Training in progress, epoch 1, checkpoint
Browse files- last-checkpoint/config.json +1 -1
- last-checkpoint/model.safetensors +1 -1
- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +280 -1956
- last-checkpoint/training_args.bin +1 -1
last-checkpoint/config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"architectures": [
|
| 4 |
"T5ForConditionalGeneration"
|
| 5 |
],
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "dq158/finalMeat",
|
| 3 |
"architectures": [
|
| 4 |
"T5ForConditionalGeneration"
|
| 5 |
],
|
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3132668808
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8f858a5136dc0160c7a812fe5b8bc3292741ce22b781f9b25d06a5f83f9e844
|
| 3 |
size 3132668808
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 6265677800
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec4b854adc7be93ea54cb3ba20a1f58e1d0c24fae05ad270bc9b5bc179cdcb59
|
| 3 |
size 6265677800
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b93bcac787dd1316177bf9f0cedd7b77d1cea8c71476888b60d1fec88f24fd8a
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:917b485f18156b96c93bc0dd12087e5d8b63a3098bce56ebac8eace0334f00f3
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,827 +1,827 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.01,
|
| 13 |
-
"learning_rate":
|
| 14 |
-
"loss":
|
| 15 |
"step": 500
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.01,
|
| 19 |
-
"learning_rate":
|
| 20 |
-
"loss":
|
| 21 |
"step": 1000
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.02,
|
| 25 |
-
"learning_rate":
|
| 26 |
-
"loss":
|
| 27 |
"step": 1500
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.03,
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss":
|
| 33 |
"step": 2000
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.04,
|
| 37 |
-
"learning_rate":
|
| 38 |
-
"loss":
|
| 39 |
"step": 2500
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.04,
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss":
|
| 45 |
"step": 3000
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.05,
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss":
|
| 51 |
"step": 3500
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.06,
|
| 55 |
-
"learning_rate":
|
| 56 |
-
"loss":
|
| 57 |
"step": 4000
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.07,
|
| 61 |
-
"learning_rate":
|
| 62 |
-
"loss":
|
| 63 |
"step": 4500
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.07,
|
| 67 |
-
"learning_rate":
|
| 68 |
-
"loss":
|
| 69 |
"step": 5000
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.08,
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss":
|
| 75 |
"step": 5500
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.09,
|
| 79 |
-
"learning_rate":
|
| 80 |
-
"loss":
|
| 81 |
"step": 6000
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.1,
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss":
|
| 87 |
"step": 6500
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.1,
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss":
|
| 93 |
"step": 7000
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.11,
|
| 97 |
-
"learning_rate":
|
| 98 |
-
"loss":
|
| 99 |
"step": 7500
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.12,
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss":
|
| 105 |
"step": 8000
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.12,
|
| 109 |
-
"learning_rate":
|
| 110 |
-
"loss":
|
| 111 |
"step": 8500
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.13,
|
| 115 |
-
"learning_rate":
|
| 116 |
-
"loss":
|
| 117 |
"step": 9000
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.14,
|
| 121 |
-
"learning_rate":
|
| 122 |
-
"loss":
|
| 123 |
"step": 9500
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 0.15,
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss":
|
| 129 |
"step": 10000
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.15,
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss":
|
| 135 |
"step": 10500
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.16,
|
| 139 |
-
"learning_rate":
|
| 140 |
-
"loss":
|
| 141 |
"step": 11000
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.17,
|
| 145 |
-
"learning_rate":
|
| 146 |
-
"loss":
|
| 147 |
"step": 11500
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.18,
|
| 151 |
-
"learning_rate":
|
| 152 |
-
"loss":
|
| 153 |
"step": 12000
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.18,
|
| 157 |
-
"learning_rate":
|
| 158 |
-
"loss":
|
| 159 |
"step": 12500
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.19,
|
| 163 |
-
"learning_rate":
|
| 164 |
-
"loss":
|
| 165 |
"step": 13000
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 0.2,
|
| 169 |
-
"learning_rate":
|
| 170 |
-
"loss":
|
| 171 |
"step": 13500
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.21,
|
| 175 |
-
"learning_rate":
|
| 176 |
-
"loss":
|
| 177 |
"step": 14000
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.21,
|
| 181 |
-
"learning_rate":
|
| 182 |
-
"loss":
|
| 183 |
"step": 14500
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.22,
|
| 187 |
-
"learning_rate":
|
| 188 |
-
"loss":
|
| 189 |
"step": 15000
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.23,
|
| 193 |
-
"learning_rate":
|
| 194 |
-
"loss":
|
| 195 |
"step": 15500
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 0.23,
|
| 199 |
-
"learning_rate":
|
| 200 |
-
"loss":
|
| 201 |
"step": 16000
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 0.24,
|
| 205 |
-
"learning_rate":
|
| 206 |
-
"loss":
|
| 207 |
"step": 16500
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 0.25,
|
| 211 |
-
"learning_rate":
|
| 212 |
-
"loss":
|
| 213 |
"step": 17000
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.26,
|
| 217 |
-
"learning_rate":
|
| 218 |
-
"loss":
|
| 219 |
"step": 17500
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.26,
|
| 223 |
-
"learning_rate":
|
| 224 |
-
"loss":
|
| 225 |
"step": 18000
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.27,
|
| 229 |
-
"learning_rate":
|
| 230 |
-
"loss":
|
| 231 |
"step": 18500
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.28,
|
| 235 |
-
"learning_rate":
|
| 236 |
-
"loss":
|
| 237 |
"step": 19000
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.29,
|
| 241 |
-
"learning_rate":
|
| 242 |
-
"loss":
|
| 243 |
"step": 19500
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.29,
|
| 247 |
-
"learning_rate":
|
| 248 |
-
"loss":
|
| 249 |
"step": 20000
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.3,
|
| 253 |
-
"learning_rate":
|
| 254 |
-
"loss":
|
| 255 |
"step": 20500
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.31,
|
| 259 |
-
"learning_rate":
|
| 260 |
-
"loss":
|
| 261 |
"step": 21000
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.32,
|
| 265 |
-
"learning_rate":
|
| 266 |
-
"loss":
|
| 267 |
"step": 21500
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 0.32,
|
| 271 |
-
"learning_rate":
|
| 272 |
-
"loss":
|
| 273 |
"step": 22000
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 0.33,
|
| 277 |
-
"learning_rate":
|
| 278 |
-
"loss":
|
| 279 |
"step": 22500
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.34,
|
| 283 |
-
"learning_rate":
|
| 284 |
-
"loss":
|
| 285 |
"step": 23000
|
| 286 |
},
|
| 287 |
{
|
| 288 |
"epoch": 0.34,
|
| 289 |
-
"learning_rate":
|
| 290 |
-
"loss":
|
| 291 |
"step": 23500
|
| 292 |
},
|
| 293 |
{
|
| 294 |
"epoch": 0.35,
|
| 295 |
-
"learning_rate":
|
| 296 |
-
"loss":
|
| 297 |
"step": 24000
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.36,
|
| 301 |
-
"learning_rate":
|
| 302 |
-
"loss":
|
| 303 |
"step": 24500
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"epoch": 0.37,
|
| 307 |
-
"learning_rate":
|
| 308 |
-
"loss":
|
| 309 |
"step": 25000
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.37,
|
| 313 |
-
"learning_rate":
|
| 314 |
-
"loss":
|
| 315 |
"step": 25500
|
| 316 |
},
|
| 317 |
{
|
| 318 |
"epoch": 0.38,
|
| 319 |
-
"learning_rate":
|
| 320 |
-
"loss":
|
| 321 |
"step": 26000
|
| 322 |
},
|
| 323 |
{
|
| 324 |
"epoch": 0.39,
|
| 325 |
-
"learning_rate":
|
| 326 |
-
"loss":
|
| 327 |
"step": 26500
|
| 328 |
},
|
| 329 |
{
|
| 330 |
"epoch": 0.4,
|
| 331 |
-
"learning_rate":
|
| 332 |
-
"loss":
|
| 333 |
"step": 27000
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.4,
|
| 337 |
-
"learning_rate":
|
| 338 |
-
"loss":
|
| 339 |
"step": 27500
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.41,
|
| 343 |
-
"learning_rate":
|
| 344 |
-
"loss":
|
| 345 |
"step": 28000
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"epoch": 0.42,
|
| 349 |
-
"learning_rate":
|
| 350 |
-
"loss":
|
| 351 |
"step": 28500
|
| 352 |
},
|
| 353 |
{
|
| 354 |
"epoch": 0.43,
|
| 355 |
-
"learning_rate":
|
| 356 |
-
"loss":
|
| 357 |
"step": 29000
|
| 358 |
},
|
| 359 |
{
|
| 360 |
"epoch": 0.43,
|
| 361 |
-
"learning_rate":
|
| 362 |
-
"loss":
|
| 363 |
"step": 29500
|
| 364 |
},
|
| 365 |
{
|
| 366 |
"epoch": 0.44,
|
| 367 |
-
"learning_rate":
|
| 368 |
-
"loss":
|
| 369 |
"step": 30000
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.45,
|
| 373 |
-
"learning_rate":
|
| 374 |
-
"loss":
|
| 375 |
"step": 30500
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.45,
|
| 379 |
-
"learning_rate":
|
| 380 |
-
"loss":
|
| 381 |
"step": 31000
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.46,
|
| 385 |
-
"learning_rate":
|
| 386 |
-
"loss":
|
| 387 |
"step": 31500
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"epoch": 0.47,
|
| 391 |
-
"learning_rate":
|
| 392 |
-
"loss":
|
| 393 |
"step": 32000
|
| 394 |
},
|
| 395 |
{
|
| 396 |
"epoch": 0.48,
|
| 397 |
-
"learning_rate":
|
| 398 |
-
"loss":
|
| 399 |
"step": 32500
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.48,
|
| 403 |
-
"learning_rate":
|
| 404 |
-
"loss":
|
| 405 |
"step": 33000
