |
size 1583544840
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:449e44f9adf4d083aec6625b9110f6a9a09baba982e3a32de94ff0c135c00f4d
|
| 3 |
size 14645
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d0df6d48ac6c8d2a3fe965d9b7a645f9b425ec23c31765b3bbc57f64cf0fee9
|
| 3 |
size 1465
|
trainer_state.json
CHANGED
|
@@ -2,7 +2,7 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 1000,
|
| 7 |
"global_step": 25000,
|
| 8 |
"is_hyper_param_search": false,
|
|
@@ -10,1965 +10,1965 @@
|
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
-
"epoch":
|
| 14 |
-
"grad_norm":
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
-
"loss":
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
-
"epoch": 0.
|
| 21 |
-
"grad_norm": 1.
|
| 22 |
-
"learning_rate":
|
| 23 |
-
"loss":
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"epoch": 0.
|
| 28 |
-
"grad_norm":
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"loss":
|
| 31 |
"step": 200
|
| 32 |
},
|
| 33 |
{
|
| 34 |
-
"epoch": 0.
|
| 35 |
-
"grad_norm":
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss":
|
| 38 |
"step": 300
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 0.
|
| 42 |
-
"grad_norm":
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss":
|
| 45 |
"step": 400
|
| 46 |
},
|
| 47 |
{
|
| 48 |
-
"epoch": 0.
|
| 49 |
-
"grad_norm": 1.
|
| 50 |
-
"learning_rate":
|
| 51 |
-
"loss":
|
| 52 |
"step": 500
|
| 53 |
},
|
| 54 |
{
|
| 55 |
-
"epoch": 0.
|
| 56 |
-
"grad_norm":
|
| 57 |
-
"learning_rate":
|
| 58 |
-
"loss":
|
| 59 |
"step": 600
|
| 60 |
},
|
| 61 |
{
|
| 62 |
-
"epoch": 0.
|
| 63 |
-
"grad_norm":
|
| 64 |
-
"learning_rate":
|
| 65 |
-
"loss":
|
| 66 |
"step": 700
|
| 67 |
},
|
| 68 |
{
|
| 69 |
-
"epoch": 0.
|
| 70 |
-
"grad_norm":
|
| 71 |
-
"learning_rate":
|
| 72 |
-
"loss":
|
| 73 |
"step": 800
|
| 74 |
},
|
| 75 |
{
|
| 76 |
-
"epoch": 0.
|
| 77 |
-
"grad_norm":
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss":
|
| 80 |
"step": 900
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch": 0.
|
| 84 |
-
"grad_norm":
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss":
|
| 87 |
"step": 1000
|
| 88 |
},
|
| 89 |
{
|
| 90 |
-
"epoch": 0.
|
| 91 |
-
"eval_loss": 1.
|
| 92 |
-
"eval_runtime":
|
| 93 |
-
"eval_samples_per_second":
|
| 94 |
-
"eval_steps_per_second": 2.
|
| 95 |
"step": 1000
|
| 96 |
},
|
| 97 |
{
|
| 98 |
-
"epoch": 0.
|
| 99 |
-
"grad_norm":
|
| 100 |
-
"learning_rate":
|
| 101 |
-
"loss":
|
| 102 |
"step": 1100
|
| 103 |
},
|
| 104 |
{
|
| 105 |
-
"epoch": 0.
|
| 106 |
-
"grad_norm":
|
| 107 |
-
"learning_rate":
|
| 108 |
-
"loss":
|
| 109 |
"step": 1200
|
| 110 |
},
|
| 111 |
{
|
| 112 |
-
"epoch": 0.
|
| 113 |
-
"grad_norm":
|
| 114 |
-
"learning_rate":
|
| 115 |
-
"loss":
|
| 116 |
"step": 1300
|
| 117 |
},
|
| 118 |
{
|
| 119 |
-
"epoch": 0.
|
| 120 |
-
"grad_norm":
|
| 121 |
-
"learning_rate":
|
| 122 |
-
"loss":
|
| 123 |
"step": 1400
|
| 124 |
},
|
| 125 |
{
|
| 126 |
-
"epoch": 0.
|
| 127 |
-
"grad_norm": 1.
|
| 128 |
-
"learning_rate":
|
| 129 |
-
"loss":
|
| 130 |
"step": 1500
|
| 131 |
},
|
| 132 |
{
|
| 133 |
-
"epoch": 0.
|
| 134 |
-
"grad_norm":
|
| 135 |
-
"learning_rate":
|
| 136 |
-
"loss":
|
| 137 |
"step": 1600
|
| 138 |
},
|
| 139 |
{
|
| 140 |
-
"epoch": 0.
|
| 141 |
-
"grad_norm":
|
| 142 |
-
"learning_rate":
|
| 143 |
-
"loss":
|
| 144 |
"step": 1700
|
| 145 |
},
|
| 146 |
{
|
| 147 |
-
"epoch": 0.
|
| 148 |
-
"grad_norm":
|
| 149 |
-
"learning_rate":
|
| 150 |
-
"loss":
|
| 151 |
"step": 1800
|
| 152 |
},
|
| 153 |
{
|
| 154 |
-
"epoch": 0.
|
| 155 |
-
"grad_norm":
|
| 156 |
-
"learning_rate":
|
| 157 |
-
"loss":
|
| 158 |
"step": 1900
|
| 159 |
},
|
| 160 |
{
|
| 161 |
-
"epoch": 0.
|
| 162 |
-
"grad_norm":
|
| 163 |
-
"learning_rate":
|
| 164 |
-
"loss":
|
| 165 |
"step": 2000
|
| 166 |
},
|
| 167 |
{
|
| 168 |
-
"epoch": 0.
|
| 169 |
-
"eval_loss": 1.
|
| 170 |
-
"eval_runtime":
|
| 171 |
-
"eval_samples_per_second":
|
| 172 |
-
"eval_steps_per_second": 2.
|
| 173 |
"step": 2000
|
| 174 |
},
|
| 175 |
{
|
| 176 |
-
"epoch": 0.
|
| 177 |
-
"grad_norm":
|
| 178 |
-
"learning_rate":
|
| 179 |
-
"loss":
|
| 180 |
"step": 2100
|
| 181 |
},
|
| 182 |
{
|
| 183 |
-
"epoch": 0.
|
| 184 |
-
"grad_norm":
|
| 185 |
-
"learning_rate":
|
| 186 |
-
"loss":
|
| 187 |
"step": 2200
|
| 188 |
},
|
| 189 |
{
|
| 190 |
-
"epoch": 0.
|
| 191 |
-
"grad_norm":
|
| 192 |
-
"learning_rate":
|
| 193 |
-
"loss":
|
| 194 |
"step": 2300
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch": 0.
|
| 198 |
-
"grad_norm":
|
| 199 |
-
"learning_rate":
|
| 200 |
-
"loss":
|
| 201 |
"step": 2400
|
| 202 |
},
|
| 203 |
{
|
| 204 |
-
"epoch": 0.
|
| 205 |
-
"grad_norm":
|
| 206 |
-
"learning_rate":
|
| 207 |
-
"loss":
|
| 208 |
"step": 2500
|
| 209 |
},
|
| 210 |
{
|
| 211 |
-
"epoch": 0.
|
| 212 |
-
"grad_norm":
|
| 213 |
-
"learning_rate":
|
| 214 |
-
"loss":
|
| 215 |
"step": 2600
|
| 216 |
},
|
| 217 |
{
|
| 218 |
-
"epoch": 0.
|
| 219 |
-
"grad_norm":
|
| 220 |
-
"learning_rate":
|
| 221 |
-
"loss":
|
| 222 |
"step": 2700
|
| 223 |
},
|
| 224 |
{
|
| 225 |
-
"epoch": 0.
|
| 226 |
-
"grad_norm":
|
| 227 |
-
"learning_rate":
|
| 228 |
-
"loss":
|
| 229 |
"step": 2800
|
| 230 |
},
|
| 231 |
{
|
| 232 |
-
"epoch": 0.
|
| 233 |
-
"grad_norm":
|
| 234 |
-
"learning_rate":
|
| 235 |
-
"loss":
|
| 236 |
"step": 2900
|
| 237 |
},
|
| 238 |
{
|
| 239 |
-
"epoch": 0.
|
| 240 |
-
"grad_norm":
|
| 241 |
-
"learning_rate":
|
| 242 |
-
"loss":
|
| 243 |
"step": 3000
|
| 244 |
},
|
| 245 |
{
|
| 246 |
-
"epoch": 0.
|
| 247 |
-
"eval_loss": 1.
|
| 248 |
-
"eval_runtime":
|
| 249 |
-
"eval_samples_per_second":
|
| 250 |
-
"eval_steps_per_second": 2.
|
| 251 |
"step": 3000
|
| 252 |
},
|
| 253 |
{
|
| 254 |
-
"epoch": 0.
|
| 255 |
-
"grad_norm":
|
| 256 |
-
"learning_rate":
|
| 257 |
-
"loss":
|
| 258 |
"step": 3100
|
| 259 |
},
|
| 260 |
{
|
| 261 |
-
"epoch": 0.
|
| 262 |
-
"grad_norm":
|
| 263 |
-
"learning_rate":
|
| 264 |
-
"loss":
|
| 265 |
"step": 3200
|
| 266 |
},
|
| 267 |
{
|
| 268 |
-
"epoch": 0.
|
| 269 |
-
"grad_norm":
|
| 270 |
-
"learning_rate":
|
| 271 |
-
"loss":
|
| 272 |
"step": 3300
|
| 273 |
},
|
| 274 |
{
|
| 275 |
-
"epoch": 0.
|
| 276 |
-
"grad_norm":
|
| 277 |
-
"learning_rate":
|
| 278 |
-
"loss":
|
| 279 |
"step": 3400
|
| 280 |
},
|
| 281 |
{
|
| 282 |
-
"epoch": 0.
|
| 283 |
-
"grad_norm":
|
| 284 |
-
"learning_rate":
|
| 285 |
-
"loss":
|
| 286 |
"step": 3500
|
| 287 |
},
|
| 288 |
{
|
| 289 |
-
"epoch": 0.
|
| 290 |
-
"grad_norm":
|
| 291 |
-
"learning_rate":
|
| 292 |
-
"loss":
|
| 293 |
"step": 3600
|
| 294 |
},
|
| 295 |
{
|
| 296 |
-
"epoch": 0.
|
| 297 |
-
"grad_norm":
|
| 298 |
-
"learning_rate":
|
| 299 |
-
"loss":
|
| 300 |
"step": 3700
|
| 301 |
},
|
| 302 |
{
|
| 303 |
-
"epoch": 0.
|
| 304 |
-
"grad_norm":
|
| 305 |
-
"learning_rate":
|
| 306 |
-
"loss":
|
| 307 |
"step": 3800
|
| 308 |
},
|
| 309 |
{
|
| 310 |
-
"epoch": 0.
|
| 311 |
-
"grad_norm":
|
| 312 |
-
"learning_rate":
|
| 313 |
-
"loss":
|
| 314 |
"step": 3900
|
| 315 |
},
|
| 316 |
{
|
| 317 |
-
"epoch": 0.
|
| 318 |
-
"grad_norm":
|
| 319 |
-
"learning_rate":
|
| 320 |
-
"loss":
|
| 321 |
"step": 4000
|
| 322 |
},
|
| 323 |
{
|
| 324 |
-
"epoch": 0.
|
| 325 |
-
"eval_loss": 1.
|
| 326 |
-
"eval_runtime":
|
| 327 |
-
"eval_samples_per_second":
|
| 328 |
-
"eval_steps_per_second": 2.
|
| 329 |
"step": 4000
|
| 330 |
},
|
| 331 |
{
|
| 332 |
-
"epoch": 0.
|
| 333 |
-
"grad_norm":
|
| 334 |
-
"learning_rate":
|
| 335 |
-
"loss":
|
| 336 |
"step": 4100
|
| 337 |
},
|
| 338 |
{
|
| 339 |
-
"epoch": 0.
|
| 340 |
-
"grad_norm":
|
| 341 |
-
"learning_rate":
|
| 342 |
-
"loss":
|
| 343 |
"step": 4200
|
| 344 |
},
|
| 345 |
{
|
| 346 |
-
"epoch": 0.
|
| 347 |
-
"grad_norm":
|
| 348 |
-
"learning_rate":
|
| 349 |
-
"loss":
|
| 350 |
"step": 4300
|
| 351 |
},
|
| 352 |
{
|
| 353 |
-
"epoch": 0.
|
| 354 |
-
"grad_norm":
|
| 355 |
-
"learning_rate":
|
| 356 |
-
"loss":
|
| 357 |
"step": 4400
|
| 358 |
},
|
| 359 |
{
|
| 360 |
-
"epoch": 0.
|
| 361 |
-
"grad_norm":
|
| 362 |
-
"learning_rate":
|
| 363 |
-
"loss":
|
| 364 |
"step": 4500
|
| 365 |
},
|
| 366 |
{
|
| 367 |
-
"epoch": 0.
|
| 368 |
-
"grad_norm":
|
| 369 |
-
"learning_rate":
|
| 370 |
-
"loss":
|
| 371 |
"step": 4600
|
| 372 |
},
|
| 373 |
{
|
| 374 |
-
"epoch": 0.
|
| 375 |
-
"grad_norm":
|
| 376 |
-
"learning_rate":
|
| 377 |
-
"loss":
|
| 378 |
"step": 4700
|
| 379 |
},
|
| 380 |
{
|
| 381 |
-
"epoch": 0.
|
| 382 |
-
"grad_norm":
|
| 383 |
-
"learning_rate":
|
| 384 |
-
"loss":
|
| 385 |
"step": 4800
|
| 386 |
},
|
| 387 |
{
|
| 388 |
-
"epoch": 0.
|
| 389 |
-
"grad_norm":
|
| 390 |
-
"learning_rate":
|
| 391 |
-
"loss":
|
| 392 |
"step": 4900
|
| 393 |
},
|
| 394 |
{
|
| 395 |
-
"epoch": 0.
|
| 396 |
-
"grad_norm":
|
| 397 |
-
"learning_rate":
|
| 398 |
-
"loss":
|
| 399 |
"step": 5000
|
| 400 |
},
|
| 401 |
{
|
| 402 |
-
"epoch": 0.
|
| 403 |
-
"eval_loss": 1.
|
| 404 |
-
"eval_runtime": 98.
|
| 405 |
-
"eval_samples_per_second":
|
| 406 |
-
"eval_steps_per_second": 2.
|
| 407 |
"step": 5000
|
| 408 |
},
|
| 409 |
{
|
| 410 |
-
"epoch": 0.
|
| 411 |
-
"grad_norm":
|
| 412 |
-
"learning_rate":
|
| 413 |
-
"loss":
|
| 414 |
"step": 5100
|
| 415 |
},
|
| 416 |
{
|
| 417 |
-
"epoch": 0.
|
| 418 |
-
"grad_norm":
|
| 419 |
-
"learning_rate":
|
| 420 |
-
"loss":
|
| 421 |
"step": 5200
|
| 422 |
},
|
| 423 |
{
|
| 424 |
-
"epoch": 0.
|
| 425 |
-
"grad_norm":
|
| 426 |
-
"learning_rate":
|
| 427 |
-
"loss":
|
| 428 |
"step": 5300
|
| 429 |
},
|
| 430 |
{
|
| 431 |
-
"epoch": 0.
|
| 432 |
-
"grad_norm":
|
| 433 |
-
"learning_rate":
|
| 434 |
-
"loss":
|
| 435 |
"step": 5400
|
| 436 |
},
|
| 437 |
{
|
| 438 |
-
"epoch": 0.
|
| 439 |
-
"grad_norm":
|
| 440 |
-
"learning_rate":
|
| 441 |
-
"loss":
|
| 442 |
"step": 5500
|
| 443 |
},
|
| 444 |
{
|
| 445 |
-
"epoch": 0.
|
| 446 |
-
"grad_norm":
|
| 447 |
-
"learning_rate":
|
| 448 |
-
"loss":
|
| 449 |
"step": 5600
|
| 450 |
},
|
| 451 |
{
|
| 452 |
-
"epoch": 0.
|
| 453 |
-
"grad_norm":
|
| 454 |
-
"learning_rate":
|
| 455 |
-
"loss":
|
| 456 |
"step": 5700
|
| 457 |
},
|
| 458 |
{
|
| 459 |
-
"epoch": 0.
|
| 460 |
-
"grad_norm":
|
| 461 |
-
"learning_rate":
|
| 462 |
-
"loss":
|
| 463 |
"step": 5800
|
| 464 |
},
|
| 465 |
{
|
| 466 |
-
"epoch": 0.
|
| 467 |
-
"grad_norm":
|
| 468 |
-
"learning_rate":
|
| 469 |
-
"loss":
|
| 470 |
"step": 5900
|
| 471 |
},
|
| 472 |
{
|
| 473 |
-
"epoch": 0.
|
| 474 |
-
"grad_norm":
|
| 475 |
-
"learning_rate":
|
| 476 |
-
"loss":
|
| 477 |
"step": 6000
|
| 478 |
},
|
| 479 |
{
|
| 480 |
-
"epoch": 0.
|
| 481 |
-
"eval_loss": 1.
|
| 482 |
-
"eval_runtime":
|
| 483 |
-
"eval_samples_per_second":
|
| 484 |
-
"eval_steps_per_second": 2.
|
| 485 |
"step": 6000
|
| 486 |
},
|
| 487 |
{
|
| 488 |
-
"epoch": 0.
|
| 489 |
-
"grad_norm": 1.
|
| 490 |
-
"learning_rate":
|
| 491 |
-
"loss":
|
| 492 |
"step": 6100
|
| 493 |
},
|
| 494 |
{
|
| 495 |
-
"epoch": 0.
|
| 496 |
-
"grad_norm":
|
| 497 |
-
"learning_rate":
|
| 498 |
-
"loss":
|
| 499 |
"step": 6200
|
| 500 |
},
|
| 501 |
{
|
| 502 |
-
"epoch": 0.
|
| 503 |
-
"grad_norm":
|
| 504 |
-
"learning_rate":
|
| 505 |
-
"loss":
|
| 506 |
"step": 6300
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch": 0.
|
| 510 |
-
"grad_norm":
|
| 511 |
-
"learning_rate":
|
| 512 |
-
"loss":
|
| 513 |
"step": 6400
|
| 514 |
},
|
| 515 |
{
|
| 516 |
-
"epoch": 0.
|
| 517 |
-
"grad_norm":
|
| 518 |
-
"learning_rate":
|
| 519 |
-
"loss":
|
| 520 |
"step": 6500
|
| 521 |
},
|
| 522 |
{
|
| 523 |
-
"epoch": 0.
|
| 524 |
-
"grad_norm":
|
| 525 |
-
"learning_rate":
|
| 526 |
-
"loss":
|
| 527 |
"step": 6600
|
| 528 |
},
|
| 529 |
{
|
| 530 |
-
"epoch": 0.
|
| 531 |
-
"grad_norm":
|
| 532 |
-
"learning_rate":
|
| 533 |
-
"loss":
|
| 534 |
"step": 6700
|
| 535 |
},
|
| 536 |
{
|
| 537 |
-
"epoch": 0.
|
| 538 |
-
"grad_norm":
|
| 539 |
-
"learning_rate":
|
| 540 |
-
"loss":
|
| 541 |
"step": 6800
|
| 542 |
},
|
| 543 |
{
|
| 544 |
-
"epoch": 0.
|
| 545 |
-
"grad_norm":
|
| 546 |
-
"learning_rate":
|
| 547 |
-
"loss":
|
| 548 |
"step": 6900
|
| 549 |
},
|
| 550 |
{
|
| 551 |
-
"epoch": 0.
|
| 552 |
-
"grad_norm":
|
| 553 |
-
"learning_rate":
|
| 554 |
-
"loss":
|
| 555 |
"step": 7000
|
| 556 |
},
|
| 557 |
{
|
| 558 |
-
"epoch": 0.
|
| 559 |
-
"eval_loss": 1.
|
| 560 |
-
"eval_runtime":
|
| 561 |
-
"eval_samples_per_second":
|
| 562 |
-
"eval_steps_per_second": 2.
|
| 563 |
"step": 7000
|
| 564 |
},
|
| 565 |
{
|
| 566 |
-
"epoch": 0.
|
| 567 |
-
"grad_norm":
|
| 568 |
-
"learning_rate":
|
| 569 |
-
"loss":
|
| 570 |
"step": 7100
|
| 571 |
},
|
| 572 |
{
|
| 573 |
-
"epoch": 0.
|
| 574 |
-
"grad_norm":
|
| 575 |
-
"learning_rate":
|
| 576 |
-
"loss":
|
| 577 |
"step": 7200
|
| 578 |
},
|
| 579 |
{
|
| 580 |
-
"epoch": 0.
|
| 581 |
-
"grad_norm": 1.
|
| 582 |
-
"learning_rate":
|
| 583 |
-
"loss":
|
| 584 |
"step": 7300
|
| 585 |
},
|
| 586 |
{
|
| 587 |
-
"epoch": 0.
|
| 588 |
-
"grad_norm":
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss":
|
| 591 |
"step": 7400
|
| 592 |
},
|
| 593 |
{
|
| 594 |
-
"epoch": 0.
|
| 595 |
-
"grad_norm":
|
| 596 |
-
"learning_rate":
|
| 597 |
-
"loss":
|
| 598 |
"step": 7500
|
| 599 |
},
|
| 600 |
{
|
| 601 |
-
"epoch": 0.
|
| 602 |
-
"grad_norm":
|
| 603 |
-
"learning_rate":
|
| 604 |
-
"loss":
|
| 605 |
"step": 7600
|
| 606 |
},
|
| 607 |
{
|
| 608 |
-
"epoch": 0.
|
| 609 |
-
"grad_norm":
|
| 610 |
-
"learning_rate":
|
| 611 |
-
"loss":
|
| 612 |
"step": 7700
|
| 613 |
},
|
| 614 |
{
|
| 615 |
-
"epoch": 0.
|
| 616 |
-
"grad_norm":
|
| 617 |
-
"learning_rate":
|
| 618 |
-
"loss":
|
| 619 |
"step": 7800
|
| 620 |
},
|
| 621 |
{
|
| 622 |
-
"epoch": 0.
|
| 623 |
-
"grad_norm":
|
| 624 |
-
"learning_rate":
|
| 625 |
-
"loss":
|
| 626 |
"step": 7900
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch": 0.
|
| 630 |
-
"grad_norm":
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss":
|
| 633 |
"step": 8000
|
| 634 |
},
|
| 635 |
{
|
| 636 |
-
"epoch": 0.
|
| 637 |
-
"eval_loss": 1.
|
| 638 |
-
"eval_runtime":
|
| 639 |
-
"eval_samples_per_second":
|
| 640 |
-
"eval_steps_per_second": 2.
|
| 641 |
"step": 8000
|
| 642 |
},
|
| 643 |
{
|
| 644 |
-
"epoch": 0.
|
| 645 |
-
"grad_norm":
|
| 646 |
-
"learning_rate":
|
| 647 |
-
"loss":
|
| 648 |
"step": 8100
|
| 649 |
},
|
| 650 |
{
|
| 651 |
-
"epoch": 0.
|
| 652 |
-
"grad_norm":
|
| 653 |
-
"learning_rate":
|
| 654 |
-
"loss":
|
| 655 |
"step": 8200
|
| 656 |
},
|
| 657 |
{
|
| 658 |
-
"epoch": 0.
|
| 659 |
-
"grad_norm":
|
| 660 |
-
"learning_rate":
|
| 661 |
-
"loss":
|
| 662 |
"step": 8300
|
| 663 |
},
|
| 664 |
{
|
| 665 |
-
"epoch": 0.
|
| 666 |
-
"grad_norm":
|
| 667 |
-
"learning_rate":
|
| 668 |
-
"loss":
|
| 669 |
"step": 8400
|
| 670 |
},
|
| 671 |
{
|
| 672 |
-
"epoch": 0.
|
| 673 |
-
"grad_norm":
|
| 674 |
-
"learning_rate":
|
| 675 |
-
"loss":
|
| 676 |
"step": 8500
|
| 677 |
},
|
| 678 |
{
|
| 679 |
-
"epoch": 0.
|
| 680 |
-
"grad_norm":
|
| 681 |
-
"learning_rate":
|
| 682 |
-
"loss":
|
| 683 |
"step": 8600
|
| 684 |
},
|
| 685 |
{
|
| 686 |
-
"epoch": 0.
|
| 687 |
-
"grad_norm":
|
| 688 |
-
"learning_rate":
|
| 689 |
-
"loss":
|
| 690 |
"step": 8700
|
| 691 |
},
|
| 692 |
{
|
| 693 |
-
"epoch": 0.
|
| 694 |
-
"grad_norm":
|
| 695 |
-
"learning_rate":
|
| 696 |
-
"loss":
|
| 697 |
"step": 8800
|
| 698 |
},
|
| 699 |
{
|
| 700 |
-
"epoch": 0.
|
| 701 |
-
"grad_norm": 1.
|
| 702 |
-
"learning_rate":
|
| 703 |
-
"loss":
|
| 704 |
"step": 8900
|
| 705 |
},
|
| 706 |
{
|
| 707 |
-
"epoch": 0.
|
| 708 |
-
"grad_norm":
|
| 709 |
-
"learning_rate":
|
| 710 |
-
"loss":
|
| 711 |
"step": 9000
|
| 712 |
},
|
| 713 |
{
|
| 714 |
-
"epoch": 0.
|
| 715 |
-
"eval_loss": 1.
|
| 716 |
-
"eval_runtime":
|
| 717 |
-
"eval_samples_per_second":
|
| 718 |
-
"eval_steps_per_second": 2.
|
| 719 |
"step": 9000
|
| 720 |
},
|
| 721 |
{
|
| 722 |
-
"epoch": 0.
|
| 723 |
-
"grad_norm":
|
| 724 |
-
"learning_rate":
|
| 725 |
-
"loss":
|
| 726 |
"step": 9100
|
| 727 |
},
|
| 728 |
{
|
| 729 |
-
"epoch": 0.
|
| 730 |
-
"grad_norm":
|
| 731 |
-
"learning_rate":
|
| 732 |
-
"loss":
|
| 733 |
"step": 9200
|
| 734 |
},
|
| 735 |
{
|
| 736 |
-
"epoch": 0.
|
| 737 |
-
"grad_norm": 1.
|
| 738 |
-
"learning_rate":
|
| 739 |
-
"loss":
|
| 740 |
"step": 9300
|
| 741 |
},
|
| 742 |
{
|
| 743 |
-
"epoch": 0.
|
| 744 |
-
"grad_norm":
|
| 745 |
-
"learning_rate":
|
| 746 |
-
"loss":
|
| 747 |
"step": 9400
|
| 748 |
},
|
| 749 |
{
|
| 750 |
-
"epoch": 0.
|
| 751 |
-
"grad_norm":
|
| 752 |
-
"learning_rate":
|
| 753 |
-
"loss":
|
| 754 |
"step": 9500
|
| 755 |
},
|
| 756 |
{
|
| 757 |
-
"epoch": 0.
|
| 758 |
-
"grad_norm":
|
| 759 |
-
"learning_rate":
|
| 760 |
-
"loss":
|
| 761 |
"step": 9600
|
| 762 |
},
|
| 763 |
{
|
| 764 |
-
"epoch": 0.
|
| 765 |
-
"grad_norm":
|
| 766 |
-
"learning_rate":
|
| 767 |
-
"loss":
|
| 768 |
"step": 9700
|
| 769 |
},
|
| 770 |
{
|
| 771 |
-
"epoch": 0.
|
| 772 |
-
"grad_norm":
|
| 773 |
-
"learning_rate":
|
| 774 |
-
"loss":
|
| 775 |
"step": 9800
|
| 776 |
},
|
| 777 |
{
|
| 778 |
-
"epoch": 0.
|
| 779 |
-
"grad_norm": 0.
|
| 780 |
-
"learning_rate":
|
| 781 |
-
"loss":
|
| 782 |
"step": 9900
|
| 783 |
},
|
| 784 |
{
|
| 785 |
-
"epoch": 0.
|
| 786 |
-
"grad_norm":
|
| 787 |
-
"learning_rate":
|
| 788 |
-
"loss":
|
| 789 |
"step": 10000
|
| 790 |
},
|
| 791 |
{
|
| 792 |
-
"epoch": 0.
|
| 793 |
-
"eval_loss": 1.
|
| 794 |
-
"eval_runtime":
|
| 795 |
-
"eval_samples_per_second":
|
| 796 |
-
"eval_steps_per_second": 2.
|
| 797 |
"step": 10000
|
| 798 |
},
|
| 799 |
{
|
| 800 |
-
"epoch": 0.
|
| 801 |
-
"grad_norm":
|
| 802 |
-
"learning_rate":
|
| 803 |
-
"loss":
|
| 804 |
"step": 10100
|
| 805 |
},
|
| 806 |
{
|
| 807 |
-
"epoch": 0.
|
| 808 |
-
"grad_norm":
|
| 809 |
-
"learning_rate":
|
| 810 |
-
"loss":
|
| 811 |
"step": 10200
|
| 812 |
},
|
| 813 |
{
|
| 814 |
-
"epoch": 0.
|
| 815 |
-
"grad_norm":
|
| 816 |
-
"learning_rate":
|
| 817 |
-
"loss":
|
| 818 |
"step": 10300
|
| 819 |
},
|
| 820 |
{
|
| 821 |
-
"epoch": 0.
|
| 822 |
-
"grad_norm": 1.
|
| 823 |
-
"learning_rate":
|
| 824 |
-
"loss":
|
| 825 |
"step": 10400
|
| 826 |
},
|
| 827 |
{
|
| 828 |
-
"epoch": 0.
|
| 829 |
-
"grad_norm":
|
| 830 |
-
"learning_rate":
|
| 831 |
-
"loss":
|
| 832 |
"step": 10500
|
| 833 |
},
|
| 834 |
{
|
| 835 |
-
"epoch": 0.
|
| 836 |
-
"grad_norm":
|
| 837 |
-
"learning_rate":
|
| 838 |
-
"loss":
|
| 839 |
"step": 10600
|
| 840 |
},
|
| 841 |
{
|
| 842 |
-
"epoch": 0.
|
| 843 |
-
"grad_norm":
|
| 844 |
-
"learning_rate":
|
| 845 |
-
"loss":
|
| 846 |
"step": 10700
|
| 847 |
},
|
| 848 |
{
|
| 849 |
-
"epoch": 0.
|
| 850 |
-
"grad_norm":
|
| 851 |
-
"learning_rate":
|
| 852 |
-
"loss":
|
| 853 |
"step": 10800
|
| 854 |
},
|
| 855 |
{
|
| 856 |
-
"epoch": 0.
|
| 857 |
-
"grad_norm":
|
| 858 |
-
"learning_rate":
|
| 859 |
-
"loss":
|
| 860 |
"step": 10900
|
| 861 |
},
|
| 862 |
{
|
| 863 |
-
"epoch": 0.
|
| 864 |
-
"grad_norm": 1.
|
| 865 |
-
"learning_rate":
|
| 866 |
-
"loss":
|
| 867 |
"step": 11000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
-
"epoch": 0.
|
| 871 |
-
"eval_loss": 1.
|
| 872 |
-
"eval_runtime":
|
| 873 |
-
"eval_samples_per_second":
|
| 874 |
-
"eval_steps_per_second": 2.
|
| 875 |
"step": 11000
|
| 876 |
},
|
| 877 |
{
|
| 878 |
-
"epoch": 0.
|
| 879 |
-
"grad_norm":
|
| 880 |
-
"learning_rate":
|
| 881 |
-
"loss":
|
| 882 |
"step": 11100
|
| 883 |
},
|
| 884 |
{
|
| 885 |
-
"epoch": 0.
|
| 886 |
-
"grad_norm": 1.
|
| 887 |
-
"learning_rate":
|
| 888 |
-
"loss":
|
| 889 |
"step": 11200
|
| 890 |
},
|
| 891 |
{
|
| 892 |
-
"epoch": 0.
|
| 893 |
-
"grad_norm":
|
| 894 |
-
"learning_rate":
|
| 895 |
-
"loss":
|
| 896 |
"step": 11300
|
| 897 |
},
|
| 898 |
{
|
| 899 |
-
"epoch": 0.
|
| 900 |
-
"grad_norm": 1.
|
| 901 |
-
"learning_rate":
|
| 902 |
-
"loss":
|
| 903 |
"step": 11400
|
| 904 |
},
|
| 905 |
{
|
| 906 |
-
"epoch": 0.
|
| 907 |
-
"grad_norm":
|
| 908 |
-
"learning_rate":
|
| 909 |
-
"loss":
|
| 910 |
"step": 11500
|
| 911 |
},
|
| 912 |
{
|
| 913 |
-
"epoch": 0.
|
| 914 |
-
"grad_norm":
|
| 915 |
-
"learning_rate":
|
| 916 |
-
"loss":
|
| 917 |
"step": 11600
|
| 918 |
},
|
| 919 |
{
|
| 920 |
-
"epoch": 0.
|
| 921 |
-
"grad_norm":
|
| 922 |
-
"learning_rate":
|
| 923 |
-
"loss":
|
| 924 |
"step": 11700
|
| 925 |
},
|
| 926 |
{
|
| 927 |
-
"epoch": 0.
|
| 928 |
-
"grad_norm":
|
| 929 |
-
"learning_rate":
|
| 930 |
-
"loss":
|
| 931 |
"step": 11800
|
| 932 |
},
|
| 933 |
{
|
| 934 |
-
"epoch": 0.
|
| 935 |
-
"grad_norm":
|
| 936 |
-
"learning_rate":
|
| 937 |
-
"loss":
|
| 938 |
"step": 11900
|
| 939 |
},
|
| 940 |
{
|
| 941 |
-
"epoch": 0.
|
| 942 |
-
"grad_norm":
|
| 943 |
-
"learning_rate":
|
| 944 |
-
"loss":
|
| 945 |
"step": 12000
|
| 946 |
},
|
| 947 |
{
|
| 948 |
-
"epoch": 0.
|
| 949 |
-
"eval_loss": 1.
|
| 950 |
-
"eval_runtime":
|
| 951 |
-
"eval_samples_per_second":
|
| 952 |
-
"eval_steps_per_second": 2.
|
| 953 |
"step": 12000
|
| 954 |
},
|
| 955 |
{
|
| 956 |
-
"epoch": 0.
|
| 957 |
-
"grad_norm":
|
| 958 |
-
"learning_rate":
|
| 959 |
-
"loss":
|
| 960 |
"step": 12100
|
| 961 |
},
|
| 962 |
{
|
| 963 |
-
"epoch": 0.
|
| 964 |
-
"grad_norm":
|
| 965 |
-
"learning_rate":
|
| 966 |
-
"loss":
|
| 967 |
"step": 12200
|
| 968 |
},
|
| 969 |
{
|
| 970 |
-
"epoch": 0.
|
| 971 |
-
"grad_norm": 1.
|
| 972 |
-
"learning_rate":
|
| 973 |
-
"loss":
|
| 974 |
"step": 12300
|
| 975 |
},
|
| 976 |
{
|
| 977 |
-
"epoch": 0.
|
| 978 |
-
"grad_norm":
|
| 979 |
-
"learning_rate":
|
| 980 |
-
"loss":
|
| 981 |
"step": 12400
|
| 982 |
},
|
| 983 |
{
|
| 984 |
-
"epoch": 0.
|
| 985 |
-
"grad_norm":
|
| 986 |
-
"learning_rate":
|
| 987 |
-
"loss":
|
| 988 |
"step": 12500
|
| 989 |
},
|
| 990 |
{
|
| 991 |
-
"epoch": 0.
|
| 992 |
-
"grad_norm":
|
| 993 |
-
"learning_rate":
|
| 994 |
-
"loss":
|
| 995 |
"step": 12600
|
| 996 |
},
|
| 997 |
{
|
| 998 |
-
"epoch": 0.
|
| 999 |
-
"grad_norm":
|
| 1000 |
-
"learning_rate":
|
| 1001 |
-
"loss":
|
| 1002 |
"step": 12700
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
-
"epoch": 0.
|
| 1006 |
-
"grad_norm":
|
| 1007 |
-
"learning_rate":
|
| 1008 |
-
"loss":
|
| 1009 |
"step": 12800
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
-
"epoch": 0.
|
| 1013 |
-
"grad_norm":
|
| 1014 |
-
"learning_rate":
|
| 1015 |
-
"loss":
|
| 1016 |
"step": 12900
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
-
"epoch": 0.
|
| 1020 |
-
"grad_norm":
|
| 1021 |
-
"learning_rate":
|
| 1022 |
-
"loss":
|
| 1023 |
"step": 13000
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
-
"epoch": 0.
|
| 1027 |
-
"eval_loss": 1.
|
| 1028 |
-
"eval_runtime":
|
| 1029 |
-
"eval_samples_per_second":
|
| 1030 |
-
"eval_steps_per_second": 2.
|
| 1031 |
"step": 13000
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
-
"epoch": 0.
|
| 1035 |
-
"grad_norm": 1.
|
| 1036 |
-
"learning_rate":
|
| 1037 |
-
"loss":
|
| 1038 |
"step": 13100
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
-
"epoch": 0.
|
| 1042 |
-
"grad_norm":
|
| 1043 |
-
"learning_rate":
|
| 1044 |
-
"loss":
|
| 1045 |
"step": 13200
|
| 1046 |
},
|
| 1047 |
{
|
| 1048 |
-
"epoch": 0.
|
| 1049 |
-
"grad_norm":
|
| 1050 |
-
"learning_rate":
|
| 1051 |
-
"loss":
|
| 1052 |
"step": 13300
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
-
"epoch": 0.
|
| 1056 |
-
"grad_norm": 1.
|
| 1057 |
-
"learning_rate":
|
| 1058 |
-
"loss":
|
| 1059 |
"step": 13400
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
-
"epoch": 0.
|
| 1063 |
-
"grad_norm":
|
| 1064 |
-
"learning_rate":
|
| 1065 |
-
"loss":
|
| 1066 |
"step": 13500
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
-
"epoch": 0.
|
| 1070 |
-
"grad_norm":
|
| 1071 |
-
"learning_rate":
|
| 1072 |
-
"loss":
|
| 1073 |
"step": 13600
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
-
"epoch": 0.
|
| 1077 |
-
"grad_norm":
|
| 1078 |
-
"learning_rate":
|
| 1079 |
-
"loss":
|
| 1080 |
"step": 13700
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
-
"epoch": 0.
|
| 1084 |
-
"grad_norm":
|
| 1085 |
-
"learning_rate":
|
| 1086 |
-
"loss":
|
| 1087 |
"step": 13800
|
| 1088 |
},
|
| 1089 |
{
|
| 1090 |
-
"epoch": 0.
|
| 1091 |
-
"grad_norm":
|
| 1092 |
-
"learning_rate":
|
| 1093 |
-
"loss":
|
| 1094 |
"step": 13900
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
-
"epoch": 0.
|
| 1098 |
-
"grad_norm":
|
| 1099 |
-
"learning_rate":
|
| 1100 |
-
"loss":
|
| 1101 |
"step": 14000
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
-
"epoch": 0.
|
| 1105 |
-
"eval_loss": 1.
|
| 1106 |
-
"eval_runtime":
|
| 1107 |
-
"eval_samples_per_second":
|
| 1108 |
-
"eval_steps_per_second": 2.
|
| 1109 |
"step": 14000
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
-
"epoch": 0.
|
| 1113 |
-
"grad_norm": 1.
|
| 1114 |
-
"learning_rate":
|
| 1115 |
-
"loss":
|
| 1116 |
"step": 14100
|
| 1117 |
},
|
| 1118 |
{
|
| 1119 |
-
"epoch": 0.
|
| 1120 |
-
"grad_norm":
|
| 1121 |
-
"learning_rate":
|
| 1122 |
-
"loss":
|
| 1123 |
"step": 14200
|
| 1124 |
},
|
| 1125 |
{
|
| 1126 |
-
"epoch": 0.
|
| 1127 |
-
"grad_norm":
|
| 1128 |
-
"learning_rate":
|
| 1129 |
-
"loss":
|
| 1130 |
"step": 14300
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
-
"epoch": 0.
|
| 1134 |
-
"grad_norm":
|
| 1135 |
-
"learning_rate":
|
| 1136 |
-
"loss":
|
| 1137 |
"step": 14400
|
| 1138 |
},
|
| 1139 |
{
|
| 1140 |
-
"epoch": 0.
|
| 1141 |
-
"grad_norm":
|
| 1142 |
-
"learning_rate":
|
| 1143 |
-
"loss":
|
| 1144 |
"step": 14500
|
| 1145 |
},
|
| 1146 |
{
|
| 1147 |
-
"epoch": 0.
|
| 1148 |
-
"grad_norm":
|
| 1149 |
-
"learning_rate":
|
| 1150 |
-
"loss":
|
| 1151 |
"step": 14600
|
| 1152 |
},
|
| 1153 |
{
|
| 1154 |
-
"epoch": 0.
|
| 1155 |
-
"grad_norm":
|
| 1156 |
-
"learning_rate":
|
| 1157 |
-
"loss":
|
| 1158 |
"step": 14700
|
| 1159 |
},
|
| 1160 |
{
|
| 1161 |
-
"epoch": 0.
|
| 1162 |
-
"grad_norm": 1.