|
| 406 |
},
|
| 407 |
{
|
| 408 |
"epoch": 0.49,
|
| 409 |
-
"learning_rate":
|
| 410 |
-
"loss":
|
| 411 |
"step": 33500
|
| 412 |
},
|
| 413 |
{
|
| 414 |
"epoch": 0.5,
|
| 415 |
-
"learning_rate":
|
| 416 |
-
"loss":
|
| 417 |
"step": 34000
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.51,
|
| 421 |
-
"learning_rate":
|
| 422 |
-
"loss":
|
| 423 |
"step": 34500
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.51,
|
| 427 |
-
"learning_rate":
|
| 428 |
-
"loss":
|
| 429 |
"step": 35000
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.52,
|
| 433 |
-
"learning_rate":
|
| 434 |
-
"loss":
|
| 435 |
"step": 35500
|
| 436 |
},
|
| 437 |
{
|
| 438 |
"epoch": 0.53,
|
| 439 |
-
"learning_rate":
|
| 440 |
-
"loss":
|
| 441 |
"step": 36000
|
| 442 |
},
|
| 443 |
{
|
| 444 |
"epoch": 0.54,
|
| 445 |
-
"learning_rate":
|
| 446 |
-
"loss":
|
| 447 |
"step": 36500
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"epoch": 0.54,
|
| 451 |
-
"learning_rate":
|
| 452 |
-
"loss":
|
| 453 |
"step": 37000
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 0.55,
|
| 457 |
-
"learning_rate":
|
| 458 |
-
"loss":
|
| 459 |
"step": 37500
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.56,
|
| 463 |
-
"learning_rate":
|
| 464 |
-
"loss":
|
| 465 |
"step": 38000
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.56,
|
| 469 |
-
"learning_rate":
|
| 470 |
-
"loss":
|
| 471 |
"step": 38500
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"epoch": 0.57,
|
| 475 |
-
"learning_rate":
|
| 476 |
-
"loss":
|
| 477 |
"step": 39000
|
| 478 |
},
|
| 479 |
{
|
| 480 |
"epoch": 0.58,
|
| 481 |
-
"learning_rate":
|
| 482 |
-
"loss":
|
| 483 |
"step": 39500
|
| 484 |
},
|
| 485 |
{
|
| 486 |
"epoch": 0.59,
|
| 487 |
-
"learning_rate":
|
| 488 |
-
"loss":
|
| 489 |
"step": 40000
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.59,
|
| 493 |
-
"learning_rate":
|
| 494 |
-
"loss":
|
| 495 |
"step": 40500
|
| 496 |
},
|
| 497 |
{
|
| 498 |
"epoch": 0.6,
|
| 499 |
-
"learning_rate":
|
| 500 |
-
"loss":
|
| 501 |
"step": 41000
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.61,
|
| 505 |
-
"learning_rate":
|
| 506 |
-
"loss":
|
| 507 |
"step": 41500
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.62,
|
| 511 |
-
"learning_rate":
|
| 512 |
-
"loss":
|
| 513 |
"step": 42000
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 0.62,
|
| 517 |
-
"learning_rate":
|
| 518 |
-
"loss":
|
| 519 |
"step": 42500
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.63,
|
| 523 |
-
"learning_rate":
|
| 524 |
-
"loss":
|
| 525 |
"step": 43000
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 0.64,
|
| 529 |
-
"learning_rate":
|
| 530 |
-
"loss":
|
| 531 |
"step": 43500
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 0.64,
|
| 535 |
-
"learning_rate":
|
| 536 |
-
"loss":
|
| 537 |
"step": 44000
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.65,
|
| 541 |
-
"learning_rate":
|
| 542 |
-
"loss":
|
| 543 |
"step": 44500
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.66,
|
| 547 |
-
"learning_rate":
|
| 548 |
-
"loss":
|
| 549 |
"step": 45000
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.67,
|
| 553 |
-
"learning_rate":
|
| 554 |
-
"loss":
|
| 555 |
"step": 45500
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 0.67,
|
| 559 |
-
"learning_rate":
|
| 560 |
-
"loss":
|
| 561 |
"step": 46000
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 0.68,
|
| 565 |
-
"learning_rate":
|
| 566 |
-
"loss":
|
| 567 |
"step": 46500
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.69,
|
| 571 |
-
"learning_rate":
|
| 572 |
-
"loss":
|
| 573 |
"step": 47000
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 0.7,
|
| 577 |
-
"learning_rate":
|
| 578 |
-
"loss":
|
| 579 |
"step": 47500
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.7,
|
| 583 |
-
"learning_rate":
|
| 584 |
-
"loss":
|
| 585 |
"step": 48000
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 0.71,
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss":
|
| 591 |
"step": 48500
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.72,
|
| 595 |
-
"learning_rate":
|
| 596 |
-
"loss":
|
| 597 |
"step": 49000
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.73,
|
| 601 |
-
"learning_rate":
|
| 602 |
-
"loss":
|
| 603 |
"step": 49500
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 0.73,
|
| 607 |
-
"learning_rate":
|
| 608 |
-
"loss":
|
| 609 |
"step": 50000
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.74,
|
| 613 |
-
"learning_rate":
|
| 614 |
-
"loss":
|
| 615 |
"step": 50500
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.75,
|
| 619 |
-
"learning_rate":
|
| 620 |
-
"loss":
|
| 621 |
"step": 51000
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 0.75,
|
| 625 |
-
"learning_rate":
|
| 626 |
-
"loss":
|
| 627 |
"step": 51500
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 0.76,
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss":
|
| 633 |
"step": 52000
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.77,
|
| 637 |
-
"learning_rate":
|
| 638 |
-
"loss":
|
| 639 |
"step": 52500
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.78,
|
| 643 |
-
"learning_rate":
|
| 644 |
-
"loss":
|
| 645 |
"step": 53000
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 0.78,
|
| 649 |
-
"learning_rate":
|
| 650 |
-
"loss":
|
| 651 |
"step": 53500
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 0.79,
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss":
|
| 657 |
"step": 54000
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.8,
|
| 661 |
-
"learning_rate":
|
| 662 |
-
"loss":
|
| 663 |
"step": 54500
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 0.81,
|
| 667 |
-
"learning_rate":
|
| 668 |
-
"loss":
|
| 669 |
"step": 55000
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.81,
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss":
|
| 675 |
"step": 55500
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.82,
|
| 679 |
-
"learning_rate":
|
| 680 |
-
"loss":
|
| 681 |
"step": 56000
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 0.83,
|
| 685 |
-
"learning_rate":
|
| 686 |
-
"loss":
|
| 687 |
"step": 56500
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 0.84,
|
| 691 |
-
"learning_rate":
|
| 692 |
-
"loss":
|
| 693 |
"step": 57000
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 0.84,
|
| 697 |
-
"learning_rate":
|
| 698 |
-
"loss":
|
| 699 |
"step": 57500
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.85,
|
| 703 |
-
"learning_rate":
|
| 704 |
-
"loss":
|
| 705 |
"step": 58000
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 0.86,
|
| 709 |
-
"learning_rate":
|
| 710 |
-
"loss":
|
| 711 |
"step": 58500
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 0.86,
|
| 715 |
-
"learning_rate":
|
| 716 |
-
"loss":
|
| 717 |
"step": 59000
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 0.87,
|
| 721 |
-
"learning_rate":
|
| 722 |
-
"loss":
|
| 723 |
"step": 59500
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 0.88,
|
| 727 |
-
"learning_rate":
|
| 728 |
-
"loss":
|
| 729 |
"step": 60000
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.89,
|
| 733 |
-
"learning_rate":
|
| 734 |
-
"loss":
|
| 735 |
"step": 60500
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 0.89,
|
| 739 |
-
"learning_rate":
|
| 740 |
-
"loss":
|
| 741 |
"step": 61000
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 0.9,
|
| 745 |
-
"learning_rate":
|
| 746 |
-
"loss":
|
| 747 |
"step": 61500
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 0.91,
|
| 751 |
-
"learning_rate":
|
| 752 |
-
"loss":
|
| 753 |
"step": 62000
|
| 754 |
},
|
| 755 |
{
|
| 756 |
"epoch": 0.92,
|
| 757 |
-
"learning_rate":
|
| 758 |
-
"loss":
|
| 759 |
"step": 62500
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.92,
|
| 763 |
-
"learning_rate":
|
| 764 |
-
"loss":
|
| 765 |
"step": 63000
|
| 766 |
},
|
| 767 |
{
|
| 768 |
"epoch": 0.93,
|
| 769 |
-
"learning_rate":
|
| 770 |
-
"loss":
|
| 771 |
"step": 63500
|
| 772 |
},
|
| 773 |
{
|
| 774 |
"epoch": 0.94,
|
| 775 |
-
"learning_rate":
|
| 776 |
-
"loss":
|
| 777 |
"step": 64000
|
| 778 |
},
|
| 779 |
{
|
| 780 |
"epoch": 0.95,
|
| 781 |
-
"learning_rate":
|
| 782 |
-
"loss":
|
| 783 |
"step": 64500
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 0.95,
|
| 787 |
-
"learning_rate":
|
| 788 |
-
"loss": 2.
|
| 789 |
"step": 65000
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 0.96,
|
| 793 |
-
"learning_rate":
|
| 794 |
-
"loss":
|
| 795 |
"step": 65500
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 0.97,
|
| 799 |
-
"learning_rate":
|
| 800 |
-
"loss":
|
| 801 |
"step": 66000
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 0.97,
|
| 805 |
-
"learning_rate":
|
| 806 |
-
"loss":
|
| 807 |
"step": 66500
|
| 808 |
},
|
| 809 |
{
|
| 810 |
"epoch": 0.98,
|
| 811 |
-
"learning_rate":
|
| 812 |
-
"loss":
|
| 813 |
"step": 67000
|
| 814 |
},
|
| 815 |
{
|
| 816 |
"epoch": 0.99,
|
| 817 |
-
"learning_rate":
|
| 818 |
-
"loss":
|
| 819 |
"step": 67500
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 1.0,
|
| 823 |
-
"learning_rate":
|
| 824 |
-
"loss":
|
| 825 |
"step": 68000
|
| 826 |
},
|
| 827 |
{
|
|
@@ -829,7 +829,7 @@
|
|
| 829 |
"eval_bleu": 1.0,
|
| 830 |
"eval_brevity_penalty": 1.0,
|
| 831 |
"eval_length_ratio": 1.0,
|
| 832 |
-
"eval_loss": 2.
|
| 833 |
"eval_precisions": [
|
| 834 |
1.0,
|
| 835 |
1.0,
|
|
@@ -837,1694 +837,18 @@
|
|
| 837 |
1.0
|
| 838 |
],
|
| 839 |
"eval_reference_length": 7761920,
|
| 840 |
-
"eval_runtime":
|
| 841 |
-
"eval_samples_per_second": 0.
|
| 842 |
-
"eval_steps_per_second": 0.
|
| 843 |
"eval_translation_length": 7761920,
|
| 844 |
"step": 68219
|
| 845 |
-
},
|
| 846 |
-
{
|
| 847 |
-
"epoch": 1.0,
|
| 848 |
-
"learning_rate": 9.048308407381583e-05,
|
| 849 |
-
"loss": 2.9246,
|
| 850 |
-
"step": 68500
|
| 851 |
-
},
|
| 852 |