|
| 1163 |
-
"learning_rate":
|
| 1164 |
-
"loss":
|
| 1165 |
"step": 14800
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
-
"epoch": 0.
|
| 1169 |
-
"grad_norm": 1.
|
| 1170 |
-
"learning_rate":
|
| 1171 |
-
"loss":
|
| 1172 |
"step": 14900
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
-
"epoch": 0.
|
| 1176 |
-
"grad_norm":
|
| 1177 |
-
"learning_rate":
|
| 1178 |
-
"loss":
|
| 1179 |
"step": 15000
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
-
"epoch": 0.
|
| 1183 |
-
"eval_loss": 1.
|
| 1184 |
-
"eval_runtime":
|
| 1185 |
-
"eval_samples_per_second":
|
| 1186 |
-
"eval_steps_per_second": 2.
|
| 1187 |
"step": 15000
|
| 1188 |
},
|
| 1189 |
{
|
| 1190 |
-
"epoch": 0.
|
| 1191 |
-
"grad_norm":
|
| 1192 |
-
"learning_rate":
|
| 1193 |
-
"loss":
|
| 1194 |
"step": 15100
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
-
"epoch": 0.
|
| 1198 |
-
"grad_norm":
|
| 1199 |
-
"learning_rate":
|
| 1200 |
-
"loss":
|
| 1201 |
"step": 15200
|
| 1202 |
},
|
| 1203 |
{
|
| 1204 |
-
"epoch": 0.
|
| 1205 |
-
"grad_norm":
|
| 1206 |
-
"learning_rate":
|
| 1207 |
-
"loss":
|
| 1208 |
"step": 15300
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
-
"epoch": 0.
|
| 1212 |
-
"grad_norm":
|
| 1213 |
-
"learning_rate":
|
| 1214 |
-
"loss":
|
| 1215 |
"step": 15400
|
| 1216 |
},
|
| 1217 |
{
|
| 1218 |
-
"epoch": 0.
|
| 1219 |
-
"grad_norm":
|
| 1220 |
-
"learning_rate":
|
| 1221 |
-
"loss":
|
| 1222 |
"step": 15500
|
| 1223 |
},
|
| 1224 |
{
|
| 1225 |
-
"epoch": 0.
|
| 1226 |
-
"grad_norm":
|
| 1227 |
-
"learning_rate":
|
| 1228 |
-
"loss":
|
| 1229 |
"step": 15600
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
-
"epoch": 0.
|
| 1233 |
-
"grad_norm":
|
| 1234 |
-
"learning_rate":
|
| 1235 |
-
"loss":
|
| 1236 |
"step": 15700
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
-
"epoch": 0.
|
| 1240 |
-
"grad_norm":
|
| 1241 |
-
"learning_rate":
|
| 1242 |
-
"loss":
|
| 1243 |
"step": 15800
|
| 1244 |
},
|
| 1245 |
{
|
| 1246 |
-
"epoch": 0.
|
| 1247 |
-
"grad_norm": 1.
|
| 1248 |
-
"learning_rate":
|
| 1249 |
-
"loss":
|
| 1250 |
"step": 15900
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
-
"epoch": 0.
|
| 1254 |
-
"grad_norm":
|
| 1255 |
-
"learning_rate":
|
| 1256 |
-
"loss":
|
| 1257 |
"step": 16000
|
| 1258 |
},
|
| 1259 |
{
|
| 1260 |
-
"epoch": 0.
|
| 1261 |
-
"eval_loss": 1.
|
| 1262 |
-
"eval_runtime":
|
| 1263 |
-
"eval_samples_per_second":
|
| 1264 |
-
"eval_steps_per_second": 2.
|
| 1265 |
"step": 16000
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
-
"epoch": 0.
|
| 1269 |
-
"grad_norm":
|
| 1270 |
-
"learning_rate":
|
| 1271 |
-
"loss":
|
| 1272 |
"step": 16100
|
| 1273 |
},
|
| 1274 |
{
|
| 1275 |
-
"epoch": 0.
|
| 1276 |
-
"grad_norm":
|
| 1277 |
-
"learning_rate":
|
| 1278 |
-
"loss":
|
| 1279 |
"step": 16200
|
| 1280 |
},
|
| 1281 |
{
|
| 1282 |
-
"epoch": 0.
|
| 1283 |
-
"grad_norm":
|
| 1284 |
-
"learning_rate":
|
| 1285 |
-
"loss":
|
| 1286 |
"step": 16300
|
| 1287 |
},
|
| 1288 |
{
|
| 1289 |
-
"epoch": 0.
|
| 1290 |
-
"grad_norm":
|
| 1291 |
-
"learning_rate":
|
| 1292 |
-
"loss":
|
| 1293 |
"step": 16400
|
| 1294 |
},
|
| 1295 |
{
|
| 1296 |
-
"epoch": 0.
|
| 1297 |
-
"grad_norm":
|
| 1298 |
-
"learning_rate":
|
| 1299 |
-
"loss":
|
| 1300 |
"step": 16500
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
-
"epoch": 0.
|
| 1304 |
-
"grad_norm":
|
| 1305 |
-
"learning_rate":
|
| 1306 |
-
"loss":
|
| 1307 |
"step": 16600
|
| 1308 |
},
|
| 1309 |
{
|
| 1310 |
-
"epoch": 0.
|
| 1311 |
-
"grad_norm":
|
| 1312 |
-
"learning_rate":
|
| 1313 |
-
"loss":
|
| 1314 |
"step": 16700
|
| 1315 |
},
|
| 1316 |
{
|
| 1317 |
-
"epoch": 0.
|
| 1318 |
-
"grad_norm":
|
| 1319 |
-
"learning_rate":
|
| 1320 |
-
"loss":
|
| 1321 |
"step": 16800
|
| 1322 |
},
|
| 1323 |
{
|
| 1324 |
-
"epoch": 0.
|
| 1325 |
-
"grad_norm":
|
| 1326 |
-
"learning_rate":
|
| 1327 |
-
"loss":
|
| 1328 |
"step": 16900
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
-
"epoch": 0.
|
| 1332 |
-
"grad_norm":
|
| 1333 |
-
"learning_rate":
|
| 1334 |
-
"loss":
|
| 1335 |
"step": 17000
|
| 1336 |
},
|
| 1337 |
{
|
| 1338 |
-
"epoch": 0.
|
| 1339 |
-
"eval_loss": 1.
|
| 1340 |
-
"eval_runtime":
|
| 1341 |
-
"eval_samples_per_second":
|
| 1342 |
-
"eval_steps_per_second": 2.
|
| 1343 |
"step": 17000
|
| 1344 |
},
|
| 1345 |
{
|
| 1346 |
-
"epoch": 0.
|
| 1347 |
-
"grad_norm":
|
| 1348 |
-
"learning_rate":
|
| 1349 |
-
"loss":
|
| 1350 |
"step": 17100
|
| 1351 |
},
|
| 1352 |
{
|
| 1353 |
-
"epoch": 0.
|
| 1354 |
-
"grad_norm":
|
| 1355 |
-
"learning_rate":
|
| 1356 |
-
"loss":
|
| 1357 |
"step": 17200
|
| 1358 |
},
|
| 1359 |
{
|
| 1360 |
-
"epoch": 0.
|
| 1361 |
-
"grad_norm":
|
| 1362 |
-
"learning_rate":
|
| 1363 |
-
"loss":
|
| 1364 |
"step": 17300
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
-
"epoch": 0.
|
| 1368 |
-
"grad_norm":
|
| 1369 |
-
"learning_rate":
|
| 1370 |
-
"loss":
|
| 1371 |
"step": 17400
|
| 1372 |
},
|
| 1373 |
{
|
| 1374 |
-
"epoch": 0.
|
| 1375 |
-
"grad_norm":
|
| 1376 |
-
"learning_rate":
|
| 1377 |
-
"loss":
|
| 1378 |
"step": 17500
|
| 1379 |
},
|
| 1380 |
{
|
| 1381 |
-
"epoch": 0.
|
| 1382 |
-
"grad_norm":
|
| 1383 |
-
"learning_rate":
|
| 1384 |
-
"loss":
|
| 1385 |
"step": 17600
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
-
"epoch": 0.
|
| 1389 |
-
"grad_norm":
|
| 1390 |
-
"learning_rate":
|
| 1391 |
-
"loss":
|
| 1392 |
"step": 17700
|
| 1393 |
},
|
| 1394 |
{
|
| 1395 |
-
"epoch": 0.
|
| 1396 |
-
"grad_norm":
|
| 1397 |
-
"learning_rate":
|
| 1398 |
-
"loss":
|
| 1399 |
"step": 17800
|
| 1400 |
},
|
| 1401 |
{
|
| 1402 |
-
"epoch": 0.
|
| 1403 |
-
"grad_norm":
|
| 1404 |
-
"learning_rate":
|
| 1405 |
-
"loss":
|
| 1406 |
"step": 17900
|
| 1407 |
},
|
| 1408 |
{
|
| 1409 |
-
"epoch": 0.
|
| 1410 |
-
"grad_norm":
|
| 1411 |
-
"learning_rate":
|
| 1412 |
-
"loss":
|
| 1413 |
"step": 18000
|
| 1414 |
},
|
| 1415 |
{
|
| 1416 |
-
"epoch": 0.
|
| 1417 |
-
"eval_loss": 1.
|
| 1418 |
-
"eval_runtime":
|
| 1419 |
-
"eval_samples_per_second":
|
| 1420 |
-
"eval_steps_per_second": 2.
|
| 1421 |
"step": 18000
|
| 1422 |
},
|
| 1423 |
{
|
| 1424 |
-
"epoch": 0.
|
| 1425 |
-
"grad_norm":
|
| 1426 |
-
"learning_rate":
|
| 1427 |
-
"loss":
|
| 1428 |
"step": 18100
|
| 1429 |
},
|
| 1430 |
{
|
| 1431 |
-
"epoch": 0.
|
| 1432 |
-
"grad_norm": 1.
|
| 1433 |
-
"learning_rate":
|
| 1434 |
-
"loss":
|
| 1435 |
"step": 18200
|
| 1436 |
},
|
| 1437 |
{
|
| 1438 |
-
"epoch": 0.
|
| 1439 |
-
"grad_norm":
|
| 1440 |
-
"learning_rate":
|
| 1441 |
-
"loss":
|
| 1442 |
"step": 18300
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
-
"epoch": 0.
|
| 1446 |
-
"grad_norm":
|
| 1447 |
-
"learning_rate":
|
| 1448 |
-
"loss":
|
| 1449 |
"step": 18400
|
| 1450 |
},
|
| 1451 |
{
|
| 1452 |
-
"epoch": 0.
|
| 1453 |
-
"grad_norm":
|
| 1454 |
-
"learning_rate":
|
| 1455 |
-
"loss":
|
| 1456 |
"step": 18500
|
| 1457 |
},
|
| 1458 |
{
|
| 1459 |
-
"epoch": 0.
|
| 1460 |
-
"grad_norm":
|
| 1461 |
-
"learning_rate":
|
| 1462 |
-
"loss":
|
| 1463 |
"step": 18600
|
| 1464 |
},
|
| 1465 |
{
|
| 1466 |
-
"epoch": 0.
|
| 1467 |
-
"grad_norm":
|
| 1468 |
-
"learning_rate":
|
| 1469 |
-
"loss":
|
| 1470 |
"step": 18700
|
| 1471 |
},
|
| 1472 |
{
|
| 1473 |
-
"epoch": 0.
|
| 1474 |
-
"grad_norm":
|
| 1475 |
-
"learning_rate":
|
| 1476 |
-
"loss":
|
| 1477 |
"step": 18800
|
| 1478 |
},
|
| 1479 |
{
|
| 1480 |
-
"epoch": 0.
|
| 1481 |
-
"grad_norm":
|
| 1482 |
-
"learning_rate":
|
| 1483 |
-
"loss":
|
| 1484 |
"step": 18900
|
| 1485 |
},
|
| 1486 |
{
|
| 1487 |
-
"epoch": 0.
|
| 1488 |
-
"grad_norm":
|
| 1489 |
-
"learning_rate":
|
| 1490 |
-
"loss":
|
| 1491 |
"step": 19000
|
| 1492 |
},
|
| 1493 |
{
|
| 1494 |
-
"epoch": 0.
|
| 1495 |
-
"eval_loss": 1.
|
| 1496 |
-
"eval_runtime":
|
| 1497 |
-
"eval_samples_per_second":
|
| 1498 |
-
"eval_steps_per_second": 2.
|
| 1499 |
"step": 19000
|
| 1500 |
},
|
| 1501 |
{
|
| 1502 |
-
"epoch": 0.
|
| 1503 |
-
"grad_norm":
|
| 1504 |
-
"learning_rate":
|
| 1505 |
-
"loss":
|
| 1506 |
"step": 19100
|
| 1507 |
},
|
| 1508 |
{
|
| 1509 |
-
"epoch": 0.
|
| 1510 |
-
"grad_norm":
|
| 1511 |
-
"learning_rate":
|
| 1512 |
-
"loss":
|
| 1513 |
"step": 19200
|
| 1514 |
},
|
| 1515 |
{
|
| 1516 |
-
"epoch": 0.
|
| 1517 |
-
"grad_norm":
|
| 1518 |
-
"learning_rate":
|
| 1519 |
-
"loss":
|
| 1520 |
"step": 19300
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
-
"epoch": 0.
|
| 1524 |
-
"grad_norm":
|
| 1525 |
-
"learning_rate":
|
| 1526 |
-
"loss":
|
| 1527 |
"step": 19400
|
| 1528 |
},
|
| 1529 |
{
|
| 1530 |
-
"epoch": 0.
|
| 1531 |
-
"grad_norm":
|
| 1532 |
-
"learning_rate": 2.
|
| 1533 |
-
"loss":
|
| 1534 |
"step": 19500
|
| 1535 |
},
|
| 1536 |
{
|
| 1537 |
-
"epoch": 0.
|
| 1538 |
-
"grad_norm":
|
| 1539 |
-
"learning_rate": 2.
|
| 1540 |
-
"loss":
|
| 1541 |
"step": 19600
|
| 1542 |
},
|
| 1543 |
{
|
| 1544 |
-
"epoch": 0.
|
| 1545 |
-
"grad_norm":
|
| 1546 |
-
"learning_rate": 2.
|
| 1547 |
-
"loss":
|
| 1548 |
"step": 19700
|
| 1549 |
},
|
| 1550 |
{
|
| 1551 |
-
"epoch": 0.
|
| 1552 |
-
"grad_norm":
|
| 1553 |
-
"learning_rate": 2.
|
| 1554 |
-
"loss":
|
| 1555 |
"step": 19800
|
| 1556 |
},
|
| 1557 |
{
|
| 1558 |
-
"epoch": 0.
|
| 1559 |
-
"grad_norm":
|
| 1560 |
-
"learning_rate":
|
| 1561 |
-
"loss":
|
| 1562 |
"step": 19900
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
-
"epoch": 0.
|
| 1566 |
-
"grad_norm":
|
| 1567 |
-
"learning_rate":
|
| 1568 |
-
"loss":
|
| 1569 |
"step": 20000
|
| 1570 |
},
|
| 1571 |
{
|
| 1572 |
-
"epoch": 0.
|
| 1573 |
-
"eval_loss": 1.
|
| 1574 |
-
"eval_runtime":
|
| 1575 |
-
"eval_samples_per_second":
|
| 1576 |
-
"eval_steps_per_second": 2.
|
| 1577 |
"step": 20000
|
| 1578 |
},
|
| 1579 |
{
|
| 1580 |
-
"epoch": 0.
|
| 1581 |
-
"grad_norm":
|
| 1582 |
-
"learning_rate":
|
| 1583 |
-
"loss":
|
| 1584 |
"step": 20100
|
| 1585 |
},
|
| 1586 |
{
|
| 1587 |
-
"epoch": 0.
|
| 1588 |
-
"grad_norm":
|
| 1589 |
-
"learning_rate":
|
| 1590 |
-
"loss":
|
| 1591 |
"step": 20200
|
| 1592 |
},
|
| 1593 |
{
|
| 1594 |
-
"epoch": 0.
|
| 1595 |
-
"grad_norm":
|
| 1596 |
-
"learning_rate":
|
| 1597 |
-
"loss":
|
| 1598 |
"step": 20300
|
| 1599 |
},
|
| 1600 |
{
|
| 1601 |
-
"epoch": 0.
|
| 1602 |
-
"grad_norm":
|
| 1603 |
-
"learning_rate":
|
| 1604 |
-
"loss":
|
| 1605 |
"step": 20400
|
| 1606 |
},
|
| 1607 |
{
|
| 1608 |
-
"epoch": 0.
|
| 1609 |
-
"grad_norm":
|
| 1610 |
-
"learning_rate": 1.
|
| 1611 |
-
"loss":
|
| 1612 |
"step": 20500
|
| 1613 |
},
|
| 1614 |
{
|
| 1615 |
-
"epoch": 0.
|
| 1616 |
-
"grad_norm":
|
| 1617 |
-
"learning_rate": 1.
|
| 1618 |
-
"loss":
|
| 1619 |
"step": 20600
|
| 1620 |
},
|
| 1621 |
{
|
| 1622 |
-
"epoch": 0.
|
| 1623 |
-
"grad_norm":
|
| 1624 |
-
"learning_rate": 1.
|
| 1625 |
-
"loss":
|
| 1626 |
"step": 20700
|
| 1627 |
},
|
| 1628 |
{
|
| 1629 |
-
"epoch": 0.
|
| 1630 |
-
"grad_norm":
|
| 1631 |
-
"learning_rate": 1.
|
| 1632 |
-
"loss":
|
| 1633 |
"step": 20800
|
| 1634 |
},
|
| 1635 |
{
|
| 1636 |
-
"epoch": 0.
|
| 1637 |
-
"grad_norm":
|
| 1638 |
-
"learning_rate": 1.
|
| 1639 |
-
"loss":
|
| 1640 |
"step": 20900
|
| 1641 |
},
|
| 1642 |
{
|
| 1643 |
-
"epoch": 0.
|
| 1644 |
-
"grad_norm":
|
| 1645 |
-
"learning_rate": 1.
|
| 1646 |
-
"loss":
|
| 1647 |
"step": 21000
|
| 1648 |
},
|
| 1649 |
{
|
| 1650 |
-
"epoch": 0.
|
| 1651 |
-
"eval_loss": 1.
|
| 1652 |
-
"eval_runtime":
|
| 1653 |
-
"eval_samples_per_second":
|
| 1654 |
-
"eval_steps_per_second": 2.
|
| 1655 |
"step": 21000
|
| 1656 |
},
|
| 1657 |
{
|
| 1658 |
-
"epoch": 0.
|
| 1659 |
-
"grad_norm":
|
| 1660 |
-
"learning_rate": 1.
|
| 1661 |
-
"loss":
|
| 1662 |
"step": 21100
|
| 1663 |
},
|
| 1664 |
{
|
| 1665 |
-
"epoch": 0.
|
| 1666 |
-
"grad_norm":
|
| 1667 |
-
"learning_rate": 1.
|
| 1668 |
-
"loss":
|
| 1669 |
"step": 21200
|
| 1670 |
},
|
| 1671 |
{
|
| 1672 |
-
"epoch": 0.
|
| 1673 |
-
"grad_norm":
|
| 1674 |
-
"learning_rate": 1.
|
| 1675 |
-
"loss":
|
| 1676 |
"step": 21300
|
| 1677 |
},
|
| 1678 |
{
|
| 1679 |
-
"epoch": 0.
|
| 1680 |
-
"grad_norm":
|
| 1681 |
-
"learning_rate": 1.
|
| 1682 |
-
"loss":
|
| 1683 |
"step": 21400
|
| 1684 |
},
|
| 1685 |
{
|
| 1686 |
-
"epoch": 0.
|
| 1687 |
-
"grad_norm": 0.
|
| 1688 |
-
"learning_rate": 1.
|
| 1689 |
-
"loss":
|
| 1690 |
"step": 21500
|
| 1691 |
},
|
| 1692 |
{
|
| 1693 |
-
"epoch": 0.
|
| 1694 |
-
"grad_norm":
|
| 1695 |
-
"learning_rate": 1.
|
| 1696 |
-
"loss":
|
| 1697 |
"step": 21600
|
| 1698 |
},
|
| 1699 |
{
|
| 1700 |
-
"epoch": 0.
|
| 1701 |
-
"grad_norm":
|
| 1702 |
-
"learning_rate": 1.
|
| 1703 |
-
"loss":
|
| 1704 |
"step": 21700
|
| 1705 |
},
|
| 1706 |
{
|
| 1707 |
-
"epoch": 0.
|
| 1708 |
-
"grad_norm":
|
| 1709 |
-
"learning_rate": 1.