-
{
|
| 853 |
-
"epoch": 1.01,
|
| 854 |
-
"learning_rate": 9.034731799405654e-05,
|
| 855 |
-
"loss": 2.8047,
|
| 856 |
-
"step": 69000
|
| 857 |
-
},
|
| 858 |
-
{
|
| 859 |
-
"epoch": 1.02,
|
| 860 |
-
"learning_rate": 9.021069373680653e-05,
|
| 861 |
-
"loss": 2.8717,
|
| 862 |
-
"step": 69500
|
| 863 |
-
},
|
| 864 |
-
{
|
| 865 |
-
"epoch": 1.03,
|
| 866 |
-
"learning_rate": 9.007321420803e-05,
|
| 867 |
-
"loss": 2.8453,
|
| 868 |
-
"step": 70000
|
| 869 |
-
},
|
| 870 |
-
{
|
| 871 |
-
"epoch": 1.03,
|
| 872 |
-
"learning_rate": 8.993488233188263e-05,
|
| 873 |
-
"loss": 2.8545,
|
| 874 |
-
"step": 70500
|
| 875 |
-
},
|
| 876 |
-
{
|
| 877 |
-
"epoch": 1.04,
|
| 878 |
-
"learning_rate": 8.979570105064923e-05,
|
| 879 |
-
"loss": 2.7871,
|
| 880 |
-
"step": 71000
|
| 881 |
-
},
|
| 882 |
-
{
|
| 883 |
-
"epoch": 1.05,
|
| 884 |
-
"learning_rate": 8.965567332468128e-05,
|
| 885 |
-
"loss": 2.772,
|
| 886 |
-
"step": 71500
|
| 887 |
-
},
|
| 888 |
-
{
|
| 889 |
-
"epoch": 1.06,
|
| 890 |
-
"learning_rate": 8.951480213233397e-05,
|
| 891 |
-
"loss": 2.915,
|
| 892 |
-
"step": 72000
|
| 893 |
-
},
|
| 894 |
-
{
|
| 895 |
-
"epoch": 1.06,
|
| 896 |
-
"learning_rate": 8.937309046990271e-05,
|
| 897 |
-
"loss": 2.8076,
|
| 898 |
-
"step": 72500
|
| 899 |
-
},
|
| 900 |
-
{
|
| 901 |
-
"epoch": 1.07,
|
| 902 |
-
"learning_rate": 8.923054135155963e-05,
|
| 903 |
-
"loss": 2.693,
|
| 904 |
-
"step": 73000
|
| 905 |
-
},
|
| 906 |
-
{
|
| 907 |
-
"epoch": 1.08,
|
| 908 |
-
"learning_rate": 8.908715780928925e-05,
|
| 909 |
-
"loss": 2.8504,
|
| 910 |
-
"step": 73500
|
| 911 |
-
},
|
| 912 |
-
{
|
| 913 |
-
"epoch": 1.08,
|
| 914 |
-
"learning_rate": 8.89429428928241e-05,
|
| 915 |
-
"loss": 2.7593,
|
| 916 |
-
"step": 74000
|
| 917 |
-
},
|
| 918 |
-
{
|
| 919 |
-
"epoch": 1.09,
|
| 920 |
-
"learning_rate": 8.879789966957988e-05,
|
| 921 |
-
"loss": 2.7953,
|
| 922 |
-
"step": 74500
|
| 923 |
-
},
|
| 924 |
-
{
|
| 925 |
-
"epoch": 1.1,
|
| 926 |
-
"learning_rate": 8.865203122459008e-05,
|
| 927 |
-
"loss": 2.7169,
|
| 928 |
-
"step": 75000
|
| 929 |
-
},
|
| 930 |
-
{
|
| 931 |
-
"epoch": 1.11,
|
| 932 |
-
"learning_rate": 8.850534066044054e-05,
|
| 933 |
-
"loss": 2.6889,
|
| 934 |
-
"step": 75500
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 1.11,
|
| 938 |
-
"learning_rate": 8.835783109720333e-05,
|
| 939 |
-
"loss": 2.8703,
|
| 940 |
-
"step": 76000
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 1.12,
|
| 944 |
-
"learning_rate": 8.820950567237043e-05,
|
| 945 |
-
"loss": 2.8989,
|
| 946 |
-
"step": 76500
|
| 947 |
-
},
|
| 948 |
-
{
|
| 949 |
-
"epoch": 1.13,
|
| 950 |
-
"learning_rate": 8.806036754078701e-05,
|
| 951 |
-
"loss": 2.7834,
|
| 952 |
-
"step": 77000
|
| 953 |
-
},
|
| 954 |
-
{
|
| 955 |
-
"epoch": 1.14,
|
| 956 |
-
"learning_rate": 8.791041987458431e-05,
|
| 957 |
-
"loss": 2.84,
|
| 958 |
-
"step": 77500
|
| 959 |
-
},
|
| 960 |
-
{
|
| 961 |
-
"epoch": 1.14,
|
| 962 |
-
"learning_rate": 8.775966586311212e-05,
|
| 963 |
-
"loss": 2.6913,
|
| 964 |
-
"step": 78000
|
| 965 |
-
},
|
| 966 |
-
{
|
| 967 |
-
"epoch": 1.15,
|
| 968 |
-
"learning_rate": 8.760810871287108e-05,
|
| 969 |
-
"loss": 2.7712,
|
| 970 |
-
"step": 78500
|
| 971 |
-
},
|
| 972 |
-
{
|
| 973 |
-
"epoch": 1.16,
|
| 974 |
-
"learning_rate": 8.745575164744435e-05,
|
| 975 |
-
"loss": 2.8229,
|
| 976 |
-
"step": 79000
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 1.17,
|
| 980 |
-
"learning_rate": 8.730259790742906e-05,
|
| 981 |
-
"loss": 2.8877,
|
| 982 |
-
"step": 79500
|
| 983 |
-
},
|
| 984 |
-
{
|
| 985 |
-
"epoch": 1.17,
|
| 986 |
-
"learning_rate": 8.714865075036746e-05,
|
| 987 |
-
"loss": 2.7965,
|
| 988 |
-
"step": 80000
|
| 989 |
-
},
|
| 990 |
-
{
|
| 991 |
-
"epoch": 1.18,
|
| 992 |
-
"learning_rate": 8.699391345067759e-05,
|
| 993 |
-
"loss": 2.9252,
|
| 994 |
-
"step": 80500
|
| 995 |
-
},
|
| 996 |
-
{
|
| 997 |
-
"epoch": 1.19,
|
| 998 |
-
"learning_rate": 8.683838929958356e-05,
|
| 999 |
-
"loss": 2.8585,
|
| 1000 |
-
"step": 81000
|
| 1001 |
-
},
|
| 1002 |
-
{
|
| 1003 |
-
"epoch": 1.19,
|
| 1004 |
-
"learning_rate": 8.66820816050457e-05,
|
| 1005 |
-
"loss": 2.7542,
|
| 1006 |
-
"step": 81500
|
| 1007 |
-
},
|
| 1008 |
-
{
|
| 1009 |
-
"epoch": 1.2,
|
| 1010 |
-
"learning_rate": 8.652499369169005e-05,
|
| 1011 |
-
"loss": 2.8124,
|
| 1012 |
-
"step": 82000
|
| 1013 |
-
},
|
| 1014 |
-
{
|
| 1015 |
-
"epoch": 1.21,
|
| 1016 |
-
"learning_rate": 8.636712890073772e-05,
|
| 1017 |
-
"loss": 2.8251,
|
| 1018 |
-
"step": 82500
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 1.22,
|
| 1022 |
-
"learning_rate": 8.62084905899339e-05,
|
| 1023 |
-
"loss": 2.7148,
|
| 1024 |
-
"step": 83000
|
| 1025 |
-
},
|
| 1026 |
-
{
|
| 1027 |
-
"epoch": 1.22,
|
| 1028 |
-
"learning_rate": 8.604908213347622e-05,
|
| 1029 |
-
"loss": 2.8276,
|
| 1030 |
-
"step": 83500
|
| 1031 |
-
},
|
| 1032 |
-
{
|
| 1033 |
-
"epoch": 1.23,
|
| 1034 |
-
"learning_rate": 8.58889069219432e-05,
|
| 1035 |
-
"loss": 2.8799,
|
| 1036 |
-
"step": 84000
|
| 1037 |
-
},
|
| 1038 |
-
{
|
| 1039 |
-
"epoch": 1.24,
|
| 1040 |
-
"learning_rate": 8.572796836222206e-05,
|
| 1041 |
-
"loss": 2.7959,
|
| 1042 |
-
"step": 84500
|
| 1043 |
-
},
|
| 1044 |
-
{
|
| 1045 |
-
"epoch": 1.25,
|
| 1046 |
-
"learning_rate": 8.556626987743621e-05,
|
| 1047 |
-
"loss": 2.756,
|
| 1048 |
-
"step": 85000
|
| 1049 |
-
},
|
| 1050 |
-
{
|
| 1051 |
-
"epoch": 1.25,
|
| 1052 |
-
"learning_rate": 8.54038149068725e-05,
|
| 1053 |
-
"loss": 2.8956,
|
| 1054 |
-
"step": 85500
|
| 1055 |
-
},
|
| 1056 |
-
{
|
| 1057 |
-
"epoch": 1.26,
|
| 1058 |
-
"learning_rate": 8.524060690590803e-05,
|
| 1059 |
-
"loss": 2.8619,
|
| 1060 |
-
"step": 86000
|
| 1061 |
-
},
|
| 1062 |
-
{
|
| 1063 |
-
"epoch": 1.27,
|
| 1064 |
-
"learning_rate": 8.507664934593668e-05,
|
| 1065 |
-
"loss": 2.8005,
|
| 1066 |
-
"step": 86500
|
| 1067 |
-
},
|
| 1068 |
-
{
|
| 1069 |
-
"epoch": 1.28,
|
| 1070 |
-
"learning_rate": 8.491194571429526e-05,
|
| 1071 |
-
"loss": 2.8213,
|
| 1072 |
-
"step": 87000
|
| 1073 |
-
},
|
| 1074 |
-
{
|
| 1075 |
-
"epoch": 1.28,
|
| 1076 |
-
"learning_rate": 8.474649951418936e-05,
|
| 1077 |
-
"loss": 2.8356,
|
| 1078 |
-
"step": 87500
|
| 1079 |
-
},
|
| 1080 |
-
{
|
| 1081 |
-
"epoch": 1.29,
|
| 1082 |
-
"learning_rate": 8.458031426461878e-05,
|
| 1083 |
-
"loss": 2.877,
|
| 1084 |
-
"step": 88000
|
| 1085 |
-
},
|
| 1086 |
-
{
|
| 1087 |
-
"epoch": 1.3,
|
| 1088 |
-
"learning_rate": 8.441339350030278e-05,
|
| 1089 |
-
"loss": 2.9189,
|
| 1090 |
-
"step": 88500
|
| 1091 |
-
},
|
| 1092 |
-
{
|
| 1093 |
-
"epoch": 1.3,
|
| 1094 |
-
"learning_rate": 8.424574077160476e-05,
|
| 1095 |
-
"loss": 2.8497,
|
| 1096 |
-
"step": 89000
|
| 1097 |
-
},
|
| 1098 |
-
{
|
| 1099 |
-
"epoch": 1.31,
|
| 1100 |
-
"learning_rate": 8.407735964445689e-05,
|
| 1101 |
-
"loss": 2.807,
|
| 1102 |
-
"step": 89500
|
| 1103 |
-
},
|
| 1104 |
-
{
|
| 1105 |
-
"epoch": 1.32,
|
| 1106 |
-
"learning_rate": 8.390825370028414e-05,
|
| 1107 |
-
"loss": 2.8745,
|
| 1108 |
-
"step": 90000
|
| 1109 |
-
},
|
| 1110 |
-
{
|
| 1111 |
-
"epoch": 1.33,
|
| 1112 |
-
"learning_rate": 8.373842653592818e-05,
|
| 1113 |
-
"loss": 2.8054,
|
| 1114 |
-
"step": 90500
|
| 1115 |
-
},
|
| 1116 |
-
{
|
| 1117 |
-
"epoch": 1.33,
|
| 1118 |
-
"learning_rate": 8.35678817635709e-05,
|
| 1119 |
-
"loss": 2.7019,
|
| 1120 |
-
"step": 91000
|
| 1121 |
-
},
|
| 1122 |
-
{
|
| 1123 |
-
"epoch": 1.34,
|
| 1124 |
-
"learning_rate": 8.339662301065747e-05,
|
| 1125 |
-
"loss": 2.8497,
|
| 1126 |
-
"step": 91500
|
| 1127 |
-
},
|
| 1128 |
-
{
|
| 1129 |
-
"epoch": 1.35,
|
| 1130 |
-
"learning_rate": 8.322465391981927e-05,
|
| 1131 |
-
"loss": 2.8442,
|
| 1132 |
-
"step": 92000
|
| 1133 |
-
},
|
| 1134 |
-
{
|
| 1135 |
-
"epoch": 1.36,
|
| 1136 |
-
"learning_rate": 8.30519781487964e-05,
|
| 1137 |
-
"loss": 2.7681,
|
| 1138 |
-
"step": 92500
|
| 1139 |
-
},
|
| 1140 |
-
{
|
| 1141 |
-
"epoch": 1.36,
|
| 1142 |
-
"learning_rate": 8.287859937035989e-05,
|
| 1143 |
-
"loss": 2.8219,
|
| 1144 |
-
"step": 93000
|
| 1145 |
-
},
|
| 1146 |
-
{
|
| 1147 |
-
"epoch": 1.37,
|
| 1148 |
-
"learning_rate": 8.270452127223352e-05,
|
| 1149 |
-
"loss": 2.8409,
|
| 1150 |
-
"step": 93500
|
| 1151 |
-
},
|
| 1152 |
-
{
|
| 1153 |
-
"epoch": 1.38,
|
| 1154 |
-
"learning_rate": 8.252974755701546e-05,
|
| 1155 |
-
"loss": 2.7358,
|
| 1156 |
-
"step": 94000
|
| 1157 |
-
},
|
| 1158 |
-
{
|
| 1159 |
-
"epoch": 1.39,
|
| 1160 |
-
"learning_rate": 8.235428194209947e-05,
|
| 1161 |
-
"loss": 2.6975,
|
| 1162 |
-
"step": 94500
|
| 1163 |
-
},
|
| 1164 |
-
{
|
| 1165 |
-
"epoch": 1.39,
|
| 1166 |
-
"learning_rate": 8.217812815959588e-05,
|
| 1167 |
-
"loss": 2.7972,
|
| 1168 |
-
"step": 95000
|
| 1169 |
-
},
|
| 1170 |
-
{
|
| 1171 |
-
"epoch": 1.4,
|
| 1172 |
-
"learning_rate": 8.200128995625211e-05,
|
| 1173 |
-
"loss": 2.8172,
|
| 1174 |
-
"step": 95500
|
| 1175 |
-
},
|
| 1176 |
-
{
|
| 1177 |
-
"epoch": 1.41,
|
| 1178 |
-
"learning_rate": 8.182377109337309e-05,
|
| 1179 |
-
"loss": 2.8309,
|
| 1180 |
-
"step": 96000
|
| 1181 |
-
},
|
| 1182 |
-
{
|
| 1183 |
-
"epoch": 1.41,
|
| 1184 |
-
"learning_rate": 8.164557534674122e-05,
|
| 1185 |
-
"loss": 2.9371,
|
| 1186 |
-
"step": 96500
|
| 1187 |
-
},
|
| 1188 |
-
{
|
| 1189 |
-
"epoch": 1.42,
|
| 1190 |
-
"learning_rate": 8.1466706506536e-05,
|
| 1191 |
-
"loss": 2.738,
|
| 1192 |
-
"step": 97000
|
| 1193 |
-
},
|
| 1194 |
-
{
|
| 1195 |
-
"epoch": 1.43,
|
| 1196 |
-
"learning_rate": 8.12871683772535e-05,
|
| 1197 |
-
"loss": 2.8989,
|
| 1198 |
-
"step": 97500
|
| 1199 |
-
},
|
| 1200 |
-
{
|
| 1201 |
-
"epoch": 1.44,
|
| 1202 |
-
"learning_rate": 8.110696477762538e-05,
|
| 1203 |
-
"loss": 2.7852,
|
| 1204 |
-
"step": 98000
|
| 1205 |
-
},
|
| 1206 |
-
{
|
| 1207 |
-
"epoch": 1.44,
|
| 1208 |
-
"learning_rate": 8.092609954053776e-05,
|
| 1209 |
-
"loss": 2.862,
|
| 1210 |
-
"step": 98500
|
| 1211 |
-
},
|
| 1212 |
-
{
|
| 1213 |
-
"epoch": 1.45,
|
| 1214 |
-
"learning_rate": 8.07445765129495e-05,
|
| 1215 |
-
"loss": 2.8234,
|
| 1216 |
-
"step": 99000
|
| 1217 |
-
},
|
| 1218 |
-
{
|
| 1219 |
-
"epoch": 1.46,
|
| 1220 |
-
"learning_rate": 8.056239955581064e-05,
|
| 1221 |
-
"loss": 2.844,
|
| 1222 |
-
"step": 99500
|
| 1223 |
-
},
|
| 1224 |
-
{
|
| 1225 |
-
"epoch": 1.47,
|
| 1226 |
-
"learning_rate": 8.037957254398004e-05,
|
| 1227 |
-
"loss": 2.7922,
|
| 1228 |
-
"step": 100000
|
| 1229 |
-
},
|
| 1230 |
-
{
|
| 1231 |
-
"epoch": 1.47,
|
| 1232 |
-
"learning_rate": 8.01960993661431e-05,
|
| 1233 |
-
"loss": 2.817,
|
| 1234 |
-
"step": 100500
|
| 1235 |
-
},
|
| 1236 |
-
{
|
| 1237 |
-
"epoch": 1.48,
|
| 1238 |
-
"learning_rate": 8.0011983924729e-05,
|
| 1239 |
-
"loss": 2.8955,
|
| 1240 |
-
"step": 101000
|
| 1241 |
-
},
|
| 1242 |
-
{
|
| 1243 |
-
"epoch": 1.49,
|
| 1244 |
-
"learning_rate": 7.982723013582772e-05,
|
| 1245 |
-
"loss": 2.8279,
|
| 1246 |
-
"step": 101500
|
| 1247 |
-
},
|
| 1248 |
-
{
|
| 1249 |
-
"epoch": 1.5,
|
| 1250 |
-
"learning_rate": 7.964184192910672e-05,
|
| 1251 |
-
"loss": 2.8259,
|
| 1252 |
-
"step": 102000
|
| 1253 |
-
},
|
| 1254 |
-
{
|
| 1255 |
-
"epoch": 1.5,
|
| 1256 |
-
"learning_rate": 7.94558232477274e-05,
|
| 1257 |
-
"loss": 2.9235,
|
| 1258 |
-
"step": 102500
|
| 1259 |
-
},
|
| 1260 |
-
{
|
| 1261 |
-
"epoch": 1.51,
|
| 1262 |
-
"learning_rate": 7.926917804826117e-05,
|
| 1263 |
-
"loss": 2.8261,
|
| 1264 |
-
"step": 103000
|
| 1265 |
-
},
|
| 1266 |
-
{
|
| 1267 |
-
"epoch": 1.52,
|
| 1268 |
-
"learning_rate": 7.908191030060532e-05,
|
| 1269 |
-
"loss": 2.7384,
|
| 1270 |
-
"step": 103500
|
| 1271 |
-
},
|
| 1272 |
-
{
|
| 1273 |
-
"epoch": 1.52,
|
| 1274 |
-
"learning_rate": 7.889402398789863e-05,
|
| 1275 |
-
"loss": 2.8064,
|
| 1276 |
-
"step": 104000
|
| 1277 |
-
},
|
| 1278 |
-
{
|
| 1279 |
-
"epoch": 1.53,
|
| 1280 |
-
"learning_rate": 7.870552310643656e-05,
|
| 1281 |
-
"loss": 2.8212,
|
| 1282 |
-
"step": 104500
|
| 1283 |
-
},
|
| 1284 |
-
{
|
| 1285 |
-
"epoch": 1.54,
|
| 1286 |
-
"learning_rate": 7.851641166558628e-05,
|
| 1287 |
-
"loss": 2.8605,
|
| 1288 |
-
"step": 105000
|
| 1289 |
-
},
|
| 1290 |
-
{
|
| 1291 |
-
"epoch": 1.55,
|
| 1292 |
-
"learning_rate": 7.832669368770149e-05,
|
| 1293 |
-
"loss": 2.7907,
|
| 1294 |
-
"step": 105500
|
| 1295 |
-
},
|
| 1296 |
-
{
|
| 1297 |
-
"epoch": 1.55,
|
| 1298 |
-
"learning_rate": 7.813637320803671e-05,
|
| 1299 |
-
"loss": 2.7278,
|
| 1300 |
-
"step": 106000
|
| 1301 |
-
},
|
| 1302 |
-
{
|
| 1303 |
-
"epoch": 1.56,
|
| 1304 |
-
"learning_rate": 7.794545427466153e-05,
|
| 1305 |
-
"loss": 2.7808,
|
| 1306 |
-
"step": 106500
|
| 1307 |
-
},
|
| 1308 |
-
{
|
| 1309 |
-
"epoch": 1.57,
|
| 1310 |
-
"learning_rate": 7.775394094837455e-05,
|
| 1311 |
-
"loss": 2.8643,
|
| 1312 |
-
"step": 107000
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 1.58,
|
| 1316 |
-
"learning_rate": 7.756183730261694e-05,
|
| 1317 |
-
"loss": 2.738,
|
| 1318 |
-
"step": 107500
|
| 1319 |
-
},
|
| 1320 |
-
{
|
| 1321 |
-
"epoch": 1.58,
|
| 1322 |
-
"learning_rate": 7.736914742338577e-05,
|
| 1323 |
-
"loss": 2.7307,
|
| 1324 |
-
"step": 108000
|
| 1325 |
-
},
|
| 1326 |
-
{
|
| 1327 |
-
"epoch": 1.59,
|
| 1328 |
-
"learning_rate": 7.71758754091472e-05,
|
| 1329 |
-
"loss": 2.8986,
|
| 1330 |
-
"step": 108500
|
| 1331 |
-
},
|
| 1332 |
-
{
|
| 1333 |
-
"epoch": 1.6,
|
| 1334 |
-
"learning_rate": 7.698202537074926e-05,
|
| 1335 |
-
"loss": 2.7493,
|
| 1336 |
-
"step": 109000
|
| 1337 |
-
},
|
| 1338 |
-
{
|
| 1339 |
-
"epoch": 1.61,
|
| 1340 |
-
"learning_rate": 7.678760143133436e-05,
|
| 1341 |
-
"loss": 2.8837,
|
| 1342 |
-
"step": 109500
|
| 1343 |
-
},
|
| 1344 |
-
{
|
| 1345 |
-
"epoch": 1.61,
|
| 1346 |
-
"learning_rate": 7.659260772625176e-05,
|
| 1347 |
-
"loss": 2.778,
|
| 1348 |
-
"step": 110000
|
| 1349 |
-
},
|
| 1350 |
-
{
|
| 1351 |
-
"epoch": 1.62,
|
| 1352 |
-
"learning_rate": 7.639704840296936e-05,
|
| 1353 |
-
"loss": 2.8215,
|
| 1354 |
-
"step": 110500
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 1.63,
|
| 1358 |
-
"learning_rate": 7.620092762098568e-05,
|
| 1359 |
-
"loss": 2.8389,
|
| 1360 |
-
"step": 111000
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 1.63,
|
| 1364 |
-
"learning_rate": 7.600424955174132e-05,
|
| 1365 |
-
"loss": 2.8627,
|
| 1366 |
-
"step": 111500
|
| 1367 |
-
},
|
| 1368 |
-
{
|
| 1369 |
-
"epoch": 1.64,
|
| 1370 |
-
"learning_rate": 7.580701837853028e-05,
|
| 1371 |
-
"loss": 2.8321,
|
| 1372 |
-
"step": 112000
|
| 1373 |
-
},
|
| 1374 |
-
{
|
| 1375 |
-
"epoch": 1.65,
|
| 1376 |
-
"learning_rate": 7.560923829641088e-05,
|
| 1377 |
-
"loss": 2.9496,
|
| 1378 |
-
"step": 112500
|
| 1379 |
-
},
|
| 1380 |
-
{
|
| 1381 |
-
"epoch": 1.66,
|
| 1382 |
-
"learning_rate": 7.541091351211666e-05,
|
| 1383 |
-
"loss": 2.8576,
|
| 1384 |
-
"step": 113000
|
| 1385 |
-
},
|
| 1386 |
-
{
|
| 1387 |
-
"epoch": 1.66,
|
| 1388 |
-
"learning_rate": 7.521204824396678e-05,
|
| 1389 |
-
"loss": 2.7548,
|
| 1390 |
-
"step": 113500
|
| 1391 |
-
},
|
| 1392 |
-
{
|
| 1393 |
-
"epoch": 1.67,
|
| 1394 |
-
"learning_rate": 7.501264672177637e-05,
|
| 1395 |
-
"loss": 2.7738,
|
| 1396 |
-
"step": 114000
|
| 1397 |
-
},
|
| 1398 |
-
{
|
| 1399 |
-
"epoch": 1.68,
|
| 1400 |
-
"learning_rate": 7.481271318676662e-05,
|
| 1401 |
-
"loss": 2.8088,
|
| 1402 |
-
"step": 114500
|
| 1403 |
-
},
|
| 1404 |
-
{
|
| 1405 |
-
"epoch": 1.69,
|
| 1406 |
-
"learning_rate": 7.46122518914744e-05,
|
| 1407 |
-
"loss": 2.8243,
|
| 1408 |
-
"step": 115000
|
| 1409 |
-
},
|
| 1410 |
-
{
|
| 1411 |
-
"epoch": 1.69,
|
| 1412 |
-
"learning_rate": 7.441126709966203e-05,
|
| 1413 |
-
"loss": 2.6794,
|
| 1414 |
-
"step": 115500
|
| 1415 |
-
},
|
| 1416 |
-
{
|
| 1417 |
-
"epoch": 1.7,
|
| 1418 |
-
"learning_rate": 7.420976308622632e-05,
|
| 1419 |
-
"loss": 2.8277,
|
| 1420 |
-
"step": 116000
|
| 1421 |
-
},
|
| 1422 |
-
{
|
| 1423 |
-
"epoch": 1.71,
|
| 1424 |
-
"learning_rate": 7.400774413710793e-05,
|
| 1425 |
-
"loss": 2.8341,
|
| 1426 |
-
"step": 116500
|
| 1427 |
-
},
|
| 1428 |
-
{
|
| 1429 |
-
"epoch": 1.72,
|
| 1430 |
-
"learning_rate": 7.380521454920001e-05,
|
| 1431 |
-
"loss": 2.8003,
|
| 1432 |
-
"step": 117000
|
| 1433 |
-
},
|
| 1434 |
-
{
|
| 1435 |
-
"epoch": 1.72,
|
| 1436 |
-
"learning_rate": 7.360217863025687e-05,
|
| 1437 |
-
"loss": 2.724,
|
| 1438 |
-
"step": 117500
|
| 1439 |
-
},
|
| 1440 |
-
{
|
| 1441 |
-
"epoch": 1.73,
|
| 1442 |
-
"learning_rate": 7.33986406988024e-05,
|
| 1443 |
-
"loss": 2.829,
|
| 1444 |
-
"step": 118000
|
| 1445 |
-
},
|
| 1446 |
-
{
|
| 1447 |
-
"epoch": 1.74,
|
| 1448 |
-
"learning_rate": 7.319460508403811e-05,
|
| 1449 |
-
"loss": 2.812,
|
| 1450 |
-
"step": 118500
|
| 1451 |
-
},
|
| 1452 |
-
{
|
| 1453 |
-
"epoch": 1.74,
|
| 1454 |
-
"learning_rate": 7.299007612575117e-05,
|
| 1455 |
-
"loss": 2.8172,
|
| 1456 |
-
"step": 119000
|
| 1457 |
-
},
|
| 1458 |
-
{
|
| 1459 |
-
"epoch": 1.75,
|
| 1460 |
-
"learning_rate": 7.278505817422199e-05,
|
| 1461 |
-
"loss": 2.8063,
|
| 1462 |
-
"step": 119500
|
| 1463 |
-
},
|
| 1464 |
-
{
|
| 1465 |
-
"epoch": 1.76,
|
| 1466 |
-
"learning_rate": 7.257955559013181e-05,
|
| 1467 |
-
"loss": 2.8775,
|
| 1468 |
-
"step": 120000
|
| 1469 |
-
},
|
| 1470 |
-
{
|
| 1471 |
-
"epoch": 1.77,
|
| 1472 |
-
"learning_rate": 7.23735727444698e-05,
|
| 1473 |
-
"loss": 2.7225,
|
| 1474 |
-
"step": 120500
|
| 1475 |
-
},
|
| 1476 |
-
{
|
| 1477 |
-
"epoch": 1.77,
|
| 1478 |
-
"learning_rate": 7.216711401844028e-05,
|
| 1479 |
-
"loss": 2.7283,
|
| 1480 |
-
"step": 121000
|
| 1481 |
-
},
|
| 1482 |
-
{
|
| 1483 |
-
"epoch": 1.78,
|
| 1484 |
-
"learning_rate": 7.196018380336934e-05,
|
| 1485 |
-
"loss": 2.8786,
|
| 1486 |
-
"step": 121500
|
| 1487 |
-
},
|
| 1488 |
-
{
|
| 1489 |
-
"epoch": 1.79,
|
| 1490 |
-
"learning_rate": 7.175278650061156e-05,
|
| 1491 |
-
"loss": 2.7372,
|
| 1492 |
-
"step": 122000
|
| 1493 |
-
},
|
| 1494 |
-
{
|
| 1495 |
-
"epoch": 1.8,
|
| 1496 |
-
"learning_rate": 7.154492652145635e-05,
|
| 1497 |
-
"loss": 2.7176,
|
| 1498 |
-
"step": 122500
|
| 1499 |
-
},
|
| 1500 |
-
{
|
| 1501 |
-
"epoch": 1.8,
|
| 1502 |
-
"learning_rate": 7.133660828703414e-05,
|
| 1503 |
-
"loss": 2.8233,
|
| 1504 |
-
"step": 123000
|
| 1505 |
-
},
|
| 1506 |
-
{
|
| 1507 |
-
"epoch": 1.81,
|
| 1508 |
-
"learning_rate": 7.112783622822235e-05,
|
| 1509 |
-
"loss": 2.8269,
|
| 1510 |
-
"step": 123500
|
| 1511 |
-
},
|
| 1512 |
-
{
|
| 1513 |
-
"epoch": 1.82,
|
| 1514 |
-
"learning_rate": 7.091861478555114e-05,
|
| 1515 |
-
"loss": 2.8155,
|
| 1516 |
-
"step": 124000
|
| 1517 |
-
},
|
| 1518 |
-
{
|
| 1519 |
-
"epoch": 1.83,
|
| 1520 |
-
"learning_rate": 7.070894840910887e-05,
|
| 1521 |
-
"loss": 2.7793,
|
| 1522 |
-
"step": 124500
|
| 1523 |
-
},
|
| 1524 |
-
{
|
| 1525 |
-
"epoch": 1.83,
|
| 1526 |
-
"learning_rate": 7.049884155844762e-05,
|
| 1527 |
-
"loss": 2.8524,
|
| 1528 |
-
"step": 125000
|
| 1529 |
-
},
|
| 1530 |
-
{
|
| 1531 |
-
"epoch": 1.84,
|
| 1532 |
-
"learning_rate": 7.028829870248824e-05,
|
| 1533 |
-
"loss": 2.8373,
|
| 1534 |
-
"step": 125500
|
| 1535 |
-
},
|
| 1536 |
-
{
|
| 1537 |
-
"epoch": 1.85,
|
| 1538 |
-
"learning_rate": 7.007732431942529e-05,
|
| 1539 |
-
"loss": 2.8219,
|
| 1540 |
-
"step": 126000
|
| 1541 |
-
},
|
| 1542 |
-
{
|
| 1543 |
-
"epoch": 1.85,
|
| 1544 |
-
"learning_rate": 6.986592289663177e-05,
|
| 1545 |
-
"loss": 2.6958,
|
| 1546 |
-
"step": 126500
|
| 1547 |
-
},
|
| 1548 |
-
{
|
| 1549 |
-
"epoch": 1.86,
|
| 1550 |
-
"learning_rate": 6.965409893056375e-05,
|
| 1551 |
-
"loss": 2.8492,
|
| 1552 |
-
"step": 127000
|
| 1553 |
-
},
|
| 1554 |
-
{
|
| 1555 |
-
"epoch": 1.87,
|
| 1556 |
-
"learning_rate": 6.944185692666472e-05,
|
| 1557 |
-
"loss": 2.8947,
|
| 1558 |
-
"step": 127500
|
| 1559 |
-
},
|
| 1560 |
-
{
|
| 1561 |
-
"epoch": 1.88,
|
| 1562 |
-
"learning_rate": 6.922920139926964e-05,