|
| 1710 |
-
"loss":
|
| 1711 |
"step": 21800
|
| 1712 |
},
|
| 1713 |
{
|
| 1714 |
-
"epoch": 0.
|
| 1715 |
-
"grad_norm":
|
| 1716 |
-
"learning_rate":
|
| 1717 |
-
"loss":
|
| 1718 |
"step": 21900
|
| 1719 |
},
|
| 1720 |
{
|
| 1721 |
-
"epoch": 0.
|
| 1722 |
-
"grad_norm":
|
| 1723 |
-
"learning_rate":
|
| 1724 |
-
"loss":
|
| 1725 |
"step": 22000
|
| 1726 |
},
|
| 1727 |
{
|
| 1728 |
-
"epoch": 0.
|
| 1729 |
-
"eval_loss": 1.
|
| 1730 |
-
"eval_runtime":
|
| 1731 |
-
"eval_samples_per_second":
|
| 1732 |
-
"eval_steps_per_second": 2.
|
| 1733 |
"step": 22000
|
| 1734 |
},
|
| 1735 |
{
|
| 1736 |
-
"epoch": 0.
|
| 1737 |
-
"grad_norm":
|
| 1738 |
-
"learning_rate":
|
| 1739 |
-
"loss":
|
| 1740 |
"step": 22100
|
| 1741 |
},
|
| 1742 |
{
|
| 1743 |
-
"epoch": 0.
|
| 1744 |
-
"grad_norm":
|
| 1745 |
-
"learning_rate":
|
| 1746 |
-
"loss":
|
| 1747 |
"step": 22200
|
| 1748 |
},
|
| 1749 |
{
|
| 1750 |
-
"epoch": 0.
|
| 1751 |
-
"grad_norm":
|
| 1752 |
-
"learning_rate":
|
| 1753 |
-
"loss":
|
| 1754 |
"step": 22300
|
| 1755 |
},
|
| 1756 |
{
|
| 1757 |
-
"epoch": 0.
|
| 1758 |
-
"grad_norm":
|
| 1759 |
-
"learning_rate":
|
| 1760 |
-
"loss":
|
| 1761 |
"step": 22400
|
| 1762 |
},
|
| 1763 |
{
|
| 1764 |
-
"epoch": 0.
|
| 1765 |
-
"grad_norm":
|
| 1766 |
-
"learning_rate":
|
| 1767 |
-
"loss":
|
| 1768 |
"step": 22500
|
| 1769 |
},
|
| 1770 |
{
|
| 1771 |
-
"epoch": 0.
|
| 1772 |
-
"grad_norm":
|
| 1773 |
-
"learning_rate":
|
| 1774 |
-
"loss":
|
| 1775 |
"step": 22600
|
| 1776 |
},
|
| 1777 |
{
|
| 1778 |
-
"epoch": 0.
|
| 1779 |
-
"grad_norm":
|
| 1780 |
-
"learning_rate":
|
| 1781 |
-
"loss":
|
| 1782 |
"step": 22700
|
| 1783 |
},
|
| 1784 |
{
|
| 1785 |
-
"epoch": 0.
|
| 1786 |
-
"grad_norm": 1.
|
| 1787 |
-
"learning_rate":
|
| 1788 |
-
"loss":
|
| 1789 |
"step": 22800
|
| 1790 |
},
|
| 1791 |
{
|
| 1792 |
-
"epoch":
|
| 1793 |
-
"grad_norm":
|
| 1794 |
-
"learning_rate":
|
| 1795 |
-
"loss":
|
| 1796 |
"step": 22900
|
| 1797 |
},
|
| 1798 |
{
|
| 1799 |
-
"epoch":
|
| 1800 |
-
"grad_norm":
|
| 1801 |
-
"learning_rate":
|
| 1802 |
-
"loss":
|
| 1803 |
"step": 23000
|
| 1804 |
},
|
| 1805 |
{
|
| 1806 |
-
"epoch":
|
| 1807 |
-
"eval_loss": 1.
|
| 1808 |
-
"eval_runtime":
|
| 1809 |
-
"eval_samples_per_second":
|
| 1810 |
-
"eval_steps_per_second": 2.
|
| 1811 |
"step": 23000
|
| 1812 |
},
|
| 1813 |
{
|
| 1814 |
-
"epoch":
|
| 1815 |
-
"grad_norm":
|
| 1816 |
-
"learning_rate":
|
| 1817 |
-
"loss":
|
| 1818 |
"step": 23100
|
| 1819 |
},
|
| 1820 |
{
|
| 1821 |
-
"epoch":
|
| 1822 |
-
"grad_norm":
|
| 1823 |
-
"learning_rate":
|
| 1824 |
-
"loss":
|
| 1825 |
"step": 23200
|
| 1826 |
},
|
| 1827 |
{
|
| 1828 |
-
"epoch":
|
| 1829 |
-
"grad_norm":
|
| 1830 |
-
"learning_rate":
|
| 1831 |
-
"loss":
|
| 1832 |
"step": 23300
|
| 1833 |
},
|
| 1834 |
{
|
| 1835 |
-
"epoch":
|
| 1836 |
-
"grad_norm":
|
| 1837 |
-
"learning_rate":
|
| 1838 |
-
"loss":
|
| 1839 |
"step": 23400
|
| 1840 |
},
|
| 1841 |
{
|
| 1842 |
-
"epoch":
|
| 1843 |
-
"grad_norm":
|
| 1844 |
-
"learning_rate":
|
| 1845 |
-
"loss":
|
| 1846 |
"step": 23500
|
| 1847 |
},
|
| 1848 |
{
|
| 1849 |
-
"epoch":
|
| 1850 |
-
"grad_norm":
|
| 1851 |
-
"learning_rate": 1.
|
| 1852 |
-
"loss":
|
| 1853 |
"step": 23600
|
| 1854 |
},
|
| 1855 |
{
|
| 1856 |
-
"epoch":
|
| 1857 |
-
"grad_norm":
|
| 1858 |
-
"learning_rate": 1.
|
| 1859 |
-
"loss":
|
| 1860 |
"step": 23700
|
| 1861 |
},
|
| 1862 |
{
|
| 1863 |
-
"epoch":
|
| 1864 |
-
"grad_norm":
|
| 1865 |
-
"learning_rate": 1.
|
| 1866 |
-
"loss":
|
| 1867 |
"step": 23800
|
| 1868 |
},
|
| 1869 |
{
|
| 1870 |
-
"epoch":
|
| 1871 |
-
"grad_norm":
|
| 1872 |
-
"learning_rate": 1.
|
| 1873 |
-
"loss":
|
| 1874 |
"step": 23900
|
| 1875 |
},
|
| 1876 |
{
|
| 1877 |
-
"epoch":
|
| 1878 |
-
"grad_norm":
|
| 1879 |
-
"learning_rate": 1.
|
| 1880 |
-
"loss":
|
| 1881 |
"step": 24000
|
| 1882 |
},
|
| 1883 |
{
|
| 1884 |
-
"epoch":
|
| 1885 |
-
"eval_loss": 1.
|
| 1886 |
-
"eval_runtime":
|
| 1887 |
-
"eval_samples_per_second":
|
| 1888 |
-
"eval_steps_per_second": 2.
|
| 1889 |
"step": 24000
|
| 1890 |
},
|
| 1891 |
{
|
| 1892 |
-
"epoch":
|
| 1893 |
-
"grad_norm":
|
| 1894 |
-
"learning_rate":
|
| 1895 |
-
"loss":
|
| 1896 |
"step": 24100
|
| 1897 |
},
|
| 1898 |
{
|
| 1899 |
-
"epoch":
|
| 1900 |
-
"grad_norm":
|
| 1901 |
-
"learning_rate":
|
| 1902 |
-
"loss":
|
| 1903 |
"step": 24200
|
| 1904 |
},
|
| 1905 |
{
|
| 1906 |
-
"epoch":
|
| 1907 |
-
"grad_norm":
|
| 1908 |
-
"learning_rate":
|
| 1909 |
-
"loss":
|
| 1910 |
"step": 24300
|
| 1911 |
},
|
| 1912 |
{
|
| 1913 |
-
"epoch":
|
| 1914 |
-
"grad_norm":
|
| 1915 |
-
"learning_rate":
|
| 1916 |
-
"loss":
|
| 1917 |
"step": 24400
|
| 1918 |
},
|
| 1919 |
{
|
| 1920 |
-
"epoch":
|
| 1921 |
-
"grad_norm":
|
| 1922 |
-
"learning_rate":
|
| 1923 |
-
"loss":
|
| 1924 |
"step": 24500
|
| 1925 |
},
|
| 1926 |
{
|
| 1927 |
-
"epoch":
|
| 1928 |
-
"grad_norm":
|
| 1929 |
-
"learning_rate": 1.
|
| 1930 |
-
"loss":
|
| 1931 |
"step": 24600
|
| 1932 |
},
|
| 1933 |
{
|
| 1934 |
-
"epoch":
|
| 1935 |
-
"grad_norm":
|
| 1936 |
-
"learning_rate":
|
| 1937 |
-
"loss":
|
| 1938 |
"step": 24700
|
| 1939 |
},
|
| 1940 |
{
|
| 1941 |
-
"epoch":
|
| 1942 |
-
"grad_norm": 0.
|
| 1943 |
-
"learning_rate":
|
| 1944 |
-
"loss":
|
| 1945 |
"step": 24800
|
| 1946 |
},
|
| 1947 |
{
|
| 1948 |
-
"epoch":
|
| 1949 |
-
"grad_norm":
|
| 1950 |
-
"learning_rate":
|
| 1951 |
-
"loss":
|
| 1952 |
"step": 24900
|
| 1953 |
},
|
| 1954 |
{
|
| 1955 |
-
"epoch":
|
| 1956 |
-
"grad_norm":
|
| 1957 |
-
"learning_rate":
|
| 1958 |
-
"loss":
|
| 1959 |
"step": 25000
|
| 1960 |
},
|
| 1961 |
{
|
| 1962 |
-
"epoch":
|
| 1963 |
-
"eval_loss": 1.
|
| 1964 |
-
"eval_runtime":
|
| 1965 |
-
"eval_samples_per_second":
|
| 1966 |
-
"eval_steps_per_second": 2.
|
| 1967 |
"step": 25000
|
| 1968 |
}
|
| 1969 |
],
|
| 1970 |
"logging_steps": 100,
|
| 1971 |
-
"max_steps":
|
| 1972 |
"num_input_tokens_seen": 0,
|
| 1973 |
"num_train_epochs": 9223372036854775807,
|
| 1974 |
"save_steps": 5000,
|
|
@@ -1979,12 +1979,12 @@
|
|
| 1979 |
"should_evaluate": false,
|
| 1980 |
"should_log": false,
|
| 1981 |
"should_save": true,
|
| 1982 |
-
"should_training_stop":
|
| 1983 |
},
|
| 1984 |
"attributes": {}
|
| 1985 |
}
|
| 1986 |
},
|
| 1987 |
-
"total_flos": 3.
|
| 1988 |
"train_batch_size": 64,
|
| 1989 |
"trial_name": null,
|
| 1990 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.08784,
|
| 6 |
"eval_steps": 1000,
|
| 7 |
"global_step": 25000,
|
| 8 |
"is_hyper_param_search": false,
|
|
|
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"epoch": 4e-05,
|
| 14 |
+
"grad_norm": 0.911555290222168,
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
+
"loss": 0.7505,
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
+
"epoch": 0.004,
|
| 21 |
+
"grad_norm": 1.2557882070541382,
|
| 22 |
+
"learning_rate": 9.9e-07,
|
| 23 |
+
"loss": 0.831,
|
| 24 |
"step": 100
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.008,
|
| 28 |
+
"grad_norm": 0.9086900353431702,
|
| 29 |
+
"learning_rate": 1.9900000000000004e-06,
|
| 30 |
+
"loss": 0.8295,
|
| 31 |
"step": 200
|
| 32 |
},
|
| 33 |
{
|
| 34 |
+
"epoch": 0.012,
|
| 35 |
+
"grad_norm": 0.9221948385238647,
|
| 36 |
+
"learning_rate": 2.4999758220143106e-06,
|
| 37 |
+
"loss": 0.8411,
|
| 38 |
"step": 300
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.016,
|
| 42 |
+
"grad_norm": 0.8809811472892761,
|
| 43 |
+
"learning_rate": 2.4997764426529066e-06,
|
| 44 |
+
"loss": 0.8288,
|
| 45 |
"step": 400
|
| 46 |
},
|
| 47 |
{
|
| 48 |
+
"epoch": 0.02,
|
| 49 |
+
"grad_norm": 1.3145067691802979,
|
| 50 |
+
"learning_rate": 2.499375702067717e-06,
|
| 51 |
+
"loss": 0.8312,
|
| 52 |
"step": 500
|
| 53 |
},
|
| 54 |
{
|
| 55 |
+
"epoch": 0.024,
|
| 56 |
+
"grad_norm": 0.9034631252288818,
|
| 57 |
+
"learning_rate": 2.4987736648251815e-06,
|
| 58 |
+
"loss": 0.8385,
|
| 59 |
"step": 600
|
| 60 |
},
|
| 61 |
{
|
| 62 |
+
"epoch": 0.028,
|
| 63 |
+
"grad_norm": 0.8681179881095886,
|
| 64 |
+
"learning_rate": 2.497970427924213e-06,
|
| 65 |
+
"loss": 0.8175,
|
| 66 |
"step": 700
|
| 67 |
},
|
| 68 |
{
|
| 69 |
+
"epoch": 0.032,
|
| 70 |
+
"grad_norm": 0.9303165674209595,
|
| 71 |
+
"learning_rate": 2.496966120780569e-06,
|
| 72 |
+
"loss": 0.8281,
|
| 73 |
"step": 800
|
| 74 |
},
|
| 75 |
{
|
| 76 |
+
"epoch": 0.036,
|
| 77 |
+
"grad_norm": 0.9573058485984802,
|
| 78 |
+
"learning_rate": 2.4957609052060012e-06,
|
| 79 |
+
"loss": 0.8326,
|
| 80 |
"step": 900
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 0.04,
|
| 84 |
+
"grad_norm": 0.9730055928230286,
|
| 85 |
+
"learning_rate": 2.4943549753821847e-06,
|
| 86 |
+
"loss": 0.8391,
|
| 87 |
"step": 1000
|
| 88 |
},
|
| 89 |
{
|
| 90 |
+
"epoch": 0.04,
|
| 91 |
+
"eval_loss": 1.5264503955841064,
|
| 92 |
+
"eval_runtime": 104.8997,
|
| 93 |
+
"eval_samples_per_second": 130.553,
|
| 94 |
+
"eval_steps_per_second": 2.04,
|
| 95 |
"step": 1000
|
| 96 |
},
|
| 97 |
{
|
| 98 |
+
"epoch": 0.044,
|
| 99 |
+
"grad_norm": 0.8237825632095337,
|
| 100 |
+
"learning_rate": 2.4927485578294313e-06,
|
| 101 |
+
"loss": 0.8176,
|
| 102 |
"step": 1100
|
| 103 |
},
|
| 104 |
{
|
| 105 |
+
"epoch": 0.048,
|
| 106 |
+
"grad_norm": 0.9133234620094299,
|
| 107 |
+
"learning_rate": 2.4909419113701947e-06,
|
| 108 |
+
"loss": 0.8303,
|
| 109 |
"step": 1200
|
| 110 |
},
|
| 111 |
{
|
| 112 |
+
"epoch": 0.052,
|
| 113 |
+
"grad_norm": 0.9377557635307312,
|
| 114 |
+
"learning_rate": 2.4889353270873663e-06,
|
| 115 |
+
"loss": 0.8159,
|
| 116 |
"step": 1300
|
| 117 |
},
|
| 118 |
{
|
| 119 |
+
"epoch": 0.056,
|
| 120 |
+
"grad_norm": 0.9034435749053955,
|
| 121 |
+
"learning_rate": 2.4867291282773805e-06,
|
| 122 |
+
"loss": 0.8145,
|
| 123 |
"step": 1400
|
| 124 |
},
|
| 125 |
{
|
| 126 |
+
"epoch": 0.06,
|
| 127 |
+
"grad_norm": 1.0601003170013428,
|
| 128 |
+
"learning_rate": 2.4843236703981235e-06,
|
| 129 |
+
"loss": 0.8317,
|
| 130 |
"step": 1500
|
| 131 |
},
|
| 132 |
{
|
| 133 |
+
"epoch": 0.064,
|
| 134 |
+
"grad_norm": 0.9157763719558716,
|
| 135 |
+
"learning_rate": 2.481719341011662e-06,
|
| 136 |
+
"loss": 0.8355,
|
| 137 |
"step": 1600
|
| 138 |
},
|
| 139 |
{
|
| 140 |
+
"epoch": 0.068,
|
| 141 |
+
"grad_norm": 0.9011576175689697,
|
| 142 |
+
"learning_rate": 2.4789165597218035e-06,
|
| 143 |
+
"loss": 0.8319,
|
| 144 |
"step": 1700
|
| 145 |
},
|
| 146 |
{
|
| 147 |
+
"epoch": 0.072,
|
| 148 |
+
"grad_norm": 0.8954268097877502,
|
| 149 |
+
"learning_rate": 2.475915778106486e-06,
|
| 150 |
+
"loss": 0.8156,
|
| 151 |
"step": 1800
|
| 152 |
},
|
| 153 |
{
|
| 154 |
+
"epoch": 0.076,
|
| 155 |
+
"grad_norm": 0.8911709189414978,
|
| 156 |
+
"learning_rate": 2.4727174796450266e-06,
|
| 157 |
+
"loss": 0.8365,
|
| 158 |
"step": 1900
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 0.08,
|
| 162 |
+
"grad_norm": 0.9407449960708618,
|
| 163 |
+
"learning_rate": 2.4693221796402166e-06,
|
| 164 |
+
"loss": 0.8288,
|
| 165 |
"step": 2000
|
| 166 |
},
|
| 167 |
{
|
| 168 |
+
"epoch": 0.08,
|
| 169 |
+
"eval_loss": 1.5217734575271606,
|
| 170 |
+
"eval_runtime": 98.2235,
|
| 171 |
+
"eval_samples_per_second": 139.427,
|
| 172 |
+
"eval_steps_per_second": 2.179,
|
| 173 |
"step": 2000
|
| 174 |
},
|
| 175 |
{
|
| 176 |
+
"epoch": 0.084,
|
| 177 |
+
"grad_norm": 0.8769101500511169,
|
| 178 |
+
"learning_rate": 2.4657304251353047e-06,
|
| 179 |
+
"loss": 0.8131,
|
| 180 |
"step": 2100
|
| 181 |
},
|
| 182 |
{
|
| 183 |
+
"epoch": 0.088,
|
| 184 |
+
"grad_norm": 0.8608514070510864,
|
| 185 |
+
"learning_rate": 2.4619427948258547e-06,
|
| 186 |
+
"loss": 0.8088,
|
| 187 |
"step": 2200
|
| 188 |
},
|
| 189 |
{
|
| 190 |
+
"epoch": 0.092,
|
| 191 |
+
"grad_norm": 0.9365686178207397,
|
| 192 |
+
"learning_rate": 2.4579598989665065e-06,
|
| 193 |
+
"loss": 0.8286,
|
| 194 |
"step": 2300
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 0.096,
|
| 198 |
+
"grad_norm": 0.928945779800415,
|
| 199 |
+
"learning_rate": 2.453782379272657e-06,
|
| 200 |
+
"loss": 0.8109,
|
| 201 |
"step": 2400
|
| 202 |
},
|
| 203 |
{
|
| 204 |
+
"epoch": 0.1,
|
| 205 |
+
"grad_norm": 0.9162323474884033,
|
| 206 |
+
"learning_rate": 2.449410908817064e-06,
|
| 207 |
+
"loss": 0.806,
|
| 208 |
"step": 2500
|
| 209 |
},
|
| 210 |
{
|
| 211 |
+
"epoch": 0.104,
|
| 212 |
+
"grad_norm": 0.9436105489730835,
|
| 213 |
+
"learning_rate": 2.444846191921406e-06,
|
| 214 |
+
"loss": 0.7969,
|
| 215 |
"step": 2600
|
| 216 |
},
|
| 217 |
{
|
| 218 |
+
"epoch": 0.108,
|
| 219 |
+
"grad_norm": 0.9459385871887207,
|
| 220 |
+
"learning_rate": 2.4400889640427992e-06,
|
| 221 |
+
"loss": 0.8315,
|
| 222 |
"step": 2700
|
| 223 |
},
|
| 224 |
{
|
| 225 |
+
"epoch": 0.112,
|
| 226 |
+
"grad_norm": 0.9575082063674927,
|
| 227 |
+
"learning_rate": 2.435139991655308e-06,
|
| 228 |
+
"loss": 0.8324,
|
| 229 |
"step": 2800
|
| 230 |
},
|
| 231 |
{
|
| 232 |
+
"epoch": 0.116,
|
| 233 |
+
"grad_norm": 0.927148163318634,
|
| 234 |
+
"learning_rate": 2.4300000721264466e-06,
|
| 235 |
+
"loss": 0.8267,
|
| 236 |
"step": 2900
|
| 237 |
},
|
| 238 |
{
|
| 239 |
+
"epoch": 0.12,
|
| 240 |
+
"grad_norm": 0.9774505496025085,
|
| 241 |
+
"learning_rate": 2.4246700335887123e-06,