|
| 1563 |
-
"loss": 2.8311,
|
| 1564 |
-
"step": 128000
|
| 1565 |
-
},
|
| 1566 |
-
{
|
| 1567 |
-
"epoch": 1.88,
|
| 1568 |
-
"learning_rate": 6.90161368715091e-05,
|
| 1569 |
-
"loss": 2.8389,
|
| 1570 |
-
"step": 128500
|
| 1571 |
-
},
|
| 1572 |
-
{
|
| 1573 |
-
"epoch": 1.89,
|
| 1574 |
-
"learning_rate": 6.880266787521298e-05,
|
| 1575 |
-
"loss": 2.7657,
|
| 1576 |
-
"step": 129000
|
| 1577 |
-
},
|
| 1578 |
-
{
|
| 1579 |
-
"epoch": 1.9,
|
| 1580 |
-
"learning_rate": 6.858879895081412e-05,
|
| 1581 |
-
"loss": 2.8388,
|
| 1582 |
-
"step": 129500
|
| 1583 |
-
},
|
| 1584 |
-
{
|
| 1585 |
-
"epoch": 1.91,
|
| 1586 |
-
"learning_rate": 6.837453464725174e-05,
|
| 1587 |
-
"loss": 2.8377,
|
| 1588 |
-
"step": 130000
|
| 1589 |
-
},
|
| 1590 |
-
{
|
| 1591 |
-
"epoch": 1.91,
|
| 1592 |
-
"learning_rate": 6.815987952187466e-05,
|
| 1593 |
-
"loss": 2.8376,
|
| 1594 |
-
"step": 130500
|
| 1595 |
-
},
|
| 1596 |
-
{
|
| 1597 |
-
"epoch": 1.92,
|
| 1598 |
-
"learning_rate": 6.794483814034439e-05,
|
| 1599 |
-
"loss": 2.7048,
|
| 1600 |
-
"step": 131000
|
| 1601 |
-
},
|
| 1602 |
-
{
|
| 1603 |
-
"epoch": 1.93,
|
| 1604 |
-
"learning_rate": 6.772941507653803e-05,
|
| 1605 |
-
"loss": 2.9014,
|
| 1606 |
-
"step": 131500
|
| 1607 |
-
},
|
| 1608 |
-
{
|
| 1609 |
-
"epoch": 1.93,
|
| 1610 |
-
"learning_rate": 6.751361491245093e-05,
|
| 1611 |
-
"loss": 2.7394,
|
| 1612 |
-
"step": 132000
|
| 1613 |
-
},
|
| 1614 |
-
{
|
| 1615 |
-
"epoch": 1.94,
|
| 1616 |
-
"learning_rate": 6.729744223809929e-05,
|
| 1617 |
-
"loss": 2.7723,
|
| 1618 |
-
"step": 132500
|
| 1619 |
-
},
|
| 1620 |
-
{
|
| 1621 |
-
"epoch": 1.95,
|
| 1622 |
-
"learning_rate": 6.708090165142255e-05,
|
| 1623 |
-
"loss": 2.8502,
|
| 1624 |
-
"step": 133000
|
| 1625 |
-
},
|
| 1626 |
-
{
|
| 1627 |
-
"epoch": 1.96,
|
| 1628 |
-
"learning_rate": 6.686399775818548e-05,
|
| 1629 |
-
"loss": 2.7521,
|
| 1630 |
-
"step": 133500
|
| 1631 |
-
},
|
| 1632 |
-
{
|
| 1633 |
-
"epoch": 1.96,
|
| 1634 |
-
"learning_rate": 6.664673517188036e-05,
|
| 1635 |
-
"loss": 2.8726,
|
| 1636 |
-
"step": 134000
|
| 1637 |
-
},
|
| 1638 |
-
{
|
| 1639 |
-
"epoch": 1.97,
|
| 1640 |
-
"learning_rate": 6.642911851362873e-05,
|
| 1641 |
-
"loss": 2.9264,
|
| 1642 |
-
"step": 134500
|
| 1643 |
-
},
|
| 1644 |
-
{
|
| 1645 |
-
"epoch": 1.98,
|
| 1646 |
-
"learning_rate": 6.621115241208316e-05,
|
| 1647 |
-
"loss": 2.8401,
|
| 1648 |
-
"step": 135000
|
| 1649 |
-
},
|
| 1650 |
-
{
|
| 1651 |
-
"epoch": 1.99,
|
| 1652 |
-
"learning_rate": 6.599284150332885e-05,
|
| 1653 |
-
"loss": 2.8706,
|
| 1654 |
-
"step": 135500
|
| 1655 |
-
},
|
| 1656 |
-
{
|
| 1657 |
-
"epoch": 1.99,
|
| 1658 |
-
"learning_rate": 6.57741904307849e-05,
|
| 1659 |
-
"loss": 2.886,
|
| 1660 |
-
"step": 136000
|
| 1661 |
-
},
|
| 1662 |
-
{
|
| 1663 |
-
"epoch": 2.0,
|
| 1664 |
-
"eval_bleu": 1.0,
|
| 1665 |
-
"eval_brevity_penalty": 1.0,
|
| 1666 |
-
"eval_length_ratio": 1.0,
|
| 1667 |
-
"eval_loss": 2.8930435180664062,
|
| 1668 |
-
"eval_precisions": [
|
| 1669 |
-
1.0,
|
| 1670 |
-
1.0,
|
| 1671 |
-
1.0,
|
| 1672 |
-
1.0
|
| 1673 |
-
],
|
| 1674 |
-
"eval_reference_length": 7761920,
|
| 1675 |
-
"eval_runtime": 15438.142,
|
| 1676 |
-
"eval_samples_per_second": 0.982,
|
| 1677 |
-
"eval_steps_per_second": 0.491,
|
| 1678 |
-
"eval_translation_length": 7761920,
|
| 1679 |
-
"step": 136438
|
| 1680 |
-
},
|
| 1681 |
-
{
|
| 1682 |
-
"epoch": 2.0,
|
| 1683 |
-
"learning_rate": 6.555520384510561e-05,
|
| 1684 |
-
"loss": 2.8434,
|
| 1685 |
-
"step": 136500
|
| 1686 |
-
},
|
| 1687 |
-
{
|
| 1688 |
-
"epoch": 2.01,
|
| 1689 |
-
"learning_rate": 6.533588640408162e-05,
|
| 1690 |
-
"loss": 2.4374,
|
| 1691 |
-
"step": 137000
|
| 1692 |
-
},
|
| 1693 |
-
{
|
| 1694 |
-
"epoch": 2.02,
|
| 1695 |
-
"learning_rate": 6.511624277254071e-05,
|
| 1696 |
-
"loss": 2.421,
|
| 1697 |
-
"step": 137500
|
| 1698 |
-
},
|
| 1699 |
-
{
|
| 1700 |
-
"epoch": 2.02,
|
| 1701 |
-
"learning_rate": 6.48962776222487e-05,
|
| 1702 |
-
"loss": 2.4327,
|
| 1703 |
-
"step": 138000
|
| 1704 |
-
},
|
| 1705 |
-
{
|
| 1706 |
-
"epoch": 2.03,
|
| 1707 |
-
"learning_rate": 6.467599563181004e-05,
|
| 1708 |
-
"loss": 2.4567,
|
| 1709 |
-
"step": 138500
|
| 1710 |
-
},
|
| 1711 |
-
{
|
| 1712 |
-
"epoch": 2.04,
|
| 1713 |
-
"learning_rate": 6.445540148656825e-05,
|
| 1714 |
-
"loss": 2.3828,
|
| 1715 |
-
"step": 139000
|
| 1716 |
-
},
|
| 1717 |
-
{
|
| 1718 |
-
"epoch": 2.04,
|
| 1719 |
-
"learning_rate": 6.423449987850634e-05,
|
| 1720 |
-
"loss": 2.4582,
|
| 1721 |
-
"step": 139500
|
| 1722 |
-
},
|
| 1723 |
-
{
|
| 1724 |
-
"epoch": 2.05,
|
| 1725 |
-
"learning_rate": 6.401329550614694e-05,
|
| 1726 |
-
"loss": 2.4376,
|
| 1727 |
-
"step": 140000
|
| 1728 |
-
},
|
| 1729 |
-
{
|
| 1730 |
-
"epoch": 2.06,
|
| 1731 |
-
"learning_rate": 6.379179307445245e-05,
|
| 1732 |
-
"loss": 2.4871,
|
| 1733 |
-
"step": 140500
|
| 1734 |
-
},
|
| 1735 |
-
{
|
| 1736 |
-
"epoch": 2.07,
|
| 1737 |
-
"learning_rate": 6.35699972947249e-05,
|
| 1738 |
-
"loss": 2.4955,
|
| 1739 |
-
"step": 141000
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 2.07,
|
| 1743 |
-
"learning_rate": 6.334791288450577e-05,
|
| 1744 |
-
"loss": 2.5095,
|
| 1745 |
-
"step": 141500
|
| 1746 |
-
},
|
| 1747 |
-
{
|
| 1748 |
-
"epoch": 2.08,
|
| 1749 |
-
"learning_rate": 6.312554456747558e-05,
|
| 1750 |
-
"loss": 2.4246,
|
| 1751 |
-
"step": 142000
|
| 1752 |
-
},
|
| 1753 |
-
{
|
| 1754 |
-
"epoch": 2.09,
|
| 1755 |
-
"learning_rate": 6.290289707335355e-05,
|
| 1756 |
-
"loss": 2.567,
|
| 1757 |
-
"step": 142500
|
| 1758 |
-
},
|
| 1759 |
-
{
|
| 1760 |
-
"epoch": 2.1,
|
| 1761 |
-
"learning_rate": 6.26799751377969e-05,
|
| 1762 |
-
"loss": 2.4974,
|
| 1763 |
-
"step": 143000
|
| 1764 |
-
},
|
| 1765 |
-
{
|
| 1766 |
-
"epoch": 2.1,
|
| 1767 |
-
"learning_rate": 6.245678350230015e-05,
|
| 1768 |
-
"loss": 2.4016,
|
| 1769 |
-
"step": 143500
|
| 1770 |
-
},
|
| 1771 |
-
{
|
| 1772 |
-
"epoch": 2.11,
|
| 1773 |
-
"learning_rate": 6.223332691409428e-05,
|
| 1774 |
-
"loss": 2.4861,
|
| 1775 |
-
"step": 144000
|
| 1776 |
-
},
|
| 1777 |
-
{
|
| 1778 |
-
"epoch": 2.12,
|
| 1779 |
-
"learning_rate": 6.200961012604575e-05,
|
| 1780 |
-
"loss": 2.4795,
|
| 1781 |
-
"step": 144500
|
| 1782 |
-
},
|
| 1783 |
-
{
|
| 1784 |
-
"epoch": 2.13,
|
| 1785 |
-
"learning_rate": 6.178563789655537e-05,
|
| 1786 |
-
"loss": 2.5037,
|
| 1787 |
-
"step": 145000
|
| 1788 |
-
},
|
| 1789 |
-
{
|
| 1790 |
-
"epoch": 2.13,
|
| 1791 |
-
"learning_rate": 6.156141498945719e-05,
|
| 1792 |
-
"loss": 2.4686,
|
| 1793 |
-
"step": 145500
|
| 1794 |
-
},
|
| 1795 |
-
{
|
| 1796 |
-
"epoch": 2.14,
|
| 1797 |
-
"learning_rate": 6.133694617391704e-05,
|
| 1798 |
-
"loss": 2.5575,
|
| 1799 |
-
"step": 146000
|
| 1800 |
-
},
|
| 1801 |
-
{
|
| 1802 |
-
"epoch": 2.15,
|
| 1803 |
-
"learning_rate": 6.111223622433124e-05,
|
| 1804 |
-
"loss": 2.5201,
|
| 1805 |
-
"step": 146500
|
| 1806 |
-
},
|
| 1807 |
-
{
|
| 1808 |
-
"epoch": 2.15,
|
| 1809 |
-
"learning_rate": 6.088728992022491e-05,
|
| 1810 |
-
"loss": 2.4335,
|
| 1811 |
-
"step": 147000
|
| 1812 |
-
},
|
| 1813 |
-
{
|
| 1814 |
-
"epoch": 2.16,
|
| 1815 |
-
"learning_rate": 6.0662112046150376e-05,
|
| 1816 |
-
"loss": 2.4387,
|
| 1817 |
-
"step": 147500
|
| 1818 |
-
},
|
| 1819 |
-
{
|
| 1820 |
-
"epoch": 2.17,
|
| 1821 |
-
"learning_rate": 6.043670739158542e-05,
|
| 1822 |
-
"loss": 2.5027,
|
| 1823 |
-
"step": 148000
|
| 1824 |
-
},
|
| 1825 |
-
{
|
| 1826 |
-
"epoch": 2.18,
|
| 1827 |
-
"learning_rate": 6.021108075083142e-05,
|
| 1828 |
-
"loss": 2.4055,
|
| 1829 |
-
"step": 148500
|
| 1830 |
-
},
|
| 1831 |
-
{
|
| 1832 |
-
"epoch": 2.18,
|
| 1833 |
-
"learning_rate": 5.998523692291127e-05,
|
| 1834 |
-
"loss": 2.478,
|
| 1835 |
-
"step": 149000
|
| 1836 |
-
},
|
| 1837 |
-
{
|
| 1838 |
-
"epoch": 2.19,
|
| 1839 |
-
"learning_rate": 5.9759180711467446e-05,
|
| 1840 |
-
"loss": 2.4491,
|
| 1841 |
-
"step": 149500
|
| 1842 |
-
},
|
| 1843 |
-
{
|
| 1844 |
-
"epoch": 2.2,
|
| 1845 |
-
"learning_rate": 5.953291692465974e-05,
|
| 1846 |
-
"loss": 2.5007,
|
| 1847 |
-
"step": 150000
|
| 1848 |
-
},
|
| 1849 |
-
{
|
| 1850 |
-
"epoch": 2.21,
|
| 1851 |
-
"learning_rate": 5.930645037506301e-05,
|
| 1852 |
-
"loss": 2.4752,
|
| 1853 |
-
"step": 150500
|
| 1854 |
-
},
|
| 1855 |
-
{
|
| 1856 |
-
"epoch": 2.21,
|
| 1857 |
-
"learning_rate": 5.9079785879564876e-05,
|
| 1858 |
-
"loss": 2.4569,
|
| 1859 |
-
"step": 151000
|
| 1860 |
-
},
|
| 1861 |
-
{
|
| 1862 |
-
"epoch": 2.22,
|
| 1863 |
-
"learning_rate": 5.885292825926314e-05,
|
| 1864 |
-
"loss": 2.441,
|
| 1865 |
-
"step": 151500
|
| 1866 |
-
},
|
| 1867 |
-
{
|
| 1868 |
-
"epoch": 2.23,
|
| 1869 |
-
"learning_rate": 5.862588233936341e-05,
|
| 1870 |
-
"loss": 2.4172,
|
| 1871 |
-
"step": 152000
|
| 1872 |
-
},
|
| 1873 |
-
{
|
| 1874 |
-
"epoch": 2.24,
|
| 1875 |
-
"learning_rate": 5.8398652949076324e-05,
|
| 1876 |
-
"loss": 2.4053,
|
| 1877 |
-
"step": 152500
|
| 1878 |
-
},
|
| 1879 |
-
{
|
| 1880 |
-
"epoch": 2.24,
|
| 1881 |
-
"learning_rate": 5.81712449215149e-05,
|
| 1882 |
-
"loss": 2.5097,
|
| 1883 |
-
"step": 153000
|
| 1884 |
-
},
|
| 1885 |
-
{
|
| 1886 |
-
"epoch": 2.25,
|
| 1887 |
-
"learning_rate": 5.794366309359174e-05,
|
| 1888 |
-
"loss": 2.4697,
|
| 1889 |
-
"step": 153500
|
| 1890 |
-
},
|
| 1891 |
-
{
|
| 1892 |
-
"epoch": 2.26,
|
| 1893 |
-
"learning_rate": 5.771591230591612e-05,
|
| 1894 |
-
"loss": 2.5604,
|
| 1895 |
-
"step": 154000
|
| 1896 |
-
},
|
| 1897 |
-
{
|
| 1898 |
-
"epoch": 2.26,
|
| 1899 |
-
"learning_rate": 5.748799740269104e-05,
|
| 1900 |
-
"loss": 2.3954,
|
| 1901 |
-
"step": 154500
|
| 1902 |
-
},
|
| 1903 |
-
{
|
| 1904 |
-
"epoch": 2.27,
|
| 1905 |
-
"learning_rate": 5.7259923231610236e-05,
|
| 1906 |
-
"loss": 2.4581,
|
| 1907 |
-
"step": 155000
|
| 1908 |
-
},
|
| 1909 |
-
{
|
| 1910 |
-
"epoch": 2.28,
|
| 1911 |
-
"learning_rate": 5.703169464375498e-05,
|
| 1912 |
-
"loss": 2.5286,
|
| 1913 |
-
"step": 155500
|
| 1914 |
-
},
|
| 1915 |
-
{
|
| 1916 |
-
"epoch": 2.29,
|
| 1917 |
-
"learning_rate": 5.680331649349101e-05,
|
| 1918 |
-
"loss": 2.3529,
|
| 1919 |
-
"step": 156000
|
| 1920 |
-
},
|
| 1921 |
-
{
|
| 1922 |
-
"epoch": 2.29,
|
| 1923 |
-
"learning_rate": 5.657479363836519e-05,
|
| 1924 |
-
"loss": 2.5152,
|
| 1925 |
-
"step": 156500
|
| 1926 |
-
},
|
| 1927 |
-
{
|
| 1928 |
-
"epoch": 2.3,
|
| 1929 |
-
"learning_rate": 5.6346130939002193e-05,
|
| 1930 |
-
"loss": 2.512,
|
| 1931 |
-
"step": 157000
|
| 1932 |
-
},
|
| 1933 |
-
{
|
| 1934 |
-
"epoch": 2.31,
|
| 1935 |
-
"learning_rate": 5.6117333259001183e-05,
|
| 1936 |
-
"loss": 2.5408,
|
| 1937 |
-
"step": 157500
|
| 1938 |
-
},
|
| 1939 |
-
{
|
| 1940 |
-
"epoch": 2.32,
|
| 1941 |
-
"learning_rate": 5.5888405464832314e-05,
|
| 1942 |
-
"loss": 2.5028,
|
| 1943 |
-
"step": 158000
|
| 1944 |
-
},
|
| 1945 |
-
{
|
| 1946 |
-
"epoch": 2.32,
|
| 1947 |
-
"learning_rate": 5.565935242573323e-05,
|
| 1948 |
-
"loss": 2.5252,
|
| 1949 |
-
"step": 158500
|
| 1950 |
-
},
|
| 1951 |
-
{
|
| 1952 |
-
"epoch": 2.33,
|
| 1953 |
-
"learning_rate": 5.5430179013605544e-05,
|
| 1954 |
-
"loss": 2.4696,
|
| 1955 |
-
"step": 159000
|
| 1956 |
-
},
|
| 1957 |
-
{
|
| 1958 |
-
"epoch": 2.34,
|
| 1959 |
-
"learning_rate": 5.5200890102911096e-05,
|
| 1960 |
-
"loss": 2.4254,
|
| 1961 |
-
"step": 159500
|
| 1962 |
-
},
|
| 1963 |
-
{
|
| 1964 |
-
"epoch": 2.35,
|
| 1965 |
-
"learning_rate": 5.497149057056843e-05,
|
| 1966 |
-
"loss": 2.346,
|
| 1967 |
-
"step": 160000
|
| 1968 |
-
},
|
| 1969 |
-
{
|
| 1970 |
-
"epoch": 2.35,
|
| 1971 |
-
"learning_rate": 5.474198529584896e-05,
|
| 1972 |
-
"loss": 2.4914,
|
| 1973 |
-
"step": 160500
|
| 1974 |
-
},
|
| 1975 |
-
{
|
| 1976 |
-
"epoch": 2.36,
|
| 1977 |
-
"learning_rate": 5.451237916027319e-05,
|
| 1978 |
-
"loss": 2.5014,
|
| 1979 |
-
"step": 161000
|
| 1980 |
-
},
|
| 1981 |
-
{
|
| 1982 |
-
"epoch": 2.37,
|
| 1983 |
-
"learning_rate": 5.428267704750691e-05,
|
| 1984 |
-
"loss": 2.4718,
|
| 1985 |
-
"step": 161500
|
| 1986 |
-
},
|
| 1987 |
-
{
|
| 1988 |
-
"epoch": 2.37,
|
| 1989 |
-
"learning_rate": 5.4052883843257375e-05,
|
| 1990 |
-
"loss": 2.4263,
|
| 1991 |
-
"step": 162000
|
| 1992 |
-
},
|
| 1993 |
-
{
|
| 1994 |
-
"epoch": 2.38,
|
| 1995 |
-
"learning_rate": 5.3823004435169276e-05,
|
| 1996 |
-
"loss": 2.5898,
|
| 1997 |
-
"step": 162500
|
| 1998 |
-
},
|
| 1999 |
-
{
|
| 2000 |
-
"epoch": 2.39,
|
| 2001 |
-
"learning_rate": 5.3593043712720835e-05,
|
| 2002 |
-
"loss": 2.5174,
|
| 2003 |
-
"step": 163000
|
| 2004 |
-
},
|
| 2005 |
-
{
|
| 2006 |
-
"epoch": 2.4,
|
| 2007 |
-
"learning_rate": 5.336300656711989e-05,
|
| 2008 |
-
"loss": 2.5126,
|
| 2009 |
-
"step": 163500
|
| 2010 |
-
},
|
| 2011 |
-
{
|
| 2012 |
-
"epoch": 2.4,
|
| 2013 |
-
"learning_rate": 5.3132897891199684e-05,
|
| 2014 |
-
"loss": 2.4919,
|
| 2015 |
-
"step": 164000
|
| 2016 |
-
},
|
| 2017 |
-
{
|
| 2018 |
-
"epoch": 2.41,
|
| 2019 |
-
"learning_rate": 5.290272257931495e-05,
|
| 2020 |
-
"loss": 2.4545,
|
| 2021 |
-
"step": 164500
|
| 2022 |
-
},
|
| 2023 |
-
{
|
| 2024 |
-
"epoch": 2.42,
|
| 2025 |
-
"learning_rate": 5.267248552723771e-05,
|
| 2026 |
-
"loss": 2.5104,
|
| 2027 |
-
"step": 165000
|
| 2028 |
-
},
|
| 2029 |
-
{
|
| 2030 |
-
"epoch": 2.43,
|
| 2031 |
-
"learning_rate": 5.244219163205324e-05,
|
| 2032 |
-
"loss": 2.464,
|
| 2033 |
-
"step": 165500
|
| 2034 |
-
},
|
| 2035 |
-
{
|
| 2036 |
-
"epoch": 2.43,
|
| 2037 |
-
"learning_rate": 5.22118457920558e-05,
|
| 2038 |
-
"loss": 2.4752,
|
| 2039 |
-
"step": 166000
|
| 2040 |
-
},
|
| 2041 |
-
{
|
| 2042 |
-
"epoch": 2.44,
|
| 2043 |
-
"learning_rate": 5.198145290664456e-05,
|
| 2044 |
-
"loss": 2.5599,
|
| 2045 |
-
"step": 166500
|
| 2046 |
-
},
|
| 2047 |
-
{
|
| 2048 |
-
"epoch": 2.45,
|
| 2049 |
-
"learning_rate": 5.1751017876219256e-05,
|
| 2050 |
-
"loss": 2.4875,
|
| 2051 |
-
"step": 167000
|
| 2052 |
-
},
|
| 2053 |
-
{
|
| 2054 |
-
"epoch": 2.46,
|
| 2055 |
-
"learning_rate": 5.15205456020761e-05,
|
| 2056 |
-
"loss": 2.4102,
|
| 2057 |
-
"step": 167500
|
| 2058 |
-
},
|
| 2059 |
-
{
|
| 2060 |
-
"epoch": 2.46,
|
| 2061 |
-
"learning_rate": 5.129004098630348e-05,
|
| 2062 |
-
"loss": 2.4686,
|
| 2063 |
-
"step": 168000
|
| 2064 |
-
},
|
| 2065 |
-
{
|
| 2066 |
-
"epoch": 2.47,
|
| 2067 |
-
"learning_rate": 5.105950893167761e-05,
|
| 2068 |
-
"loss": 2.5395,
|
| 2069 |
-
"step": 168500
|
| 2070 |
-
},
|
| 2071 |
-
{
|
| 2072 |
-
"epoch": 2.48,
|
| 2073 |
-
"learning_rate": 5.082895434155841e-05,
|
| 2074 |
-
"loss": 2.4642,
|
| 2075 |
-
"step": 169000
|
| 2076 |
-
},
|
| 2077 |
-
{
|
| 2078 |
-
"epoch": 2.48,
|
| 2079 |
-
"learning_rate": 5.059838211978506e-05,
|
| 2080 |
-
"loss": 2.6105,
|
| 2081 |
-
"step": 169500
|
| 2082 |
-
},
|
| 2083 |
-
{
|
| 2084 |
-
"epoch": 2.49,
|
| 2085 |
-
"learning_rate": 5.0367797170571776e-05,
|
| 2086 |
-
"loss": 2.5126,
|
| 2087 |
-
"step": 170000
|
| 2088 |
-
},
|
| 2089 |
-
{
|
| 2090 |
-
"epoch": 2.5,
|
| 2091 |
-
"learning_rate": 5.013720439840348e-05,
|
| 2092 |
-
"loss": 2.5685,
|
| 2093 |
-
"step": 170500
|
| 2094 |
-
},
|
| 2095 |
-
{
|
| 2096 |
-
"epoch": 2.51,
|
| 2097 |
-
"learning_rate": 4.9906608707931526e-05,
|
| 2098 |
-
"loss": 2.4634,
|
| 2099 |
-
"step": 171000
|
| 2100 |
-
},
|
| 2101 |
-
{
|
| 2102 |
-
"epoch": 2.51,
|
| 2103 |
-
"learning_rate": 4.967601500386929e-05,
|
| 2104 |
-
"loss": 2.4449,
|
| 2105 |
-
"step": 171500
|
| 2106 |
-
},
|
| 2107 |
-
{
|
| 2108 |
-
"epoch": 2.52,
|
| 2109 |
-
"learning_rate": 4.9445428190887924e-05,
|
| 2110 |
-
"loss": 2.515,
|
| 2111 |
-
"step": 172000
|
| 2112 |
-
},
|
| 2113 |
-
{
|
| 2114 |
-
"epoch": 2.53,
|
| 2115 |
-
"learning_rate": 4.9214853173511975e-05,
|
| 2116 |
-
"loss": 2.5057,
|
| 2117 |
-
"step": 172500
|
| 2118 |
-
},
|
| 2119 |
-
{
|
| 2120 |
-
"epoch": 2.54,
|
| 2121 |
-
"learning_rate": 4.8984294856015164e-05,
|
| 2122 |
-
"loss": 2.5329,
|
| 2123 |
-
"step": 173000
|
| 2124 |
-
},
|
| 2125 |
-
{
|
| 2126 |
-
"epoch": 2.54,
|
| 2127 |
-
"learning_rate": 4.875375814231596e-05,
|
| 2128 |
-
"loss": 2.4823,
|
| 2129 |
-
"step": 173500
|
| 2130 |
-
},
|
| 2131 |
-
{
|
| 2132 |
-
"epoch": 2.55,
|
| 2133 |
-
"learning_rate": 4.852324793587333e-05,
|
| 2134 |
-
"loss": 2.4907,
|
| 2135 |
-
"step": 174000
|
| 2136 |
-
},
|
| 2137 |
-
{
|
| 2138 |
-
"epoch": 2.56,
|
| 2139 |
-
"learning_rate": 4.829276913958245e-05,
|
| 2140 |
-
"loss": 2.523,
|
| 2141 |
-
"step": 174500
|
| 2142 |
-
},
|
| 2143 |
-
{
|
| 2144 |
-
"epoch": 2.57,
|
| 2145 |
-
"learning_rate": 4.80623266556704e-05,
|
| 2146 |
-
"loss": 2.3798,
|
| 2147 |
-
"step": 175000
|
| 2148 |
-
},
|
| 2149 |
-
{
|
| 2150 |
-
"epoch": 2.57,
|
| 2151 |
-
"learning_rate": 4.783192538559191e-05,
|
| 2152 |
-
"loss": 2.5055,
|
| 2153 |
-
"step": 175500
|
| 2154 |
-
},
|
| 2155 |
-
{
|
| 2156 |
-
"epoch": 2.58,
|
| 2157 |
-
"learning_rate": 4.760157022992511e-05,
|
| 2158 |
-
"loss": 2.542,
|
| 2159 |
-
"step": 176000
|
| 2160 |
-
},
|
| 2161 |
-
{
|
| 2162 |
-
"epoch": 2.59,
|
| 2163 |
-
"learning_rate": 4.737126608826727e-05,
|
| 2164 |
-
"loss": 2.5768,
|
| 2165 |
-
"step": 176500
|
| 2166 |
-
},
|
| 2167 |
-
{
|
| 2168 |
-
"epoch": 2.59,
|
| 2169 |
-
"learning_rate": 4.7141017859130605e-05,
|
| 2170 |
-
"loss": 2.5816,
|
| 2171 |
-
"step": 177000
|
| 2172 |
-
},
|
| 2173 |
-
{
|
| 2174 |
-
"epoch": 2.6,
|
| 2175 |
-
"learning_rate": 4.6910830439838114e-05,
|
| 2176 |
-
"loss": 2.4947,
|
| 2177 |
-
"step": 177500
|
| 2178 |
-
},
|
| 2179 |
-
{
|
| 2180 |
-
"epoch": 2.61,
|
| 2181 |
-
"learning_rate": 4.668070872641934e-05,
|
| 2182 |
-
"loss": 2.5179,
|
| 2183 |
-
"step": 178000
|
| 2184 |
-
},
|
| 2185 |
-
{
|
| 2186 |
-
"epoch": 2.62,
|
| 2187 |
-
"learning_rate": 4.645065761350634e-05,
|
| 2188 |
-
"loss": 2.5035,
|
| 2189 |
-
"step": 178500
|
| 2190 |
-
},
|
| 2191 |
-
{
|
| 2192 |
-
"epoch": 2.62,
|
| 2193 |
-
"learning_rate": 4.622068199422945e-05,
|
| 2194 |
-
"loss": 2.4886,
|
| 2195 |
-
"step": 179000
|
| 2196 |
-
},
|
| 2197 |
-
{
|
| 2198 |
-
"epoch": 2.63,
|
| 2199 |
-
"learning_rate": 4.59907867601133e-05,
|
| 2200 |
-
"loss": 2.434,
|
| 2201 |
-
"step": 179500
|
| 2202 |
-
},
|
| 2203 |
-
{
|
| 2204 |
-
"epoch": 2.64,
|
| 2205 |
-
"learning_rate": 4.576097680097277e-05,
|
| 2206 |
-
"loss": 2.4869,
|
| 2207 |
-
"step": 180000
|
| 2208 |
-
},
|
| 2209 |
-
{
|
| 2210 |
-
"epoch": 2.65,
|
| 2211 |
-
"learning_rate": 4.553125700480892e-05,
|
| 2212 |
-
"loss": 2.4565,
|
| 2213 |
-
"step": 180500
|
| 2214 |
-
},
|
| 2215 |
-
{
|
| 2216 |
-
"epoch": 2.65,
|
| 2217 |
-
"learning_rate": 4.530163225770513e-05,
|
| 2218 |
-
"loss": 2.496,
|
| 2219 |
-
"step": 181000
|
| 2220 |
-
},
|
| 2221 |
-
{
|
| 2222 |
-
"epoch": 2.66,
|
| 2223 |
-
"learning_rate": 4.5072107443723025e-05,
|
| 2224 |
-
"loss": 2.415,
|
| 2225 |
-
"step": 181500
|
| 2226 |
-
},
|
| 2227 |
-
{
|
| 2228 |
-
"epoch": 2.67,
|
| 2229 |
-
"learning_rate": 4.484268744479875e-05,
|
| 2230 |
-
"loss": 2.4461,
|
| 2231 |
-
"step": 182000
|
| 2232 |
-
},
|
| 2233 |
-
{
|
| 2234 |
-
"epoch": 2.68,
|
| 2235 |
-
"learning_rate": 4.461337714063901e-05,
|
| 2236 |
-
"loss": 2.4858,
|
| 2237 |
-
"step": 182500
|
| 2238 |
-
},
|
| 2239 |
-
{
|
| 2240 |
-
"epoch": 2.68,
|
| 2241 |
-
"learning_rate": 4.438418140861738e-05,
|
| 2242 |
-
"loss": 2.5667,
|
| 2243 |
-
"step": 183000
|
| 2244 |
-
},
|
| 2245 |
-
{
|
| 2246 |
-
"epoch": 2.69,
|
| 2247 |
-
"learning_rate": 4.415510512367047e-05,
|
| 2248 |
-
"loss": 2.5009,
|
| 2249 |
-
"step": 183500
|
| 2250 |
-
},
|
| 2251 |
-
{
|
| 2252 |
-
"epoch": 2.7,
|
| 2253 |
-
"learning_rate": 4.392615315819427e-05,
|
| 2254 |
-
"loss": 2.575,
|
| 2255 |
-
"step": 184000
|
| 2256 |
-
},
|
| 2257 |
-
{
|
| 2258 |
-
"epoch": 2.7,
|
| 2259 |
-
"learning_rate": 4.3697330381940595e-05,
|
| 2260 |
-
"loss": 2.3703,
|
| 2261 |
-
"step": 184500
|
| 2262 |
-
},