|
| 242 |
+
"loss": 0.8262,
|
| 243 |
"step": 3000
|
| 244 |
},
|
| 245 |
{
|
| 246 |
+
"epoch": 0.12,
|
| 247 |
+
"eval_loss": 1.5202959775924683,
|
| 248 |
+
"eval_runtime": 98.5199,
|
| 249 |
+
"eval_samples_per_second": 139.007,
|
| 250 |
+
"eval_steps_per_second": 2.172,
|
| 251 |
"step": 3000
|
| 252 |
},
|
| 253 |
{
|
| 254 |
+
"epoch": 0.124,
|
| 255 |
+
"grad_norm": 0.9433075785636902,
|
| 256 |
+
"learning_rate": 2.4191507348061575e-06,
|
| 257 |
+
"loss": 0.803,
|
| 258 |
"step": 3100
|
| 259 |
},
|
| 260 |
{
|
| 261 |
+
"epoch": 0.128,
|
| 262 |
+
"grad_norm": 0.9418466091156006,
|
| 263 |
+
"learning_rate": 2.4134430650360284e-06,
|
| 264 |
+
"loss": 0.8088,
|
| 265 |
"step": 3200
|
| 266 |
},
|
| 267 |
{
|
| 268 |
+
"epoch": 0.132,
|
| 269 |
+
"grad_norm": 0.9223436713218689,
|
| 270 |
+
"learning_rate": 2.407547943885489e-06,
|
| 271 |
+
"loss": 0.8116,
|
| 272 |
"step": 3300
|
| 273 |
},
|
| 274 |
{
|
| 275 |
+
"epoch": 0.136,
|
| 276 |
+
"grad_norm": 0.9359924793243408,
|
| 277 |
+
"learning_rate": 2.4014663211634552e-06,
|
| 278 |
+
"loss": 0.8232,
|
| 279 |
"step": 3400
|
| 280 |
},
|
| 281 |
{
|
| 282 |
+
"epoch": 0.14,
|
| 283 |
+
"grad_norm": 0.9347231388092041,
|
| 284 |
+
"learning_rate": 2.395199176727567e-06,
|
| 285 |
+
"loss": 0.8131,
|
| 286 |
"step": 3500
|
| 287 |
},
|
| 288 |
{
|
| 289 |
+
"epoch": 0.144,
|
| 290 |
+
"grad_norm": 0.9255951046943665,
|
| 291 |
+
"learning_rate": 2.388747520326311e-06,
|
| 292 |
+
"loss": 0.8064,
|
| 293 |
"step": 3600
|
| 294 |
},
|
| 295 |
{
|
| 296 |
+
"epoch": 0.148,
|
| 297 |
+
"grad_norm": 0.8580342531204224,
|
| 298 |
+
"learning_rate": 2.3821123914363374e-06,
|
| 299 |
+
"loss": 0.8247,
|
| 300 |
"step": 3700
|
| 301 |
},
|
| 302 |
{
|
| 303 |
+
"epoch": 0.152,
|
| 304 |
+
"grad_norm": 0.8920683860778809,
|
| 305 |
+
"learning_rate": 2.3752948590949766e-06,
|
| 306 |
+
"loss": 0.8058,
|
| 307 |
"step": 3800
|
| 308 |
},
|
| 309 |
{
|
| 310 |
+
"epoch": 0.156,
|
| 311 |
+
"grad_norm": 0.8848472237586975,
|
| 312 |
+
"learning_rate": 2.368296021728002e-06,
|
| 313 |
+
"loss": 0.8209,
|
| 314 |
"step": 3900
|
| 315 |
},
|
| 316 |
{
|
| 317 |
+
"epoch": 0.16,
|
| 318 |
+
"grad_norm": 0.9708815217018127,
|
| 319 |
+
"learning_rate": 2.3611170069726532e-06,
|
| 320 |
+
"loss": 0.8216,
|
| 321 |
"step": 4000
|
| 322 |
},
|
| 323 |
{
|
| 324 |
+
"epoch": 0.16,
|
| 325 |
+
"eval_loss": 1.5283503532409668,
|
| 326 |
+
"eval_runtime": 98.9755,
|
| 327 |
+
"eval_samples_per_second": 138.368,
|
| 328 |
+
"eval_steps_per_second": 2.162,
|
| 329 |
"step": 4000
|
| 330 |
},
|
| 331 |
{
|
| 332 |
+
"epoch": 0.164,
|
| 333 |
+
"grad_norm": 0.8715313673019409,
|
| 334 |
+
"learning_rate": 2.3537589714959523e-06,
|
| 335 |
+
"loss": 0.8185,
|
| 336 |
"step": 4100
|
| 337 |
},
|
| 338 |
{
|
| 339 |
+
"epoch": 0.168,
|
| 340 |
+
"grad_norm": 0.9748795032501221,
|
| 341 |
+
"learning_rate": 2.346223100808346e-06,
|
| 342 |
+
"loss": 0.8172,
|
| 343 |
"step": 4200
|
| 344 |
},
|
| 345 |
{
|
| 346 |
+
"epoch": 0.172,
|
| 347 |
+
"grad_norm": 0.900182843208313,
|
| 348 |
+
"learning_rate": 2.3385106090726974e-06,
|
| 349 |
+
"loss": 0.8101,
|
| 350 |
"step": 4300
|
| 351 |
},
|
| 352 |
{
|
| 353 |
+
"epoch": 0.176,
|
| 354 |
+
"grad_norm": 0.8882376551628113,
|
| 355 |
+
"learning_rate": 2.330622738908663e-06,
|
| 356 |
+
"loss": 0.8004,
|
| 357 |
"step": 4400
|
| 358 |
},
|
| 359 |
{
|
| 360 |
+
"epoch": 0.18,
|
| 361 |
+
"grad_norm": 0.9087768793106079,
|
| 362 |
+
"learning_rate": 2.322560761192485e-06,
|
| 363 |
+
"loss": 0.8028,
|
| 364 |
"step": 4500
|
| 365 |
},
|
| 366 |
{
|
| 367 |
+
"epoch": 0.184,
|
| 368 |
+
"grad_norm": 0.9928045868873596,
|
| 369 |
+
"learning_rate": 2.3143259748522308e-06,
|
| 370 |
+
"loss": 0.8257,
|
| 371 |
"step": 4600
|
| 372 |
},
|
| 373 |
{
|
| 374 |
+
"epoch": 0.188,
|
| 375 |
+
"grad_norm": 0.9519675970077515,
|
| 376 |
+
"learning_rate": 2.3059197066585126e-06,
|
| 377 |
+
"loss": 0.817,
|
| 378 |
"step": 4700
|
| 379 |
},
|
| 380 |
{
|
| 381 |
+
"epoch": 0.192,
|
| 382 |
+
"grad_norm": 0.970738410949707,
|
| 383 |
+
"learning_rate": 2.297343311010719e-06,
|
| 384 |
+
"loss": 0.8109,
|
| 385 |
"step": 4800
|
| 386 |
},
|
| 387 |
{
|
| 388 |
+
"epoch": 0.196,
|
| 389 |
+
"grad_norm": 0.9740980267524719,
|
| 390 |
+
"learning_rate": 2.2885981697188002e-06,
|
| 391 |
+
"loss": 0.8168,
|
| 392 |
"step": 4900
|
| 393 |
},
|
| 394 |
{
|
| 395 |
+
"epoch": 0.2,
|
| 396 |
+
"grad_norm": 0.9454805850982666,
|
| 397 |
+
"learning_rate": 2.2796856917806313e-06,
|
| 398 |
+
"loss": 0.8305,
|
| 399 |
"step": 5000
|
| 400 |
},
|
| 401 |
{
|
| 402 |
+
"epoch": 0.2,
|
| 403 |
+
"eval_loss": 1.5317082405090332,
|
| 404 |
+
"eval_runtime": 98.9715,
|
| 405 |
+
"eval_samples_per_second": 138.373,
|
| 406 |
+
"eval_steps_per_second": 2.162,
|
| 407 |
"step": 5000
|
| 408 |
},
|
| 409 |
{
|
| 410 |
+
"epoch": 0.204,
|
| 411 |
+
"grad_norm": 0.9181498289108276,
|
| 412 |
+
"learning_rate": 2.270607313155e-06,
|
| 413 |
+
"loss": 0.807,
|
| 414 |
"step": 5100
|
| 415 |
},
|
| 416 |
{
|
| 417 |
+
"epoch": 0.208,
|
| 418 |
+
"grad_norm": 0.8452897071838379,
|
| 419 |
+
"learning_rate": 2.2613644965302456e-06,
|
| 420 |
+
"loss": 0.802,
|
| 421 |
"step": 5200
|
| 422 |
},
|
| 423 |
{
|
| 424 |
+
"epoch": 0.212,
|
| 425 |
+
"grad_norm": 0.8827036619186401,
|
| 426 |
+
"learning_rate": 2.251958731088596e-06,
|
| 427 |
+
"loss": 0.8001,
|
| 428 |
"step": 5300
|
| 429 |
},
|
| 430 |
{
|
| 431 |
+
"epoch": 0.216,
|
| 432 |
+
"grad_norm": 0.8728039264678955,
|
| 433 |
+
"learning_rate": 2.242391532266232e-06,
|
| 434 |
+
"loss": 0.8211,
|
| 435 |
"step": 5400
|
| 436 |
},
|
| 437 |
{
|
| 438 |
+
"epoch": 0.22,
|
| 439 |
+
"grad_norm": 0.9410618543624878,
|
| 440 |
+
"learning_rate": 2.2326644415091264e-06,
|
| 441 |
+
"loss": 0.7996,
|
| 442 |
"step": 5500
|
| 443 |
},
|
| 444 |
{
|
| 445 |
+
"epoch": 0.224,
|
| 446 |
+
"grad_norm": 0.9829330444335938,
|
| 447 |
+
"learning_rate": 2.2227790260246856e-06,
|
| 448 |
+
"loss": 0.7971,
|
| 449 |
"step": 5600
|
| 450 |
},
|
| 451 |
{
|
| 452 |
+
"epoch": 0.228,
|
| 453 |
+
"grad_norm": 0.9688398241996765,
|
| 454 |
+
"learning_rate": 2.2127368785292484e-06,
|
| 455 |
+
"loss": 0.7854,
|
| 456 |
"step": 5700
|
| 457 |
},
|
| 458 |
{
|
| 459 |
+
"epoch": 0.232,
|
| 460 |
+
"grad_norm": 0.864470362663269,
|
| 461 |
+
"learning_rate": 2.2025396169914697e-06,
|
| 462 |
+
"loss": 0.8192,
|
| 463 |
"step": 5800
|
| 464 |
},
|
| 465 |
{
|
| 466 |
+
"epoch": 0.236,
|
| 467 |
+
"grad_norm": 0.9038395881652832,
|
| 468 |
+
"learning_rate": 2.1921888843716356e-06,
|
| 469 |
+
"loss": 0.8005,
|
| 470 |
"step": 5900
|
| 471 |
},
|
| 472 |
{
|
| 473 |
+
"epoch": 0.24,
|
| 474 |
+
"grad_norm": 0.8807651996612549,
|
| 475 |
+
"learning_rate": 2.181686348356955e-06,
|
| 476 |
+
"loss": 0.806,
|
| 477 |
"step": 6000
|
| 478 |
},
|
| 479 |
{
|
| 480 |
+
"epoch": 0.24,
|
| 481 |
+
"eval_loss": 1.524116039276123,
|
| 482 |
+
"eval_runtime": 99.2477,
|
| 483 |
+
"eval_samples_per_second": 137.988,
|
| 484 |
+
"eval_steps_per_second": 2.156,
|
| 485 |
"step": 6000
|
| 486 |
},
|
| 487 |
{
|
| 488 |
+
"epoch": 0.244,
|
| 489 |
+
"grad_norm": 1.0644515752792358,
|
| 490 |
+
"learning_rate": 2.1710337010928655e-06,
|
| 491 |
+
"loss": 0.8232,
|
| 492 |
"step": 6100
|
| 493 |
},
|
| 494 |
{
|
| 495 |
+
"epoch": 0.248,
|
| 496 |
+
"grad_norm": 0.9187564253807068,
|
| 497 |
+
"learning_rate": 2.1602326589103967e-06,
|
| 498 |
+
"loss": 0.8036,
|
| 499 |
"step": 6200
|
| 500 |
},
|
| 501 |
{
|
| 502 |
+
"epoch": 0.252,
|
| 503 |
+
"grad_norm": 0.9233301877975464,
|
| 504 |
+
"learning_rate": 2.1492849620496414e-06,
|
| 505 |
+
"loss": 0.8118,
|
| 506 |
"step": 6300
|
| 507 |
},
|
| 508 |
{
|
| 509 |
+
"epoch": 0.256,
|
| 510 |
+
"grad_norm": 0.9559895396232605,
|
| 511 |
+
"learning_rate": 2.13819237437937e-06,
|
| 512 |
+
"loss": 0.7959,
|
| 513 |
"step": 6400
|
| 514 |
},
|
| 515 |
{
|
| 516 |
+
"epoch": 0.26,
|
| 517 |
+
"grad_norm": 0.8455320000648499,
|
| 518 |
+
"learning_rate": 2.126956683112842e-06,
|
| 519 |
+
"loss": 0.8254,
|
| 520 |
"step": 6500
|
| 521 |
},
|
| 522 |
{
|
| 523 |
+
"epoch": 0.264,
|
| 524 |
+
"grad_norm": 0.942471444606781,
|
| 525 |
+
"learning_rate": 2.1155796985198495e-06,
|
| 526 |
+
"loss": 0.808,
|
| 527 |
"step": 6600
|
| 528 |
},
|
| 529 |
{
|
| 530 |
+
"epoch": 0.268,
|
| 531 |
+
"grad_norm": 0.8535305261611938,
|
| 532 |
+
"learning_rate": 2.1040632536350573e-06,
|
| 533 |
+
"loss": 0.8182,
|
| 534 |
"step": 6700
|
| 535 |
},
|
| 536 |
{
|
| 537 |
+
"epoch": 0.272,
|
| 538 |
+
"grad_norm": 0.8879380226135254,
|
| 539 |
+
"learning_rate": 2.092409203962663e-06,
|
| 540 |
+
"loss": 0.8177,
|
| 541 |
"step": 6800
|
| 542 |
},
|
| 543 |
{
|
| 544 |
+
"epoch": 0.276,
|
| 545 |
+
"grad_norm": 0.8684147000312805,
|
| 546 |
+
"learning_rate": 2.080619427177443e-06,
|
| 547 |
+
"loss": 0.7982,
|
| 548 |
"step": 6900
|
| 549 |
},
|
| 550 |
{
|
| 551 |
+
"epoch": 0.28,
|
| 552 |
+
"grad_norm": 0.9437069892883301,
|
| 553 |
+
"learning_rate": 2.0686958228222298e-06,
|
| 554 |
+
"loss": 0.7984,
|
| 555 |
"step": 7000
|
| 556 |
},
|
| 557 |
{
|
| 558 |
+
"epoch": 0.28,
|
| 559 |
+
"eval_loss": 1.530232548713684,
|
| 560 |
+
"eval_runtime": 99.3518,
|
| 561 |
+
"eval_samples_per_second": 137.844,
|
| 562 |
+
"eval_steps_per_second": 2.154,
|
| 563 |
"step": 7000
|
| 564 |
},
|
| 565 |
{
|
| 566 |
+
"epoch": 0.284,
|
| 567 |
+
"grad_norm": 0.9226755499839783,
|
| 568 |
+
"learning_rate": 2.056640312001856e-06,
|
| 569 |
+
"loss": 0.8072,
|
| 570 |
"step": 7100
|
| 571 |
},
|
| 572 |
{
|
| 573 |
+
"epoch": 0.288,
|
| 574 |
+
"grad_norm": 0.9192745685577393,
|
| 575 |
+
"learning_rate": 2.0444548370736335e-06,
|
| 576 |
+
"loss": 0.8081,
|
| 577 |
"step": 7200
|
| 578 |
},
|
| 579 |
{
|
| 580 |
+
"epoch": 0.292,
|
| 581 |
+
"grad_norm": 1.026985764503479,
|
| 582 |
+
"learning_rate": 2.032141361334406e-06,
|
| 583 |
+
"loss": 0.8074,
|
| 584 |
"step": 7300
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 0.296,
|
| 588 |
+
"grad_norm": 0.8428290486335754,
|
| 589 |
+
"learning_rate": 2.019701868704224e-06,
|
| 590 |
+
"loss": 0.8081,
|
| 591 |
"step": 7400
|
| 592 |
},
|
| 593 |
{
|
| 594 |
+
"epoch": 0.3,
|
| 595 |
+
"grad_norm": 0.9866459369659424,
|
| 596 |
+
"learning_rate": 2.007138363406702e-06,
|
| 597 |
+
"loss": 0.8241,
|
| 598 |
"step": 7500
|
| 599 |
},
|
| 600 |
{
|
| 601 |
+
"epoch": 0.304,
|
| 602 |
+
"grad_norm": 0.9240759015083313,
|
| 603 |
+
"learning_rate": 1.9944528696461016e-06,
|
| 604 |
+
"loss": 0.8089,
|
| 605 |
"step": 7600
|
| 606 |
},
|
| 607 |
{
|
| 608 |
+
"epoch": 0.308,
|
| 609 |
+
"grad_norm": 0.8980386853218079,
|
| 610 |
+
"learning_rate": 1.9816474312811984e-06,
|
| 611 |
+
"loss": 0.7995,
|
| 612 |
"step": 7700
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 0.312,
|
| 616 |
+
"grad_norm": 0.9766695499420166,
|
| 617 |
+
"learning_rate": 1.9687241114959753e-06,
|
| 618 |
+
"loss": 0.7969,
|
| 619 |
"step": 7800
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"epoch": 0.316,
|
| 623 |
+
"grad_norm": 0.8739997148513794,
|
| 624 |
+
"learning_rate": 1.955684992467211e-06,
|
| 625 |
+
"loss": 0.8053,
|
| 626 |
"step": 7900
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 0.32,
|
| 630 |
+
"grad_norm": 0.9071422219276428,
|
| 631 |
+
"learning_rate": 1.942532175029003e-06,
|
| 632 |
+
"loss": 0.7896,
|
| 633 |
"step": 8000
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"epoch": 0.32,
|
| 637 |
+
"eval_loss": 1.5243619680404663,
|
| 638 |
+
"eval_runtime": 99.5243,
|
| 639 |
+
"eval_samples_per_second": 137.605,
|
| 640 |
+
"eval_steps_per_second": 2.15,
|
| 641 |
"step": 8000
|
| 642 |
},
|
| 643 |
{
|
| 644 |
+
"epoch": 0.324,
|
| 645 |
+
"grad_norm": 0.9778127670288086,
|
| 646 |
+
"learning_rate": 1.929267778334285e-06,
|
| 647 |
+
"loss": 0.7878,
|
| 648 |
"step": 8100
|
| 649 |
},
|
| 650 |
{
|
| 651 |
+
"epoch": 0.328,
|
| 652 |
+
"grad_norm": 0.9122934937477112,
|
| 653 |
+
"learning_rate": 1.915893939513396e-06,
|
| 654 |
+
"loss": 0.7967,
|
| 655 |
"step": 8200
|
| 656 |
},
|
| 657 |
{
|
| 658 |
+
"epoch": 0.332,
|
| 659 |
+
"grad_norm": 0.90513676404953,
|
| 660 |
+
"learning_rate": 1.9024128133297467e-06,
|
| 661 |
+
"loss": 0.8048,
|
| 662 |
"step": 8300
|
| 663 |
},
|
| 664 |
{
|
| 665 |
+
"epoch": 0.336,
|
| 666 |
+
"grad_norm": 0.9107154607772827,
|
| 667 |
+
"learning_rate": 1.8888265718326532e-06,
|
| 668 |
+
"loss": 0.7944,
|
| 669 |
"step": 8400
|
| 670 |
},
|
| 671 |
{
|
| 672 |
+
"epoch": 0.34,
|
| 673 |
+
"grad_norm": 0.8964477777481079,
|
| 674 |
+
"learning_rate": 1.8751374040073774e-06,
|
| 675 |
+
"loss": 0.7958,
|
| 676 |
"step": 8500
|
| 677 |
},
|
| 678 |
{
|
| 679 |
+
"epoch": 0.344,
|
| 680 |
+
"grad_norm": 0.9018213152885437,
|
| 681 |
+
"learning_rate": 1.8613475154224456e-06,
|
| 682 |
+
"loss": 0.8065,
|
| 683 |
"step": 8600
|
| 684 |
},
|
| 685 |
{
|
| 686 |
+
"epoch": 0.348,
|
| 687 |
+
"grad_norm": 0.9653429985046387,
|
| 688 |
+
"learning_rate": 1.8474591278742894e-06,
|
| 689 |
+
"loss": 0.8194,
|
| 690 |
"step": 8700
|
| 691 |
},
|
| 692 |
{
|
| 693 |
+
"epoch": 0.352,
|
| 694 |
+
"grad_norm": 0.9324017763137817,
|
| 695 |
+
"learning_rate": 1.8334744790292766e-06,
|
| 696 |
+
"loss": 0.796,
|
| 697 |
"step": 8800
|
| 698 |
},
|
| 699 |
{
|
| 700 |
+
"epoch": 0.356,
|
| 701 |
+
"grad_norm": 1.0298709869384766,
|
| 702 |
+
"learning_rate": 1.8193958220631833e-06,
|
| 703 |
+
"loss": 0.8268,
|
| 704 |
"step": 8900
|
| 705 |
},
|
| 706 |
{
|
| 707 |
+
"epoch": 0.36,
|
| 708 |
+
"grad_norm": 0.8846196532249451,
|
| 709 |
+
"learning_rate": 1.805225425298166e-06,
|
| 710 |
+
"loss": 0.825,
|
| 711 |
"step": 9000
|
| 712 |
},
|
| 713 |
{
|
| 714 |
+
"epoch": 0.36,
|
| 715 |
+
"eval_loss": 1.5243308544158936,
|
| 716 |
+
"eval_runtime": 100.5198,
|
| 717 |
+
"eval_samples_per_second": 136.242,
|
| 718 |
+
"eval_steps_per_second": 2.129,
|
| 719 |
"step": 9000
|
| 720 |
},
|
| 721 |
{
|
| 722 |
+
"epoch": 0.364,
|
| 723 |
+
"grad_norm": 0.8830705881118774,
|
| 724 |
+
"learning_rate": 1.790965571837296e-06,
|
| 725 |
+
"loss": 0.8233,
|
| 726 |
"step": 9100
|
| 727 |
},
|
| 728 |
{
|
| 729 |
+
"epoch": 0.368,
|
| 730 |
+
"grad_norm": 0.9197975993156433,
|
| 731 |
+
"learning_rate": 1.7766185591967092e-06,
|
| 732 |
+
"loss": 0.8299,
|
| 733 |
"step": 9200
|
| 734 |
},
|
| 735 |
{
|
| 736 |
+
"epoch": 0.372,
|
| 737 |
+
"grad_norm": 1.0428673028945923,
|
| 738 |
+
"learning_rate": 1.762186698935437e-06,
|
| 739 |
+
"loss": 0.8182,
|
| 740 |
"step": 9300
|
| 741 |
},
|
| 742 |
{
|
| 743 |
+
"epoch": 0.376,
|
| 744 |
+
"grad_norm": 0.9466006755828857,
|
| 745 |
+
"learning_rate": 1.7476723162829723e-06,
|
| 746 |
+
"loss": 0.8255,
|
| 747 |
"step": 9400
|
| 748 |
},
|
| 749 |
{
|
| 750 |
+
"epoch": 0.38,
|
| 751 |
+
"grad_norm": 0.9237021803855896,
|
| 752 |
+
"learning_rate": 1.7330777497646328e-06,
|
| 753 |
+
"loss": 0.7672,
|
| 754 |
"step": 9500
|
| 755 |
},
|
| 756 |
{
|
| 757 |
+
"epoch": 0.384,
|
| 758 |
+
"grad_norm": 0.917202353477478,
|
| 759 |
+
"learning_rate": 1.7184053508247853e-06,
|
| 760 |
+
"loss": 0.8427,
|
| 761 |
"step": 9600
|
| 762 |
},
|
| 763 |
{
|
| 764 |
+
"epoch": 0.388,
|
| 765 |
+
"grad_norm": 0.9462612271308899,
|
| 766 |
+
"learning_rate": 1.703657483447983e-06,
|
| 767 |
+
"loss": 0.8409,
|
| 768 |
"step": 9700
|
| 769 |
},
|
| 770 |
{
|
| 771 |
+
"epoch": 0.392,
|
| 772 |
+
"grad_norm": 0.8924245834350586,
|
| 773 |
+
"learning_rate": 1.6888365237780886e-06,
|
| 774 |
+
"loss": 0.8335,
|
| 775 |
"step": 9800
|
| 776 |
},
|
| 777 |
{
|
| 778 |
+
"epoch": 0.396,
|
| 779 |
+
"grad_norm": 0.9719087481498718,
|
| 780 |
+
"learning_rate": 1.6739448597354327e-06,
|
| 781 |
+
"loss": 0.826,
|
| 782 |
"step": 9900
|
| 783 |
},
|
| 784 |
{
|
| 785 |
+
"epoch": 0.4,
|
| 786 |
+
"grad_norm": 0.8893173336982727,
|
| 787 |
+
"learning_rate": 1.6589848906320794e-06,
|
| 788 |
+
"loss": 0.8326,
|
| 789 |
"step": 10000
|
| 790 |
},
|
| 791 |
{
|
| 792 |
+
"epoch": 0.4,
|
| 793 |
+
"eval_loss": 1.5264792442321777,
|
| 794 |
+
"eval_runtime": 101.5699,
|
| 795 |
+
"eval_samples_per_second": 134.833,
|
| 796 |
+
"eval_steps_per_second": 2.107,
|
| 797 |
"step": 10000
|
| 798 |
},
|
| 799 |
{
|
| 800 |
+
"epoch": 0.404,
|
| 801 |
+
"grad_norm": 0.8719335198402405,
|
| 802 |
+
"learning_rate": 1.6439590267852528e-06,
|
| 803 |
+
"loss": 0.8198,
|
| 804 |
"step": 10100
|
| 805 |
},
|
| 806 |
{
|
| 807 |
+
"epoch": 0.408,
|
| 808 |
+
"grad_norm": 0.8997857570648193,
|
| 809 |
+
"learning_rate": 1.6288696891289938e-06,
|
| 810 |
+
"loss": 0.8103,
|
| 811 |
"step": 10200
|
| 812 |
},
|
| 813 |
{
|
| 814 |
+
"epoch": 0.412,
|
| 815 |
+
"grad_norm": 0.9756138920783997,
|
| 816 |
+
"learning_rate": 1.6137193088241021e-06,
|
| 817 |
+
"loss": 0.8245,
|
| 818 |
"step": 10300
|
| 819 |
},
|
| 820 |
{
|
| 821 |
+
"epoch": 0.416,
|
| 822 |
+
"grad_norm": 1.009027123451233,
|
| 823 |
+
"learning_rate": 1.598510326866435e-06,
|
| 824 |
+
"loss": 0.8226,
|
| 825 |
"step": 10400
|
| 826 |
},
|
| 827 |
{
|
| 828 |
+
"epoch": 0.42,
|
| 829 |
+
"grad_norm": 0.9941139221191406,
|
| 830 |
+
"learning_rate": 1.583245193693619e-06,
|
| 831 |
+
"loss": 0.8154,
|
| 832 |
"step": 10500
|
| 833 |
},
|
| 834 |
{
|
| 835 |
+
"epoch": 0.424,
|
| 836 |
+
"grad_norm": 0.9156614542007446,
|
| 837 |
+
"learning_rate": 1.5679263687902402e-06,
|
| 838 |
+
"loss": 0.8194,
|
| 839 |
"step": 10600
|
| 840 |
},
|
| 841 |
{
|
| 842 |
+
"epoch": 0.428,
|
| 843 |
+
"grad_norm": 0.9270005226135254,
|
| 844 |
+
"learning_rate": 1.552556320291578e-06,
|
| 845 |
+
"loss": 0.8144,
|
| 846 |
"step": 10700
|
| 847 |
},
|
| 848 |
{
|
| 849 |
+
"epoch": 0.432,
|
| 850 |
+
"grad_norm": 0.9664807915687561,
|
| 851 |
+
"learning_rate": 1.5371375245859446e-06,
|
| 852 |
+
"loss": 0.823,
|
| 853 |
"step": 10800
|
| 854 |
},
|
| 855 |
{
|
| 856 |
+
"epoch": 0.436,
|
| 857 |
+
"grad_norm": 0.9909628629684448,
|
| 858 |
+
"learning_rate": 1.5216724659156944e-06,
|
| 859 |
+
"loss": 0.8319,
|
| 860 |
"step": 10900
|
| 861 |
},
|
| 862 |
{
|
| 863 |
+
"epoch": 0.44,
|
| 864 |
+
"grad_norm": 1.0144808292388916,
|
| 865 |
+
"learning_rate": 1.506163635976969e-06,
|
| 866 |
+
"loss": 0.8272,
|
| 867 |
"step": 11000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
+
"epoch": 0.44,
|
| 871 |
+
"eval_loss": 1.5209919214248657,
|
| 872 |
+
"eval_runtime": 101.3638,
|
| 873 |
+
"eval_samples_per_second": 135.107,
|
| 874 |
+
"eval_steps_per_second": 2.111,
|
| 875 |
"step": 11000
|
| 876 |
},
|
| 877 |
{
|
| 878 |
+
"epoch": 0.444,
|
| 879 |
+
"grad_norm": 0.9689117074012756,
|
| 880 |
+
"learning_rate": 1.49061353351824e-06,
|
| 881 |
+
"loss": 0.8408,
|
| 882 |
"step": 11100
|
| 883 |
},
|
| 884 |
{
|
| 885 |
+
"epoch": 0.448,
|
| 886 |
+
"grad_norm": 1.0267921686172485,
|
| 887 |
+
"learning_rate": 1.4750246639377161e-06,
|
| 888 |
+
"loss": 0.8362,
|
| 889 |
"step": 11200
|
| 890 |
},
|
| 891 |
{
|
| 892 |
+
"epoch": 0.452,
|
| 893 |
+
"grad_norm": 0.920600175857544,
|
| 894 |
+
"learning_rate": 1.4593995388796797e-06,
|
| 895 |
+
"loss": 0.8343,
|
| 896 |
"step": 11300
|
| 897 |
},
|
| 898 |
{
|
| 899 |
+
"epoch": 0.456,
|
| 900 |
+
"grad_norm": 1.025995135307312,
|
| 901 |
+
"learning_rate": 1.4437406758298156e-06,
|
| 902 |
+
"loss": 0.8255,
|
| 903 |
"step": 11400
|
| 904 |
},
|
| 905 |
{
|
| 906 |
+
"epoch": 0.46,
|
| 907 |
+
"grad_norm": 0.889402449131012,
|
| 908 |
+
"learning_rate": 1.428050597709599e-06,
|
| 909 |
+
"loss": 0.839,
|
| 910 |
"step": 11500
|
| 911 |
},
|
| 912 |
{
|
| 913 |
+
"epoch": 0.464,
|
| 914 |
+
"grad_norm": 0.8957056999206543,
|
| 915 |
+
"learning_rate": 1.412331832469809e-06,
|
| 916 |
+
"loss": 0.8304,
|
| 917 |
"step": 11600
|
| 918 |
},
|
| 919 |
{
|
| 920 |
+
"epoch": 0.468,
|
| 921 |
+
"grad_norm": 0.9389684796333313,
|
| 922 |
+
"learning_rate": 1.39658691268323e-06,
|
| 923 |
+
"loss": 0.8523,
|
| 924 |
"step": 11700
|
| 925 |
},
|
| 926 |
{
|
| 927 |
+
"epoch": 0.472,
|
| 928 |
+
"grad_norm": 0.9115435481071472,
|
| 929 |
+
"learning_rate": 1.3808183751366089e-06,
|
| 930 |
+
"loss": 0.8421,
|
| 931 |
"step": 11800
|
| 932 |
},
|
| 933 |
{
|
| 934 |
+
"epoch": 0.476,
|
| 935 |
+
"grad_norm": 0.9521908164024353,
|
| 936 |
+
"learning_rate": 1.3650287604219342e-06,
|
| 937 |
+
"loss": 0.8704,
|
| 938 |
"step": 11900
|
| 939 |
},
|
| 940 |
{
|
| 941 |
+
"epoch": 0.48,
|
| 942 |
+
"grad_norm": 0.9166862964630127,
|
| 943 |
+
"learning_rate": 1.3492206125271016e-06,
|
| 944 |
+
"loss": 0.8527,
|
| 945 |
"step": 12000
|
| 946 |
},
|
| 947 |
{
|
| 948 |
+
"epoch": 0.48,
|
| 949 |
+
"eval_loss": 1.5229912996292114,
|
| 950 |
+
"eval_runtime": 101.6086,
|
| 951 |
+
"eval_samples_per_second": 134.782,
|
| 952 |
+
"eval_steps_per_second": 2.106,
|
| 953 |
"step": 12000
|
| 954 |
},
|
| 955 |
{
|
| 956 |
+
"epoch": 0.484,
|
| 957 |
+
"grad_norm": 0.9557492733001709,
|
| 958 |
+
"learning_rate": 1.333396478426031e-06,
|
| 959 |
+
"loss": 0.8499,
|
| 960 |
"step": 12100
|
| 961 |
},
|
| 962 |
{
|
| 963 |
+
"epoch": 0.488,
|
| 964 |
+
"grad_norm": 0.9957550764083862,
|
| 965 |
+
"learning_rate": 1.317558907668306e-06,
|
| 966 |
+
"loss": 0.8534,
|
| 967 |
"step": 12200
|
| 968 |
},
|
| 969 |
{
|
| 970 |
+
"epoch": 0.492,
|
| 971 |
+
"grad_norm": 1.1370068788528442,
|
| 972 |
+
"learning_rate": 1.3017104519683932e-06,
|
| 973 |
+
"loss": 0.8336,
|
| 974 |
"step": 12300
|
| 975 |
},
|
| 976 |
{
|
| 977 |
+
"epoch": 0.496,
|
| 978 |
+
"grad_norm": 0.9006808400154114,
|
| 979 |
+
"learning_rate": 1.285853664794518e-06,
|
| 980 |
+
"loss": 0.8196,
|
| 981 |
"step": 12400
|
| 982 |
},
|
| 983 |
{
|
| 984 |
+
"epoch": 0.5,
|
| 985 |
+
"grad_norm": 0.9441719651222229,
|
| 986 |
+
"learning_rate": 1.269991100957254e-06,
|
| 987 |
+
"loss": 0.844,
|
| 988 |
"step": 12500
|
| 989 |
},
|
| 990 |
{
|
| 991 |
+
"epoch": 0.504,
|
| 992 |
+
"grad_norm": 0.8616164922714233,
|
| 993 |
+
"learning_rate": 1.2541253161978986e-06,
|
| 994 |
+
"loss": 0.8319,
|
| 995 |
"step": 12600
|
| 996 |
},
|
| 997 |
{
|
| 998 |
+
"epoch": 0.508,
|
| 999 |
+
"grad_norm": 0.9243165850639343,
|
| 1000 |
+
"learning_rate": 1.238258866776697e-06,
|
| 1001 |
+
"loss": 0.8307,
|
| 1002 |
"step": 12700
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
+
"epoch": 0.512,
|
| 1006 |
+
"grad_norm": 0.9617546796798706,
|
| 1007 |
+
"learning_rate": 1.222394309060982e-06,
|
| 1008 |
+
"loss": 0.8562,
|
| 1009 |
"step": 12800
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
+
"epoch": 0.516,
|
| 1013 |
+
"grad_norm": 0.8897221684455872,
|
| 1014 |
+
"learning_rate": 1.2065341991133013e-06,
|
| 1015 |
+
"loss": 0.8344,
|
| 1016 |
"step": 12900
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
+
"epoch": 0.52,
|
| 1020 |
+
"grad_norm": 0.8364721536636353,
|
| 1021 |
+
"learning_rate": 1.1906810922795864e-06,
|
| 1022 |
+
"loss": 0.8389,
|
| 1023 |
"step": 13000
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
+
"epoch": 0.52,
|
| 1027 |
+
"eval_loss": 1.5288289785385132,
|
| 1028 |
+
"eval_runtime": 101.5647,
|
| 1029 |
+
"eval_samples_per_second": 134.84,
|
| 1030 |
+
"eval_steps_per_second": 2.107,
|
| 1031 |
"step": 13000
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
+
"epoch": 0.524,
|
| 1035 |
+
"grad_norm": 1.0084967613220215,
|
| 1036 |
+
"learning_rate": 1.1748375427774422e-06,
|
| 1037 |
+
"loss": 0.8498,
|
| 1038 |
"step": 13100
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
+
"epoch": 0.528,
|
| 1042 |
+
"grad_norm": 0.9439749717712402,
|
| 1043 |
+
"learning_rate": 1.1590061032846182e-06,
|
| 1044 |
+
"loss": 0.8509,
|
| 1045 |
"step": 13200
|
| 1046 |
},
|
| 1047 |
{
|
| 1048 |
+
"epoch": 0.532,
|
| 1049 |
+
"grad_norm": 0.8930461406707764,
|
| 1050 |
+
"learning_rate": 1.1431893245277262e-06,
|
| 1051 |
+
"loss": 0.8384,
|
| 1052 |
"step": 13300
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
+
"epoch": 0.536,
|
| 1056 |
+
"grad_norm": 1.0605283975601196,
|
| 1057 |
+
"learning_rate": 1.1273897548712726e-06,
|
| 1058 |
+
"loss": 0.8557,
|
| 1059 |
"step": 13400
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
+
"epoch": 0.54,
|
| 1063 |
+
"grad_norm": 0.8892098069190979,
|
| 1064 |
+
"learning_rate": 1.11160993990707e-06,
|
| 1065 |
+
"loss": 0.8378,
|
| 1066 |
"step": 13500
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
+
"epoch": 0.544,
|
| 1070 |
+
"grad_norm": 0.9008782505989075,
|
| 1071 |
+
"learning_rate": 1.0958524220440999e-06,
|
| 1072 |
+
"loss": 0.8437,
|
| 1073 |
"step": 13600
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
+
"epoch": 0.548,
|
| 1077 |
+
"grad_norm": 0.8771668672561646,
|
| 1078 |
+
"learning_rate": 1.0801197400988838e-06,
|
| 1079 |
+
"loss": 0.8512,
|
| 1080 |
"step": 13700
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
+
"epoch": 0.552,
|
| 1084 |
+
"grad_norm": 0.9245998859405518,
|
| 1085 |
+
"learning_rate": 1.0644144288864352e-06,
|
| 1086 |
+
"loss": 0.8671,
|
| 1087 |
"step": 13800
|
| 1088 |
},
|
| 1089 |
{
|
| 1090 |
+
"epoch": 0.556,
|
| 1091 |
+
"grad_norm": 0.9122968912124634,
|
| 1092 |
+
"learning_rate": 1.048739018811855e-06,
|
| 1093 |
+
"loss": 0.8328,
|
| 1094 |
"step": 13900
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
+
"epoch": 0.56,
|
| 1098 |
+
"grad_norm": 0.9968782067298889,
|
| 1099 |
+
"learning_rate": 1.0330960354626384e-06,
|
| 1100 |
+
"loss": 0.851,
|
| 1101 |
"step": 14000
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
+
"epoch": 0.56,
|
| 1105 |
+
"eval_loss": 1.5260618925094604,
|
| 1106 |
+
"eval_runtime": 101.9042,
|
| 1107 |
+
"eval_samples_per_second": 134.391,
|
| 1108 |
+
"eval_steps_per_second": 2.1,
|
| 1109 |
"step": 14000
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
+
"epoch": 0.564,
|
| 1113 |
+
"grad_norm": 1.0338596105575562,
|
| 1114 |
+
"learning_rate": 1.0174879992017586e-06,
|
| 1115 |
+
"loss": 0.8374,
|
| 1116 |
"step": 14100
|
| 1117 |
},
|
| 1118 |
{
|
| 1119 |
+
"epoch": 0.568,
|
| 1120 |
+
"grad_norm": 0.9291728734970093,
|
| 1121 |
+
"learning_rate": 1.0019174247615919e-06,
|
| 1122 |
+
"loss": 0.8356,
|
| 1123 |
"step": 14200
|
| 1124 |
},
|
| 1125 |
{
|
| 1126 |
+
"epoch": 0.572,
|
| 1127 |
+
"grad_norm": 0.8955647945404053,
|
| 1128 |
+
"learning_rate": 9.863868208387473e-07,
|
| 1129 |
+
"loss": 0.839,
|
| 1130 |
"step": 14300