|
| 2263 |
-
{
|
| 2264 |
-
"epoch": 2.71,
|
| 2265 |
-
"learning_rate": 4.346864166191336e-05,
|
| 2266 |
-
"loss": 2.4276,
|
| 2267 |
-
"step": 185000
|
| 2268 |
-
},
|
| 2269 |
-
{
|
| 2270 |
-
"epoch": 2.72,
|
| 2271 |
-
"learning_rate": 4.3240091862265176e-05,
|
| 2272 |
-
"loss": 2.5392,
|
| 2273 |
-
"step": 185500
|
| 2274 |
-
},
|
| 2275 |
-
{
|
| 2276 |
-
"epoch": 2.73,
|
| 2277 |
-
"learning_rate": 4.301168584419385e-05,
|
| 2278 |
-
"loss": 2.4175,
|
| 2279 |
-
"step": 186000
|
| 2280 |
-
},
|
| 2281 |
-
{
|
| 2282 |
-
"epoch": 2.73,
|
| 2283 |
-
"learning_rate": 4.2783428465838974e-05,
|
| 2284 |
-
"loss": 2.457,
|
| 2285 |
-
"step": 186500
|
| 2286 |
-
},
|
| 2287 |
-
{
|
| 2288 |
-
"epoch": 2.74,
|
| 2289 |
-
"learning_rate": 4.2555324582178655e-05,
|
| 2290 |
-
"loss": 2.5549,
|
| 2291 |
-
"step": 187000
|
| 2292 |
-
},
|
| 2293 |
-
{
|
| 2294 |
-
"epoch": 2.75,
|
| 2295 |
-
"learning_rate": 4.232737904492613e-05,
|
| 2296 |
-
"loss": 2.5629,
|
| 2297 |
-
"step": 187500
|
| 2298 |
-
},
|
| 2299 |
-
{
|
| 2300 |
-
"epoch": 2.76,
|
| 2301 |
-
"learning_rate": 4.2099596702426705e-05,
|
| 2302 |
-
"loss": 2.529,
|
| 2303 |
-
"step": 188000
|
| 2304 |
-
},
|
| 2305 |
-
{
|
| 2306 |
-
"epoch": 2.76,
|
| 2307 |
-
"learning_rate": 4.187198239955456e-05,
|
| 2308 |
-
"loss": 2.5347,
|
| 2309 |
-
"step": 188500
|
| 2310 |
-
},
|
| 2311 |
-
{
|
| 2312 |
-
"epoch": 2.77,
|
| 2313 |
-
"learning_rate": 4.164454097760973e-05,
|
| 2314 |
-
"loss": 2.4773,
|
| 2315 |
-
"step": 189000
|
| 2316 |
-
},
|
| 2317 |
-
{
|
| 2318 |
-
"epoch": 2.78,
|
| 2319 |
-
"learning_rate": 4.141727727421508e-05,
|
| 2320 |
-
"loss": 2.4718,
|
| 2321 |
-
"step": 189500
|
| 2322 |
-
},
|
| 2323 |
-
{
|
| 2324 |
-
"epoch": 2.79,
|
| 2325 |
-
"learning_rate": 4.119019612321346e-05,
|
| 2326 |
-
"loss": 2.4831,
|
| 2327 |
-
"step": 190000
|
| 2328 |
-
},
|
| 2329 |
-
{
|
| 2330 |
-
"epoch": 2.79,
|
| 2331 |
-
"learning_rate": 4.096330235456489e-05,
|
| 2332 |
-
"loss": 2.5369,
|
| 2333 |
-
"step": 190500
|
| 2334 |
-
},
|
| 2335 |
-
{
|
| 2336 |
-
"epoch": 2.8,
|
| 2337 |
-
"learning_rate": 4.073660079424377e-05,
|
| 2338 |
-
"loss": 2.5185,
|
| 2339 |
-
"step": 191000
|
| 2340 |
-
},
|
| 2341 |
-
{
|
| 2342 |
-
"epoch": 2.81,
|
| 2343 |
-
"learning_rate": 4.051009626413636e-05,
|
| 2344 |
-
"loss": 2.4736,
|
| 2345 |
-
"step": 191500
|
| 2346 |
-
},
|
| 2347 |
-
{
|
| 2348 |
-
"epoch": 2.81,
|
| 2349 |
-
"learning_rate": 4.028379358193805e-05,
|
| 2350 |
-
"loss": 2.5441,
|
| 2351 |
-
"step": 192000
|
| 2352 |
-
},
|
| 2353 |
-
{
|
| 2354 |
-
"epoch": 2.82,
|
| 2355 |
-
"learning_rate": 4.005769756105101e-05,
|
| 2356 |
-
"loss": 2.4609,
|
| 2357 |
-
"step": 192500
|
| 2358 |
-
},
|
| 2359 |
-
{
|
| 2360 |
-
"epoch": 2.83,
|
| 2361 |
-
"learning_rate": 3.983181301048179e-05,
|
| 2362 |
-
"loss": 2.5473,
|
| 2363 |
-
"step": 193000
|
| 2364 |
-
},
|
| 2365 |
-
{
|
| 2366 |
-
"epoch": 2.84,
|
| 2367 |
-
"learning_rate": 3.960614473473898e-05,
|
| 2368 |
-
"loss": 2.5741,
|
| 2369 |
-
"step": 193500
|
| 2370 |
-
},
|
| 2371 |
-
{
|
| 2372 |
-
"epoch": 2.84,
|
| 2373 |
-
"learning_rate": 3.93806975337311e-05,
|
| 2374 |
-
"loss": 2.5476,
|
| 2375 |
-
"step": 194000
|
| 2376 |
-
},
|
| 2377 |
-
{
|
| 2378 |
-
"epoch": 2.85,
|
| 2379 |
-
"learning_rate": 3.9155476202664446e-05,
|
| 2380 |
-
"loss": 2.5093,
|
| 2381 |
-
"step": 194500
|
| 2382 |
-
},
|
| 2383 |
-
{
|
| 2384 |
-
"epoch": 2.86,
|
| 2385 |
-
"learning_rate": 3.8930485531941084e-05,
|
| 2386 |
-
"loss": 2.5792,
|
| 2387 |
-
"step": 195000
|
| 2388 |
-
},
|
| 2389 |
-
{
|
| 2390 |
-
"epoch": 2.87,
|
| 2391 |
-
"learning_rate": 3.870573030705705e-05,
|
| 2392 |
-
"loss": 2.5136,
|
| 2393 |
-
"step": 195500
|
| 2394 |
-
},
|
| 2395 |
-
{
|
| 2396 |
-
"epoch": 2.87,
|
| 2397 |
-
"learning_rate": 3.848121530850049e-05,
|
| 2398 |
-
"loss": 2.4505,
|
| 2399 |
-
"step": 196000
|
| 2400 |
-
},
|
| 2401 |
-
{
|
| 2402 |
-
"epoch": 2.88,
|
| 2403 |
-
"learning_rate": 3.825694531164997e-05,
|
| 2404 |
-
"loss": 2.4591,
|
| 2405 |
-
"step": 196500
|
| 2406 |
-
},
|
| 2407 |
-
{
|
| 2408 |
-
"epoch": 2.89,
|
| 2409 |
-
"learning_rate": 3.803292508667294e-05,
|
| 2410 |
-
"loss": 2.501,
|
| 2411 |
-
"step": 197000
|
| 2412 |
-
},
|
| 2413 |
-
{
|
| 2414 |
-
"epoch": 2.9,
|
| 2415 |
-
"learning_rate": 3.780915939842428e-05,
|
| 2416 |
-
"loss": 2.5764,
|
| 2417 |
-
"step": 197500
|
| 2418 |
-
},
|
| 2419 |
-
{
|
| 2420 |
-
"epoch": 2.9,
|
| 2421 |
-
"learning_rate": 3.7585653006344886e-05,
|
| 2422 |
-
"loss": 2.4637,
|
| 2423 |
-
"step": 198000
|
| 2424 |
-
},
|
| 2425 |
-
{
|
| 2426 |
-
"epoch": 2.91,
|
| 2427 |
-
"learning_rate": 3.736241066436059e-05,
|
| 2428 |
-
"loss": 2.5644,
|
| 2429 |
-
"step": 198500
|
| 2430 |
-
},
|
| 2431 |
-
{
|
| 2432 |
-
"epoch": 2.92,
|
| 2433 |
-
"learning_rate": 3.713943712078083e-05,
|
| 2434 |
-
"loss": 2.4658,
|
| 2435 |
-
"step": 199000
|
| 2436 |
-
},
|
| 2437 |
-
{
|
| 2438 |
-
"epoch": 2.92,
|
| 2439 |
-
"learning_rate": 3.691673711819783e-05,
|
| 2440 |
-
"loss": 2.5032,
|
| 2441 |
-
"step": 199500
|
| 2442 |
-
},
|
| 2443 |
-
{
|
| 2444 |
-
"epoch": 2.93,
|
| 2445 |
-
"learning_rate": 3.669431539338567e-05,
|
| 2446 |
-
"loss": 2.5171,
|
| 2447 |
-
"step": 200000
|
| 2448 |
-
},
|
| 2449 |
-
{
|
| 2450 |
-
"epoch": 2.94,
|
| 2451 |
-
"learning_rate": 3.647217667719949e-05,
|
| 2452 |
-
"loss": 2.5642,
|
| 2453 |
-
"step": 200500
|
| 2454 |
-
},
|
| 2455 |
-
{
|
| 2456 |
-
"epoch": 2.95,
|
| 2457 |
-
"learning_rate": 3.6250325694474943e-05,
|
| 2458 |
-
"loss": 2.5403,
|
| 2459 |
-
"step": 201000
|
| 2460 |
-
},
|
| 2461 |
-
{
|
| 2462 |
-
"epoch": 2.95,
|
| 2463 |
-
"learning_rate": 3.6028767163927654e-05,
|
| 2464 |
-
"loss": 2.5735,
|
| 2465 |
-
"step": 201500
|
| 2466 |
-
},
|
| 2467 |
-
{
|
| 2468 |
-
"epoch": 2.96,
|
| 2469 |
-
"learning_rate": 3.580750579805285e-05,
|
| 2470 |
-
"loss": 2.4973,
|
| 2471 |
-
"step": 202000
|
| 2472 |
-
},
|
| 2473 |
-
{
|
| 2474 |
-
"epoch": 2.97,
|
| 2475 |
-
"learning_rate": 3.558654630302516e-05,
|
| 2476 |
-
"loss": 2.4614,
|
| 2477 |
-
"step": 202500
|
| 2478 |
-
},
|
| 2479 |
-
{
|
| 2480 |
-
"epoch": 2.98,
|
| 2481 |
-
"learning_rate": 3.536589337859844e-05,
|
| 2482 |
-
"loss": 2.513,
|
| 2483 |
-
"step": 203000
|
| 2484 |
-
},
|
| 2485 |
-
{
|
| 2486 |
-
"epoch": 2.98,
|
| 2487 |
-
"learning_rate": 3.514555171800597e-05,
|
| 2488 |
-
"loss": 2.4906,
|
| 2489 |
-
"step": 203500
|
| 2490 |
-
},
|
| 2491 |
-
{
|
| 2492 |
-
"epoch": 2.99,
|
| 2493 |
-
"learning_rate": 3.492552600786041e-05,
|
| 2494 |
-
"loss": 2.5152,
|
| 2495 |
-
"step": 204000
|
| 2496 |
-
},
|
| 2497 |
-
{
|
| 2498 |
-
"epoch": 3.0,
|
| 2499 |
-
"learning_rate": 3.470582092805431e-05,
|
| 2500 |
-
"loss": 2.4693,
|
| 2501 |
-
"step": 204500
|
| 2502 |
-
},
|
| 2503 |
-
{
|
| 2504 |
-
"epoch": 3.0,
|
| 2505 |
-
"eval_bleu": 1.0,
|
| 2506 |
-
"eval_brevity_penalty": 1.0,
|
| 2507 |
-
"eval_length_ratio": 1.0,
|
| 2508 |
-
"eval_loss": 2.896097421646118,
|
| 2509 |
-
"eval_precisions": [
|
| 2510 |
-
1.0,
|
| 2511 |
-
1.0,
|
| 2512 |
-
1.0,
|
| 2513 |
-
1.0
|
| 2514 |
-
],
|
| 2515 |
-
"eval_reference_length": 7761920,
|
| 2516 |
-
"eval_runtime": 14493.3817,
|
| 2517 |
-
"eval_samples_per_second": 1.046,
|
| 2518 |
-
"eval_steps_per_second": 0.523,
|
| 2519 |
-
"eval_translation_length": 7761920,
|
| 2520 |
-
"step": 204657
|
| 2521 |
}
|
| 2522 |
],
|
| 2523 |
"logging_steps": 500,
|
| 2524 |
"max_steps": 341095,
|
| 2525 |
"num_train_epochs": 5,
|
| 2526 |
-
"save_steps":
|
| 2527 |
-
"total_flos":
|
| 2528 |
"trial_name": null,
|
| 2529 |
"trial_params": null
|
| 2530 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 68219,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.01,
|
| 13 |
+
"learning_rate": 8e-05,
|
| 14 |
+
"loss": 2.273,
|
| 15 |
"step": 500
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 7.999957460493864e-05,
|
| 20 |
+
"loss": 2.2497,
|
| 21 |
"step": 1000
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.02,
|
| 25 |
+
"learning_rate": 7.999829842880257e-05,
|
| 26 |
+
"loss": 2.3728,
|
| 27 |
"step": 1500
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.03,
|
| 31 |
+
"learning_rate": 7.999617149873574e-05,
|
| 32 |
+
"loss": 2.3576,
|
| 33 |
"step": 2000
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.04,
|
| 37 |
+
"learning_rate": 7.999319385997746e-05,
|
| 38 |
+
"loss": 2.3374,
|
| 39 |
"step": 2500
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.04,
|
| 43 |
+
"learning_rate": 7.998936557586135e-05,
|
| 44 |
+
"loss": 2.3538,
|
| 45 |
"step": 3000
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.05,
|
| 49 |
+
"learning_rate": 7.998468672781407e-05,
|
| 50 |
+
"loss": 2.366,
|
| 51 |
"step": 3500
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.06,
|
| 55 |
+
"learning_rate": 7.997915741535355e-05,
|
| 56 |
+
"loss": 2.3321,
|
| 57 |
"step": 4000
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.07,
|
| 61 |
+
"learning_rate": 7.997277775608694e-05,
|
| 62 |
+
"loss": 2.2838,
|
| 63 |
"step": 4500
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.07,
|
| 67 |
+
"learning_rate": 7.996554788570796e-05,
|
| 68 |
+
"loss": 2.2679,
|
| 69 |
"step": 5000
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.08,
|
| 73 |
+
"learning_rate": 7.995746795799422e-05,
|
| 74 |
+
"loss": 2.4041,
|
| 75 |
"step": 5500
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.09,
|
| 79 |
+
"learning_rate": 7.994853814480376e-05,
|
| 80 |
+
"loss": 2.3415,
|
| 81 |
"step": 6000
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.1,
|
| 85 |
+
"learning_rate": 7.99387586360715e-05,
|
| 86 |
+
"loss": 2.3589,
|
| 87 |
"step": 6500
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.1,
|
| 91 |
+
"learning_rate": 7.992812963980518e-05,
|
| 92 |
+
"loss": 2.3203,
|
| 93 |
"step": 7000
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.11,
|
| 97 |
+