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
+
"epoch": 0.576,
|
| 1134 |
+
"grad_norm": 0.9726178050041199,
|
| 1135 |
+
"learning_rate": 9.708986896898727e-07,
|
| 1136 |
+
"loss": 0.8396,
|
| 1137 |
"step": 14400
|
| 1138 |
},
|
| 1139 |
{
|
| 1140 |
+
"epoch": 0.58,
|
| 1141 |
+
"grad_norm": 0.9720205068588257,
|
| 1142 |
+
"learning_rate": 9.554555267284956e-07,
|
| 1143 |
+
"loss": 0.8334,
|
| 1144 |
"step": 14500
|
| 1145 |
},
|
| 1146 |
{
|
| 1147 |
+
"epoch": 0.584,
|
| 1148 |
+
"grad_norm": 0.9503899216651917,
|
| 1149 |
+
"learning_rate": 9.400598201229705e-07,
|
| 1150 |
+
"loss": 0.8165,
|
| 1151 |
"step": 14600
|
| 1152 |
},
|
| 1153 |
{
|
| 1154 |
+
"epoch": 0.588,
|
| 1155 |
+
"grad_norm": 0.8789735436439514,
|
| 1156 |
+
"learning_rate": 9.247140503955863e-07,
|
| 1157 |
+
"loss": 0.8262,
|
| 1158 |
"step": 14700
|
| 1159 |
},
|
| 1160 |
{
|
| 1161 |
+
"epoch": 0.592,
|
| 1162 |
+
"grad_norm": 1.4387589693069458,
|
| 1163 |
+
"learning_rate": 9.09420690022913e-07,
|
| 1164 |
+
"loss": 0.8378,
|
| 1165 |
"step": 14800
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
+
"epoch": 0.596,
|
| 1169 |
+
"grad_norm": 1.1762765645980835,
|
| 1170 |
+
"learning_rate": 8.941822030374405e-07,
|
| 1171 |
+
"loss": 0.8428,
|
| 1172 |
"step": 14900
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
+
"epoch": 0.6,
|
| 1176 |
+
"grad_norm": 0.880807638168335,
|
| 1177 |
+
"learning_rate": 8.790010446305814e-07,
|
| 1178 |
+
"loss": 0.8254,
|
| 1179 |
"step": 15000
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
+
"epoch": 0.6,
|
| 1183 |
+
"eval_loss": 1.5283499956130981,
|
| 1184 |
+
"eval_runtime": 103.2419,
|
| 1185 |
+
"eval_samples_per_second": 132.65,
|
| 1186 |
+
"eval_steps_per_second": 2.073,
|
| 1187 |
"step": 15000
|
| 1188 |
},
|
| 1189 |
{
|
| 1190 |
+
"epoch": 0.604,
|
| 1191 |
+
"grad_norm": 0.9635188579559326,
|
| 1192 |
+
"learning_rate": 8.63879660757092e-07,
|
| 1193 |
+
"loss": 0.798,
|
| 1194 |
"step": 15100
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
+
"epoch": 0.608,
|
| 1198 |
+
"grad_norm": 0.9472705721855164,
|
| 1199 |
+
"learning_rate": 8.488204877409884e-07,
|
| 1200 |
+
"loss": 0.8033,
|
| 1201 |
"step": 15200
|
| 1202 |
},
|
| 1203 |
{
|
| 1204 |
+
"epoch": 0.612,
|
| 1205 |
+
"grad_norm": 0.8378113508224487,
|
| 1206 |
+
"learning_rate": 8.338259518830106e-07,
|
| 1207 |
+
"loss": 0.8012,
|
| 1208 |
"step": 15300
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
+
"epoch": 0.616,
|
| 1212 |
+
"grad_norm": 0.9451029300689697,
|
| 1213 |
+
"learning_rate": 8.18898469069703e-07,
|
| 1214 |
+
"loss": 0.8047,
|
| 1215 |
"step": 15400
|
| 1216 |
},
|
| 1217 |
{
|
| 1218 |
+
"epoch": 0.62,
|
| 1219 |
+
"grad_norm": 0.9609344005584717,
|
| 1220 |
+
"learning_rate": 8.040404443841701e-07,
|
| 1221 |
+
"loss": 0.7927,
|
| 1222 |
"step": 15500
|
| 1223 |
},
|
| 1224 |
{
|
| 1225 |
+
"epoch": 0.624,
|
| 1226 |
+
"grad_norm": 0.8947242498397827,
|
| 1227 |
+
"learning_rate": 7.892542717185766e-07,
|
| 1228 |
+
"loss": 0.7885,
|
| 1229 |
"step": 15600
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
+
"epoch": 0.628,
|
| 1233 |
+
"grad_norm": 0.9105751514434814,
|
| 1234 |
+
"learning_rate": 7.745423333884483e-07,
|
| 1235 |
+
"loss": 0.801,
|
| 1236 |
"step": 15700
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
+
"epoch": 0.632,
|
| 1240 |
+
"grad_norm": 0.899936854839325,
|
| 1241 |
+
"learning_rate": 7.599069997488386e-07,
|
| 1242 |
+
"loss": 0.8005,
|
| 1243 |
"step": 15800
|
| 1244 |
},
|
| 1245 |
{
|
| 1246 |
+
"epoch": 0.636,
|
| 1247 |
+
"grad_norm": 1.0273375511169434,
|
| 1248 |
+
"learning_rate": 7.453506288124224e-07,
|
| 1249 |
+
"loss": 0.8015,
|
| 1250 |
"step": 15900
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
+
"epoch": 0.64,
|
| 1254 |
+
"grad_norm": 0.8960332274436951,
|
| 1255 |
+
"learning_rate": 7.308755658695775e-07,
|
| 1256 |
+
"loss": 0.8074,
|
| 1257 |
"step": 16000
|
| 1258 |
},
|
| 1259 |
{
|
| 1260 |
+
"epoch": 0.64,
|
| 1261 |
+
"eval_loss": 1.5343570709228516,
|
| 1262 |
+
"eval_runtime": 102.3372,
|
| 1263 |
+
"eval_samples_per_second": 133.822,
|
| 1264 |
+
"eval_steps_per_second": 2.091,
|
| 1265 |
"step": 16000
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
+
"epoch": 0.644,
|
| 1269 |
+
"grad_norm": 0.8942509293556213,
|
| 1270 |
+
"learning_rate": 7.164841431105172e-07,
|
| 1271 |
+
"loss": 0.796,
|
| 1272 |
"step": 16100
|
| 1273 |
},
|
| 1274 |
{
|
| 1275 |
+
"epoch": 0.648,
|
| 1276 |
+
"grad_norm": 0.9353269934654236,
|
| 1277 |
+
"learning_rate": 7.021786792495325e-07,
|
| 1278 |
+
"loss": 0.8196,
|
| 1279 |
"step": 16200
|
| 1280 |
},
|
| 1281 |
{
|
| 1282 |
+
"epoch": 0.652,
|
| 1283 |
+
"grad_norm": 0.985683262348175,
|
| 1284 |
+
"learning_rate": 6.879614791514075e-07,
|
| 1285 |
+
"loss": 0.808,
|
| 1286 |
"step": 16300
|
| 1287 |
},
|
| 1288 |
{
|
| 1289 |
+
"epoch": 0.656,
|
| 1290 |
+
"grad_norm": 0.8981220722198486,
|
| 1291 |
+
"learning_rate": 6.738348334600634e-07,
|
| 1292 |
+
"loss": 0.8015,
|
| 1293 |
"step": 16400
|
| 1294 |
},
|
| 1295 |
{
|
| 1296 |
+
"epoch": 0.66,
|
| 1297 |
+
"grad_norm": 0.9412031173706055,
|
| 1298 |
+
"learning_rate": 6.598010182294938e-07,
|
| 1299 |
+
"loss": 0.8009,
|
| 1300 |
"step": 16500
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
+
"epoch": 0.664,
|
| 1304 |
+
"grad_norm": 0.8926331996917725,
|
| 1305 |
+
"learning_rate": 6.458622945570538e-07,
|
| 1306 |
+
"loss": 0.783,
|
| 1307 |
"step": 16600
|
| 1308 |
},
|
| 1309 |
{
|
| 1310 |
+
"epoch": 0.668,
|
| 1311 |
+
"grad_norm": 0.8715830445289612,
|
| 1312 |
+
"learning_rate": 6.320209082191569e-07,
|
| 1313 |
+
"loss": 0.8127,
|
| 1314 |
"step": 16700
|
| 1315 |
},
|
| 1316 |
{
|
| 1317 |
+
"epoch": 0.672,
|
| 1318 |
+
"grad_norm": 0.8215272426605225,
|
| 1319 |
+
"learning_rate": 6.182790893094402e-07,
|
| 1320 |
+
"loss": 0.7958,
|
| 1321 |
"step": 16800
|
| 1322 |
},
|
| 1323 |
{
|
| 1324 |
+
"epoch": 0.676,
|
| 1325 |
+
"grad_norm": 0.9258244037628174,
|
| 1326 |
+
"learning_rate": 6.046390518794556e-07,
|
| 1327 |
+
"loss": 0.7931,
|
| 1328 |
"step": 16900
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
+
"epoch": 0.68,
|
| 1332 |
+
"grad_norm": 0.8930866122245789,
|
| 1333 |
+
"learning_rate": 5.911029935819468e-07,
|
| 1334 |
+
"loss": 0.7811,
|
| 1335 |
"step": 17000
|
| 1336 |
},
|
| 1337 |
{
|
| 1338 |
+
"epoch": 0.68,
|
| 1339 |
+
"eval_loss": 1.5324440002441406,
|
| 1340 |
+
"eval_runtime": 102.3251,
|
| 1341 |
+
"eval_samples_per_second": 133.838,
|
| 1342 |
+
"eval_steps_per_second": 2.091,
|
| 1343 |
"step": 17000
|
| 1344 |
},
|
| 1345 |
{
|
| 1346 |
+
"epoch": 0.684,
|
| 1347 |
+
"grad_norm": 0.9415869116783142,
|
| 1348 |
+
"learning_rate": 5.776730953167705e-07,
|
| 1349 |
+
"loss": 0.8003,
|
| 1350 |
"step": 17100
|
| 1351 |
},
|
| 1352 |
{
|
| 1353 |
+
"epoch": 0.688,
|
| 1354 |
+
"grad_norm": 0.892819344997406,
|
| 1355 |
+
"learning_rate": 5.643515208795141e-07,
|
| 1356 |
+
"loss": 0.7943,
|
| 1357 |
"step": 17200
|
| 1358 |
},
|
| 1359 |
{
|
| 1360 |
+
"epoch": 0.692,
|
| 1361 |
+
"grad_norm": 0.9383297562599182,
|
| 1362 |
+
"learning_rate": 5.511404166128647e-07,
|
| 1363 |
+
"loss": 0.7998,
|
| 1364 |
"step": 17300
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
+
"epoch": 0.696,
|
| 1368 |
+
"grad_norm": 0.8630228638648987,
|
| 1369 |
+
"learning_rate": 5.380419110608033e-07,
|
| 1370 |
+
"loss": 0.7949,
|
| 1371 |
"step": 17400
|
| 1372 |
},
|
| 1373 |
{
|
| 1374 |
+
"epoch": 0.7,
|
| 1375 |
+
"grad_norm": 0.9032106995582581,
|
| 1376 |
+
"learning_rate": 5.250581146256524e-07,
|
| 1377 |
+
"loss": 0.7928,
|
| 1378 |
"step": 17500
|
| 1379 |
},
|
| 1380 |
{
|
| 1381 |
+
"epoch": 0.704,
|
| 1382 |
+
"grad_norm": 0.9039574265480042,
|
| 1383 |
+
"learning_rate": 5.121911192280557e-07,
|
| 1384 |
+
"loss": 0.8012,
|
| 1385 |
"step": 17600
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
+
"epoch": 0.708,
|
| 1389 |
+
"grad_norm": 0.9616802334785461,
|
| 1390 |
+
"learning_rate": 4.994429979699302e-07,
|
| 1391 |
+
"loss": 0.7964,
|
| 1392 |
"step": 17700
|
| 1393 |
},
|
| 1394 |
{
|
| 1395 |
+
"epoch": 0.712,
|
| 1396 |
+
"grad_norm": 0.9427072405815125,
|
| 1397 |
+
"learning_rate": 4.868158048004537e-07,
|
| 1398 |
+
"loss": 0.805,
|
| 1399 |
"step": 17800
|
| 1400 |
},
|
| 1401 |
{
|
| 1402 |
+
"epoch": 0.716,
|
| 1403 |
+
"grad_norm": 0.9399961829185486,
|
| 1404 |
+
"learning_rate": 4.743115741851383e-07,
|
| 1405 |
+
"loss": 0.7913,
|
| 1406 |
"step": 17900
|
| 1407 |
},
|
| 1408 |
{
|
| 1409 |
+
"epoch": 0.72,
|
| 1410 |
+
"grad_norm": 0.8884769678115845,
|
| 1411 |
+
"learning_rate": 4.6193232077804006e-07,
|
| 1412 |
+
"loss": 0.7985,
|
| 1413 |
"step": 18000
|
| 1414 |
},
|
| 1415 |
{
|
| 1416 |
+
"epoch": 0.72,
|
| 1417 |
+
"eval_loss": 1.5309633016586304,
|
| 1418 |
+
"eval_runtime": 102.8962,
|
| 1419 |
+
"eval_samples_per_second": 133.095,
|
| 1420 |
+
"eval_steps_per_second": 2.08,
|
| 1421 |
"step": 18000
|
| 1422 |
},
|
| 1423 |
{
|
| 1424 |
+
"epoch": 0.724,
|
| 1425 |
+
"grad_norm": 0.9725548028945923,
|
| 1426 |
+
"learning_rate": 4.4968003909716243e-07,
|
| 1427 |
+
"loss": 0.8162,
|
| 1428 |
"step": 18100
|
| 1429 |
},
|
| 1430 |
{
|
| 1431 |
+
"epoch": 0.728,
|
| 1432 |
+
"grad_norm": 1.0075186491012573,
|
| 1433 |
+
"learning_rate": 4.3755670320310443e-07,
|
| 1434 |
+
"loss": 0.8054,
|
| 1435 |
"step": 18200
|
| 1436 |
},
|
| 1437 |
{
|
| 1438 |
+
"epoch": 0.732,
|
| 1439 |
+
"grad_norm": 0.8749048113822937,
|
| 1440 |
+
"learning_rate": 4.2556426638100555e-07,
|
| 1441 |
+
"loss": 0.8056,
|
| 1442 |
"step": 18300
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
+
"epoch": 0.736,
|
| 1446 |
+
"grad_norm": 0.9941290616989136,
|
| 1447 |
+
"learning_rate": 4.1370466082583353e-07,
|
| 1448 |
+
"loss": 0.8052,
|
| 1449 |
"step": 18400
|
| 1450 |
},
|
| 1451 |
{
|
| 1452 |
+
"epoch": 0.74,
|
| 1453 |
+
"grad_norm": 0.8676705956459045,
|
| 1454 |
+
"learning_rate": 4.0197979733107755e-07,
|
| 1455 |
+
"loss": 0.7861,
|
| 1456 |
"step": 18500
|
| 1457 |
},
|
| 1458 |
{
|
| 1459 |
+
"epoch": 0.744,
|
| 1460 |
+
"grad_norm": 0.9036993980407715,
|
| 1461 |
+
"learning_rate": 3.903915649808812e-07,
|
| 1462 |
+
"loss": 0.8081,
|
| 1463 |
"step": 18600
|
| 1464 |
},
|
| 1465 |
{
|
| 1466 |
+
"epoch": 0.748,
|
| 1467 |
+
"grad_norm": 0.9067134261131287,
|
| 1468 |
+
"learning_rate": 3.789418308456812e-07,
|
| 1469 |
+
"loss": 0.7956,
|
| 1470 |
"step": 18700
|
| 1471 |
},
|
| 1472 |
{
|
| 1473 |
+
"epoch": 0.752,
|
| 1474 |
+
"grad_norm": 0.8146563768386841,
|
| 1475 |
+
"learning_rate": 3.676324396813856e-07,
|
| 1476 |
+
"loss": 0.8031,
|
| 1477 |
"step": 18800
|
| 1478 |
},
|
| 1479 |
{
|
| 1480 |
+
"epoch": 0.756,
|
| 1481 |
+
"grad_norm": 0.9973321557044983,
|
| 1482 |
+
"learning_rate": 3.5646521363215447e-07,
|
| 1483 |
+
"loss": 0.794,
|
| 1484 |
"step": 18900
|
| 1485 |
},
|
| 1486 |
{
|
| 1487 |
+
"epoch": 0.76,
|
| 1488 |
+
"grad_norm": 0.9761902689933777,
|
| 1489 |
+
"learning_rate": 3.4544195193681615e-07,
|
| 1490 |
+
"loss": 0.7816,
|
| 1491 |
"step": 19000
|
| 1492 |
},
|
| 1493 |
{
|
| 1494 |
+
"epoch": 0.76,
|
| 1495 |
+
"eval_loss": 1.5294893980026245,
|
| 1496 |
+
"eval_runtime": 102.4113,
|
| 1497 |
+
"eval_samples_per_second": 133.726,
|
| 1498 |
+
"eval_steps_per_second": 2.09,
|
| 1499 |
"step": 19000
|
| 1500 |
},
|
| 1501 |
{
|
| 1502 |
+
"epoch": 0.764,
|
| 1503 |
+
"grad_norm": 0.8643273115158081,
|
| 1504 |
+
"learning_rate": 3.3456443063898157e-07,
|
| 1505 |
+
"loss": 0.7917,
|
| 1506 |
"step": 19100
|
| 1507 |
},
|
| 1508 |
{
|
| 1509 |
+
"epoch": 0.768,
|
| 1510 |
+
"grad_norm": 0.9306071400642395,
|
| 1511 |
+
"learning_rate": 3.238344023008888e-07,
|
| 1512 |
+
"loss": 0.8012,
|
| 1513 |
"step": 19200
|
| 1514 |
},
|
| 1515 |
{
|
| 1516 |
+
"epoch": 0.772,
|
| 1517 |
+
"grad_norm": 0.9324482083320618,
|
| 1518 |
+
"learning_rate": 3.132535957210366e-07,
|
| 1519 |
+
"loss": 0.7929,
|
| 1520 |
"step": 19300
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
+
"epoch": 0.776,
|
| 1524 |
+
"grad_norm": 0.8625467419624329,
|
| 1525 |
+
"learning_rate": 3.0282371565564324e-07,
|
| 1526 |
+
"loss": 0.7815,
|
| 1527 |
"step": 19400
|
| 1528 |
},
|
| 1529 |
{
|
| 1530 |
+
"epoch": 0.78,
|
| 1531 |
+
"grad_norm": 0.8669098019599915,
|
| 1532 |
+
"learning_rate": 2.925464425439789e-07,
|
| 1533 |
+
"loss": 0.8214,
|
| 1534 |
"step": 19500
|
| 1535 |
},
|
| 1536 |
{
|
| 1537 |
+
"epoch": 0.784,
|
| 1538 |
+
"grad_norm": 0.8781657814979553,
|
| 1539 |
+
"learning_rate": 2.824234322376185e-07,
|
| 1540 |
+
"loss": 0.7941,
|
| 1541 |
"step": 19600
|
| 1542 |
},
|
| 1543 |
{
|
| 1544 |
+
"epoch": 0.788,
|
| 1545 |
+
"grad_norm": 0.8899013996124268,
|
| 1546 |
+
"learning_rate": 2.724563157336542e-07,
|
| 1547 |
+
"loss": 0.7966,
|
| 1548 |
"step": 19700
|
| 1549 |
},
|
| 1550 |
{
|
| 1551 |
+
"epoch": 0.792,
|
| 1552 |
+
"grad_norm": 0.9773925542831421,
|
| 1553 |
+
"learning_rate": 2.626466989119131e-07,
|
| 1554 |
+
"loss": 0.8009,
|
| 1555 |
"step": 19800
|
| 1556 |
},
|
| 1557 |
{
|
| 1558 |
+
"epoch": 0.796,
|
| 1559 |
+
"grad_norm": 0.912438690662384,
|
| 1560 |
+
"learning_rate": 2.5299616227621946e-07,
|
| 1561 |
+
"loss": 0.7902,
|
| 1562 |
"step": 19900
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
+
"epoch": 0.8,
|
| 1566 |
+
"grad_norm": 0.9557161927223206,
|
| 1567 |
+
"learning_rate": 2.435062606997499e-07,
|
| 1568 |
+
"loss": 0.7889,
|
| 1569 |
"step": 20000
|
| 1570 |
},
|
| 1571 |
{
|
| 1572 |
+
"epoch": 0.8,
|
| 1573 |
+
"eval_loss": 1.5292094945907593,
|
| 1574 |
+
"eval_runtime": 102.5763,
|
| 1575 |
+
"eval_samples_per_second": 133.51,
|
| 1576 |
+
"eval_steps_per_second": 2.086,
|
| 1577 |
"step": 20000
|
| 1578 |
},
|
| 1579 |
{
|
| 1580 |
+
"epoch": 0.804,
|
| 1581 |
+
"grad_norm": 0.8561129570007324,
|
| 1582 |
+
"learning_rate": 2.3417852317451418e-07,
|
| 1583 |
+
"loss": 0.8033,
|
| 1584 |
"step": 20100
|
| 1585 |
},
|
| 1586 |
{
|
| 1587 |
+
"epoch": 0.808,
|
| 1588 |
+
"grad_norm": 0.9422599673271179,
|
| 1589 |
+
"learning_rate": 2.250144525650086e-07,
|
| 1590 |
+
"loss": 0.7985,
|
| 1591 |
"step": 20200
|
| 1592 |
},
|
| 1593 |
{
|
| 1594 |
+
"epoch": 0.812,
|
| 1595 |
+
"grad_norm": 0.8980026245117188,
|
| 1596 |
+
"learning_rate": 2.160155253660759e-07,
|
| 1597 |
+
"loss": 0.7951,
|
| 1598 |
"step": 20300
|
| 1599 |
},
|
| 1600 |
{
|
| 1601 |
+
"epoch": 0.816,
|
| 1602 |
+
"grad_norm": 0.8675551414489746,
|
| 1603 |
+
"learning_rate": 2.071831914650173e-07,
|
| 1604 |
+
"loss": 0.7994,
|
| 1605 |
"step": 20400
|
| 1606 |
},
|
| 1607 |
{
|
| 1608 |
+
"epoch": 0.82,
|
| 1609 |
+
"grad_norm": 0.8988806009292603,
|
| 1610 |
+
"learning_rate": 1.9851887390798922e-07,
|
| 1611 |
+
"loss": 0.7875,
|
| 1612 |
"step": 20500
|
| 1613 |
},
|
| 1614 |
{
|
| 1615 |
+
"epoch": 0.824,
|
| 1616 |
+
"grad_norm": 0.9102202653884888,
|
| 1617 |
+
"learning_rate": 1.9002396867072587e-07,
|
| 1618 |
+
"loss": 0.7993,
|
| 1619 |
"step": 20600
|
| 1620 |
},
|
| 1621 |
{
|
| 1622 |
+
"epoch": 0.828,
|
| 1623 |
+
"grad_norm": 0.9096868634223938,
|
| 1624 |
+
"learning_rate": 1.816998444336214e-07,
|
| 1625 |
+
"loss": 0.7704,
|
| 1626 |
"step": 20700
|
| 1627 |
},
|
| 1628 |
{
|
| 1629 |
+
"epoch": 0.832,
|
| 1630 |
+
"grad_norm": 0.9461880922317505,
|
| 1631 |
+
"learning_rate": 1.7354784236121206e-07,
|
| 1632 |
+
"loss": 0.7853,
|
| 1633 |
"step": 20800
|
| 1634 |
},
|
| 1635 |
{
|
| 1636 |
+
"epoch": 0.836,
|
| 1637 |
+
"grad_norm": 0.9219881296157837,
|
| 1638 |
+
"learning_rate": 1.6556927588609078e-07,
|
| 1639 |
+
"loss": 0.7857,
|
| 1640 |
"step": 20900
|
| 1641 |
},
|
| 1642 |
{
|
| 1643 |
+
"epoch": 0.84,
|
| 1644 |
+
"grad_norm": 0.8964762687683105,
|
| 1645 |
+
"learning_rate": 1.577654304972899e-07,
|
| 1646 |
+
"loss": 0.7872,
|
| 1647 |
"step": 21000
|
| 1648 |
},
|
| 1649 |
{
|
| 1650 |
+
"epoch": 0.84,
|
| 1651 |
+
"eval_loss": 1.524131178855896,
|
| 1652 |
+
"eval_runtime": 102.4749,
|
| 1653 |
+
"eval_samples_per_second": 133.642,
|
| 1654 |
+
"eval_steps_per_second": 2.088,
|
| 1655 |
"step": 21000
|
| 1656 |
},
|
| 1657 |
{
|
| 1658 |
+
"epoch": 0.844,
|
| 1659 |
+
"grad_norm": 0.9355736970901489,
|
| 1660 |
+
"learning_rate": 1.501375635331652e-07,
|
| 1661 |
+
"loss": 0.7957,
|
| 1662 |
"step": 21100
|
| 1663 |
},
|
| 1664 |
{
|
| 1665 |
+
"epoch": 0.848,
|
| 1666 |
+
"grad_norm": 0.8686819076538086,
|
| 1667 |
+
"learning_rate": 1.4268690397881675e-07,
|
| 1668 |
+
"loss": 0.793,
|
| 1669 |
"step": 21200
|
| 1670 |
},
|
| 1671 |
{
|
| 1672 |
+
"epoch": 0.852,
|
| 1673 |
+
"grad_norm": 0.874756395816803,
|
| 1674 |
+
"learning_rate": 1.3541465226807813e-07,
|
| 1675 |
+
"loss": 0.7878,
|
| 1676 |
"step": 21300
|
| 1677 |
},
|
| 1678 |
{
|
| 1679 |
+
"epoch": 0.856,
|
| 1680 |
+
"grad_norm": 0.9285154342651367,
|
| 1681 |
+
"learning_rate": 1.283219800901045e-07,
|
| 1682 |
+
"loss": 0.7547,
|
| 1683 |
"step": 21400
|
| 1684 |
},
|
| 1685 |
{
|
| 1686 |
+
"epoch": 0.86,
|
| 1687 |
+
"grad_norm": 0.9496791958808899,
|
| 1688 |
+
"learning_rate": 1.2141003020059273e-07,
|
| 1689 |
+
"loss": 0.7885,
|
| 1690 |
"step": 21500
|
| 1691 |
},
|
| 1692 |
{
|
| 1693 |
+
"epoch": 0.864,
|
| 1694 |
+
"grad_norm": 0.879410445690155,
|
| 1695 |
+
"learning_rate": 1.1467991623766287e-07,
|
| 1696 |
+
"loss": 0.8123,
|
| 1697 |
"step": 21600
|
| 1698 |
},
|
| 1699 |
{
|
| 1700 |
+
"epoch": 0.868,
|
| 1701 |
+
"grad_norm": 0.942361056804657,
|
| 1702 |
+
"learning_rate": 1.081327225424321e-07,
|
| 1703 |
+
"loss": 0.817,
|
| 1704 |
"step": 21700
|
| 1705 |
},
|
| 1706 |
{
|
| 1707 |
+
"epoch": 0.872,
|
| 1708 |
+
"grad_norm": 0.9548047184944153,
|
| 1709 |
+
"learning_rate": 1.0176950398430752e-07,
|
| 1710 |
+
"loss": 0.7925,
|
| 1711 |
"step": 21800
|
| 1712 |
},
|
| 1713 |
{
|
| 1714 |
+
"epoch": 0.876,
|
| 1715 |
+
"grad_norm": 0.8643764853477478,
|
| 1716 |
+
"learning_rate": 9.559128579102767e-08,
|
| 1717 |
+
"loss": 0.7985,
|
| 1718 |
"step": 21900
|
| 1719 |
},
|
| 1720 |
{
|
| 1721 |
+
"epoch": 0.88,
|
| 1722 |
+
"grad_norm": 0.9450801014900208,
|
| 1723 |
+
"learning_rate": 8.959906338348007e-08,
|
| 1724 |
+
"loss": 0.7975,
|
| 1725 |
"step": 22000
|
| 1726 |
},
|
| 1727 |
{
|
| 1728 |
+
"epoch": 0.88,
|
| 1729 |
+
"eval_loss": 1.5321519374847412,
|
| 1730 |
+
"eval_runtime": 103.5374,
|
| 1731 |
+
"eval_samples_per_second": 132.271,
|
| 1732 |
+
"eval_steps_per_second": 2.067,
|
| 1733 |
"step": 22000
|
| 1734 |
},
|
| 1735 |
{
|
| 1736 |
+
"epoch": 0.884,
|
| 1737 |
+
"grad_norm": 0.9130359292030334,
|
| 1738 |
+
"learning_rate": 8.37938022153223e-08,
|
| 1739 |
+
"loss": 0.8005,
|
| 1740 |
"step": 22100
|
| 1741 |
},
|
| 1742 |
{
|
| 1743 |
+
"epoch": 0.888,
|
| 1744 |
+
"grad_norm": 0.8732690215110779,
|
| 1745 |
+
"learning_rate": 7.817643761742891e-08,
|
| 1746 |
+
"loss": 0.7857,
|
| 1747 |
"step": 22200
|
| 1748 |
},
|
| 1749 |
{
|
| 1750 |
+
"epoch": 0.892,
|
| 1751 |
+
"grad_norm": 0.9094323515892029,
|
| 1752 |
+
"learning_rate": 7.274787464719338e-08,
|
| 1753 |
+
"loss": 0.8096,
|
| 1754 |
"step": 22300
|
| 1755 |
},
|
| 1756 |
{
|
| 1757 |
+
"epoch": 0.896,
|
| 1758 |
+
"grad_norm": 0.8987523913383484,
|
| 1759 |
+
"learning_rate": 6.75089879427078e-08,
|
| 1760 |
+
"loss": 0.8072,
|
| 1761 |
"step": 22400
|
| 1762 |
},
|
| 1763 |
{
|
| 1764 |
+
"epoch": 0.9,
|
| 1765 |
+
"grad_norm": 0.9105306267738342,
|
| 1766 |
+
"learning_rate": 6.246062158184241e-08,
|
| 1767 |
+
"loss": 0.7968,
|
| 1768 |
"step": 22500
|
| 1769 |
},
|
| 1770 |
{
|
| 1771 |
+
"epoch": 0.904,
|
| 1772 |
+
"grad_norm": 0.8889061808586121,
|
| 1773 |
+
"learning_rate": 5.7603588946250064e-08,
|
| 1774 |
+
"loss": 0.7971,
|
| 1775 |
"step": 22600
|
| 1776 |
},
|
| 1777 |
{
|
| 1778 |
+
"epoch": 0.908,
|
| 1779 |
+
"grad_norm": 0.9296440482139587,
|
| 1780 |
+
"learning_rate": 5.293867259031568e-08,
|
| 1781 |
+
"loss": 0.7896,
|
| 1782 |
"step": 22700
|
| 1783 |
},
|
| 1784 |
{
|
| 1785 |
+
"epoch": 0.912,
|
| 1786 |
+
"grad_norm": 1.0374181270599365,
|
| 1787 |
+
"learning_rate": 4.8466624115073164e-08,
|
| 1788 |
+
"loss": 0.808,
|
| 1789 |
"step": 22800
|
| 1790 |
},
|
| 1791 |
{
|
| 1792 |
+
"epoch": 1.00384,
|
| 1793 |
+
"grad_norm": 0.8791893124580383,
|
| 1794 |
+
"learning_rate": 4.4188164047108403e-08,
|
| 1795 |
+
"loss": 0.7835,
|
| 1796 |
"step": 22900
|
| 1797 |
},
|
| 1798 |
{
|
| 1799 |
+
"epoch": 1.00784,
|
| 1800 |
+
"grad_norm": 0.8789498209953308,
|
| 1801 |
+
"learning_rate": 4.010398172247104e-08,
|
| 1802 |
+
"loss": 0.7987,
|
| 1803 |
"step": 23000
|
| 1804 |
},
|
| 1805 |
{
|
| 1806 |
+
"epoch": 1.00784,
|
| 1807 |
+
"eval_loss": 1.5310029983520508,
|
| 1808 |
+
"eval_runtime": 101.8479,
|
| 1809 |
+
"eval_samples_per_second": 134.465,
|
| 1810 |
+
"eval_steps_per_second": 2.101,
|
| 1811 |
"step": 23000
|
| 1812 |
},
|
| 1813 |
{
|
| 1814 |
+
"epoch": 1.01184,
|
| 1815 |
+
"grad_norm": 0.9262071847915649,
|
| 1816 |
+
"learning_rate": 3.6214735175608004e-08,
|
| 1817 |
+
"loss": 0.7966,
|
| 1818 |
"step": 23100
|
| 1819 |
},
|
| 1820 |
{
|
| 1821 |
+
"epoch": 1.01584,
|
| 1822 |
+
"grad_norm": 0.8986383676528931,
|
| 1823 |
+
"learning_rate": 3.252105103334499e-08,
|
| 1824 |
+
"loss": 0.7954,
|
| 1825 |
"step": 23200
|
| 1826 |
},
|
| 1827 |
{
|
| 1828 |
+
"epoch": 1.01984,
|
| 1829 |
+
"grad_norm": 0.9548205733299255,
|
| 1830 |
+
"learning_rate": 2.9023524413923365e-08,
|
| 1831 |
+
"loss": 0.7934,
|
| 1832 |
"step": 23300
|
| 1833 |
},
|
| 1834 |
{
|
| 1835 |
+
"epoch": 1.02384,
|
| 1836 |
+
"grad_norm": 0.9211428165435791,
|
| 1837 |
+
"learning_rate": 2.5722718831117656e-08,
|
| 1838 |
+
"loss": 0.8068,
|
| 1839 |
"step": 23400
|
| 1840 |
},
|
| 1841 |
{
|
| 1842 |
+
"epoch": 1.02784,
|
| 1843 |
+
"grad_norm": 0.9240931272506714,
|
| 1844 |
+
"learning_rate": 2.26191661034425e-08,
|
| 1845 |
+
"loss": 0.787,
|
| 1846 |
"step": 23500
|
| 1847 |
},
|
| 1848 |
{
|
| 1849 |
+
"epoch": 1.03184,
|
| 1850 |
+
"grad_norm": 0.9866804480552673,
|
| 1851 |
+
"learning_rate": 1.9713366268468148e-08,
|
| 1852 |
+
"loss": 0.7929,
|
| 1853 |
"step": 23600
|
| 1854 |
},
|
| 1855 |
{
|
| 1856 |
+
"epoch": 1.03584,
|
| 1857 |
+
"grad_norm": 0.9947385787963867,
|
| 1858 |
+
"learning_rate": 1.700578750225432e-08,
|
| 1859 |
+
"loss": 0.7973,
|
| 1860 |
"step": 23700
|
| 1861 |
},
|
| 1862 |
{
|
| 1863 |
+
"epoch": 1.03984,
|
| 1864 |
+
"grad_norm": 0.8872534036636353,
|
| 1865 |
+
"learning_rate": 1.4496866043919865e-08,
|
| 1866 |
+
"loss": 0.7995,
|
| 1867 |
"step": 23800
|
| 1868 |
},
|
| 1869 |
{
|
| 1870 |
+
"epoch": 1.04384,
|
| 1871 |
+
"grad_norm": 0.8726480007171631,
|
| 1872 |
+
"learning_rate": 1.2187006125356087e-08,
|
| 1873 |
+
"loss": 0.7929,
|
| 1874 |
"step": 23900
|
| 1875 |
},
|
| 1876 |
{
|
| 1877 |
+
"epoch": 1.04784,
|
| 1878 |
+
"grad_norm": 0.881963849067688,
|
| 1879 |
+
"learning_rate": 1.0076579906098255e-08,
|
| 1880 |
+
"loss": 0.8044,
|
| 1881 |
"step": 24000
|
| 1882 |
},
|
| 1883 |
{
|
| 1884 |
+
"epoch": 1.04784,
|
| 1885 |
+
"eval_loss": 1.5276943445205688,
|
| 1886 |
+
"eval_runtime": 99.4561,
|
| 1887 |
+
"eval_samples_per_second": 137.699,
|
| 1888 |
+
"eval_steps_per_second": 2.152,
|
| 1889 |
"step": 24000
|
| 1890 |
},
|
| 1891 |
{
|
| 1892 |
+
"epoch": 1.0518399999999999,
|
| 1893 |
+
"grad_norm": 0.8809722065925598,
|
| 1894 |
+
"learning_rate": 8.16592741336386e-09,
|
| 1895 |
+
"loss": 0.7832,
|
| 1896 |
"step": 24100
|
| 1897 |
},
|
| 1898 |
{
|
| 1899 |
+
"epoch": 1.05584,
|
| 1900 |
+
"grad_norm": 0.8471363186836243,
|
| 1901 |
+
"learning_rate": 6.455356487267833e-09,
|
| 1902 |
+
"loss": 0.7815,
|
| 1903 |
"step": 24200
|
| 1904 |
},
|
| 1905 |
{
|
| 1906 |
+
"epoch": 1.05984,
|
| 1907 |
+
"grad_norm": 0.9595879912376404,
|
| 1908 |
+
"learning_rate": 4.9451427312251224e-09,
|
| 1909 |
+
"loss": 0.7943,
|
| 1910 |
"step": 24300
|
| 1911 |
},
|
| 1912 |
{
|
| 1913 |
+
"epoch": 1.06384,
|
| 1914 |
+
"grad_norm": 0.8937146663665771,
|
| 1915 |
+
"learning_rate": 3.635529467544696e-09,
|
| 1916 |
+
"loss": 0.8066,
|
| 1917 |
"step": 24400
|
| 1918 |
},
|
| 1919 |
{
|
| 1920 |
+
"epoch": 1.06784,
|
| 1921 |
+
"grad_norm": 0.9749945998191833,
|
| 1922 |
+
"learning_rate": 2.526727698227288e-09,
|
| 1923 |
+
"loss": 0.802,
|
| 1924 |
"step": 24500
|
| 1925 |
},
|
| 1926 |
{
|
| 1927 |
+
"epoch": 1.07184,
|
| 1928 |
+
"grad_norm": 0.919170081615448,
|
| 1929 |
+
"learning_rate": 1.6189160709680074e-09,
|
| 1930 |
+
"loss": 0.79,
|
| 1931 |
"step": 24600
|
| 1932 |
},
|
| 1933 |
{
|
| 1934 |
+
"epoch": 1.07584,
|
| 1935 |
+
"grad_norm": 0.9579231142997742,
|
| 1936 |
+
"learning_rate": 9.122408503739466e-10,
|
| 1937 |
+
"loss": 0.8092,
|
| 1938 |
"step": 24700
|
| 1939 |
},
|
| 1940 |
{
|
| 1941 |
+
"epoch": 1.07984,
|
| 1942 |
+
"grad_norm": 0.8257275223731995,
|
| 1943 |
+
"learning_rate": 4.0681589439789395e-10,
|
| 1944 |
+
"loss": 0.8028,
|
| 1945 |
"step": 24800
|
| 1946 |
},
|
| 1947 |
{
|
| 1948 |
+
"epoch": 1.08384,
|
| 1949 |
+
"grad_norm": 0.8641030788421631,
|
| 1950 |
+
"learning_rate": 1.0272263599411803e-10,
|
| 1951 |
+
"loss": 0.7852,
|
| 1952 |
"step": 24900
|
| 1953 |
},
|
| 1954 |
{
|
| 1955 |
+
"epoch": 1.08784,
|
| 1956 |
+
"grad_norm": 0.929093062877655,
|
| 1957 |
+
"learning_rate": 1.006999733599301e-14,
|
| 1958 |
+
"loss": 0.7867,
|
| 1959 |
"step": 25000
|
| 1960 |
},
|
| 1961 |
{
|
| 1962 |
+
"epoch": 1.08784,
|
| 1963 |
+
"eval_loss": 1.5288399457931519,
|
| 1964 |
+
"eval_runtime": 99.9402,
|
| 1965 |
+
"eval_samples_per_second": 137.032,
|
| 1966 |
+
"eval_steps_per_second": 2.141,
|
| 1967 |
"step": 25000
|
| 1968 |
}
|
| 1969 |
],
|
| 1970 |
"logging_steps": 100,
|
| 1971 |
+
"max_steps": 25000,
|
| 1972 |
"num_input_tokens_seen": 0,
|
| 1973 |
"num_train_epochs": 9223372036854775807,
|
| 1974 |
"save_steps": 5000,
|
|
|
|
| 1979 |
"should_evaluate": false,
|
| 1980 |
"should_log": false,
|
| 1981 |
"should_save": true,
|
| 1982 |
+
"should_training_stop": true
|
| 1983 |
},
|
| 1984 |
"attributes": {}
|
| 1985 |
}
|
| 1986 |
},
|
| 1987 |
+
"total_flos": 3.3846277778817024e+18,
|
| 1988 |
"train_batch_size": 64,
|
| 1989 |
"trial_name": null,
|
| 1990 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5841
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8468e002e14d69abcb2f7de8e401f9fa2561c9e9f59ee528d9d623ec438f38ae
|
| 3 |
size 5841
|