"learning_rate": 7.991665138208094e-05,
|
| 98 |
+
"loss": 2.3517,
|
| 99 |
"step": 7500
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.12,
|
| 103 |
+
"learning_rate": 7.990432410703848e-05,
|
| 104 |
+
"loss": 2.4286,
|
| 105 |
"step": 8000
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.12,
|
| 109 |
+
"learning_rate": 7.989114807687589e-05,
|
| 110 |
+
"loss": 2.3227,
|
| 111 |
"step": 8500
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.13,
|
| 115 |
+
"learning_rate": 7.987712357184408e-05,
|
| 116 |
+
"loss": 2.3945,
|
| 117 |
"step": 9000
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.14,
|
| 121 |
+
"learning_rate": 7.98622508902408e-05,
|
| 122 |
+
"loss": 2.3775,
|
| 123 |
"step": 9500
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 0.15,
|
| 127 |
+
"learning_rate": 7.984653034840432e-05,
|
| 128 |
+
"loss": 2.3522,
|
| 129 |
"step": 10000
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.15,
|
| 133 |
+
"learning_rate": 7.982996228070671e-05,
|
| 134 |
+
"loss": 2.334,
|
| 135 |
"step": 10500
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.16,
|
| 139 |
+
"learning_rate": 7.981254703954664e-05,
|
| 140 |
+
"loss": 2.3742,
|
| 141 |
"step": 11000
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.17,
|
| 145 |
+
"learning_rate": 7.979428499534201e-05,
|
| 146 |
+
"loss": 2.2517,
|
| 147 |
"step": 11500
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.18,
|
| 151 |
+
"learning_rate": 7.977517653652199e-05,
|
| 152 |
+
"loss": 2.3994,
|
| 153 |
"step": 12000
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.18,
|
| 157 |
+
"learning_rate": 7.975522206951876e-05,
|
| 158 |
+
"loss": 2.3606,
|
| 159 |
"step": 12500
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.19,
|
| 163 |
+
"learning_rate": 7.973442201875895e-05,
|
| 164 |
+
"loss": 2.3634,
|
| 165 |
"step": 13000
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 0.2,
|
| 169 |
+
"learning_rate": 7.971277682665446e-05,
|
| 170 |
+
"loss": 2.3061,
|
| 171 |
"step": 13500
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.21,
|
| 175 |
+
"learning_rate": 7.969028695359319e-05,
|
| 176 |
+
"loss": 2.4985,
|
| 177 |
"step": 14000
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.21,
|
| 181 |
+
"learning_rate": 7.966695287792921e-05,
|
| 182 |
+
"loss": 2.3908,
|
| 183 |
"step": 14500
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.22,
|
| 187 |
+
"learning_rate": 7.96427750959725e-05,
|
| 188 |
+
"loss": 2.3605,
|
| 189 |
"step": 15000
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.23,
|
| 193 |
+
"learning_rate": 7.961775412197857e-05,
|
| 194 |
+
"loss": 2.347,
|
| 195 |
"step": 15500
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 0.23,
|
| 199 |
+
"learning_rate": 7.959189048813735e-05,
|
| 200 |
+
"loss": 2.3233,
|
| 201 |
"step": 16000
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 0.24,
|
| 205 |
+
"learning_rate": 7.95651847445619e-05,
|
| 206 |
+
"loss": 2.3415,
|
| 207 |
"step": 16500
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 0.25,
|
| 211 |
+
"learning_rate": 7.953763745927682e-05,
|
| 212 |
+
"loss": 2.4679,
|
| 213 |
"step": 17000
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.26,
|
| 217 |
+
"learning_rate": 7.950924921820606e-05,
|
| 218 |
+
"loss": 2.443,
|
| 219 |
"step": 17500
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.26,
|
| 223 |
+
"learning_rate": 7.948002062516052e-05,
|
| 224 |
+
"loss": 2.5141,
|
| 225 |
"step": 18000
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.27,
|
| 229 |
+
"learning_rate": 7.944995230182513e-05,
|
| 230 |
+
"loss": 2.4339,
|
| 231 |
"step": 18500
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.28,
|
| 235 |
+
"learning_rate": 7.941904488774571e-05,
|
| 236 |
+
"loss": 2.4308,
|
| 237 |
"step": 19000
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.29,
|
| 241 |
+
"learning_rate": 7.938729904031533e-05,
|
| 242 |
+
"loss": 2.4312,
|
| 243 |
"step": 19500
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.29,
|
| 247 |
+
"learning_rate": 7.93547154347603e-05,
|
| 248 |
+
"loss": 2.4756,
|
| 249 |
"step": 20000
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.3,
|
| 253 |
+
"learning_rate": 7.932129476412592e-05,
|
| 254 |
+
"loss": 2.4426,
|
| 255 |
"step": 20500
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.31,
|
| 259 |
+
"learning_rate": 7.928703773926155e-05,
|
| 260 |
+
"loss": 2.4779,
|
| 261 |
"step": 21000
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.32,
|
| 265 |
+
"learning_rate": 7.925194508880567e-05,
|
| 266 |
+
"loss": 2.4671,
|
| 267 |
"step": 21500
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 0.32,
|
| 271 |
+
"learning_rate": 7.921601755917029e-05,
|
| 272 |
+
"loss": 2.4473,
|
| 273 |
"step": 22000
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 0.33,
|
| 277 |
+
"learning_rate": 7.917925591452508e-05,
|
| 278 |
+
"loss": 2.3929,
|
| 279 |
"step": 22500
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.34,
|
| 283 |
+
"learning_rate": 7.914166093678117e-05,
|
| 284 |
+
"loss": 2.4158,
|
| 285 |
"step": 23000
|
| 286 |
},
|
| 287 |
{
|
| 288 |
"epoch": 0.34,
|
| 289 |
+
"learning_rate": 7.910323342557442e-05,
|
| 290 |
+
"loss": 2.4607,
|
| 291 |
"step": 23500
|
| 292 |
},
|
| 293 |
{
|
| 294 |
"epoch": 0.35,
|
| 295 |
+
"learning_rate": 7.906397419824855e-05,
|
| 296 |
+
"loss": 2.4866,
|
| 297 |
"step": 24000
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 0.36,
|
| 301 |
+
"learning_rate": 7.902388408983759e-05,
|
| 302 |
+
"loss": 2.3708,
|
| 303 |
"step": 24500
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"epoch": 0.37,
|
| 307 |
+
"learning_rate": 7.898296395304824e-05,
|
| 308 |
+
"loss": 2.4718,
|
| 309 |
"step": 25000
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.37,
|
| 313 |
+
"learning_rate": 7.894121465824175e-05,
|
| 314 |
+
"loss": 2.4436,
|
| 315 |
"step": 25500
|
| 316 |
},
|
| 317 |
{
|
| 318 |
"epoch": 0.38,
|
| 319 |
+
"learning_rate": 7.889863709341528e-05,
|
| 320 |
+
"loss": 2.498,
|
| 321 |
"step": 26000
|
| 322 |
},
|
| 323 |
{
|
| 324 |
"epoch": 0.39,
|
| 325 |
+
"learning_rate": 7.885523216418312e-05,
|
| 326 |
+
"loss": 2.4418,
|
| 327 |
"step": 26500
|
| 328 |
},
|
| 329 |
{
|
| 330 |
"epoch": 0.4,
|
| 331 |
+
"learning_rate": 7.881100079375742e-05,
|
| 332 |
+
"loss": 2.3653,
|
| 333 |
"step": 27000
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.4,
|
| 337 |
+
"learning_rate": 7.876594392292848e-05,
|
| 338 |
+
"loss": 2.5256,
|
| 339 |
"step": 27500
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.41,
|
| 343 |
+
"learning_rate": 7.872006251004482e-05,
|
| 344 |
+
"loss": 2.549,
|
| 345 |
"step": 28000
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"epoch": 0.42,
|
| 349 |
+
"learning_rate": 7.867335753099278e-05,
|
| 350 |
+
"loss": 2.4543,
|
| 351 |
"step": 28500
|
| 352 |
},
|
| 353 |
{
|
| 354 |
"epoch": 0.43,
|
| 355 |
+
"learning_rate": 7.86258299791757e-05,
|
| 356 |
+
"loss": 2.4647,
|
| 357 |
"step": 29000
|
| 358 |
},
|
| 359 |
{
|
| 360 |
"epoch": 0.43,
|
| 361 |
+
"learning_rate": 7.857748086549292e-05,
|
| 362 |
+
"loss": 2.5375,
|
| 363 |
"step": 29500
|
| 364 |
},
|
| 365 |
{
|
| 366 |
"epoch": 0.44,
|
| 367 |
+
"learning_rate": 7.852831121831812e-05,
|
| 368 |
+
"loss": 2.4895,
|
| 369 |
"step": 30000
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.45,
|
| 373 |
+
"learning_rate": 7.847832208347754e-05,
|
| 374 |
+
"loss": 2.4156,
|
| 375 |
"step": 30500
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.45,
|
| 379 |
+
"learning_rate": 7.842751452422775e-05,
|
| 380 |
+
"loss": 2.5006,
|
| 381 |
"step": 31000
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 0.46,
|
| 385 |
+
"learning_rate": 7.8375889621233e-05,
|
| 386 |
+
"loss": 2.459,
|
| 387 |
"step": 31500
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"epoch": 0.47,
|
| 391 |
+
"learning_rate": 7.83234484725422e-05,
|
| 392 |
+
"loss": 2.469,
|
| 393 |
"step": 32000
|
| 394 |
},
|
| 395 |
{
|
| 396 |
"epoch": 0.48,
|
| 397 |
+
"learning_rate": 7.827019219356568e-05,
|
| 398 |
+
"loss": 2.4331,
|
| 399 |
"step": 32500
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.48,
|
| 403 |
+
"learning_rate": 7.821612191705128e-05,
|
| 404 |
+
"loss": 2.487,
|
| 405 |
"step": 33000
|
| 406 |
},
|
| 407 |
{
|
| 408 |
"epoch": 0.49,
|
| 409 |
+
"learning_rate": 7.816123879306048e-05,
|
| 410 |
+
"loss": 2.5139,
|
| 411 |
"step": 33500
|
| 412 |
},
|
| 413 |
{
|
| 414 |
"epoch": 0.5,
|
| 415 |
+
"learning_rate": 7.810554398894376e-05,
|
| 416 |
+
"loss": 2.5117,
|
| 417 |
"step": 34000
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.51,
|
| 421 |
+
"learning_rate": 7.804903868931584e-05,
|
| 422 |
+
"loss": 2.5537,
|
| 423 |
"step": 34500
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 0.51,
|
| 427 |
+
"learning_rate": 7.79917240960305e-05,
|
| 428 |
+
"loss": 2.4394,
|
| 429 |
"step": 35000
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.52,
|
| 433 |
+
"learning_rate": 7.7933601428155e-05,
|
| 434 |
+
"loss": 2.4285,
|
| 435 |
"step": 35500
|
| 436 |
},
|
| 437 |
{
|
| 438 |
"epoch": 0.53,
|
| 439 |
+
"learning_rate": 7.78746719219441e-05,
|
| 440 |
+
"loss": 2.5693,
|
| 441 |
"step": 36000
|
| 442 |
},
|
| 443 |
{
|
| 444 |
"epoch": 0.54,
|
| 445 |
+
"learning_rate": 7.781493683081388e-05,
|
| 446 |
+
"loss": 2.4932,
|
| 447 |
"step": 36500
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"epoch": 0.54,
|
| 451 |
+
"learning_rate": 7.775439742531495e-05,
|
| 452 |
+
"loss": 2.4468,
|
| 453 |
"step": 37000
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 0.55,
|
| 457 |
+
"learning_rate": 7.769305499310553e-05,
|
| 458 |
+
"loss": 2.4614,
|
| 459 |
"step": 37500
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.56,
|
| 463 |
+
"learning_rate": 7.763091083892402e-05,
|
| 464 |
+
"loss": 2.6549,
|
| 465 |
"step": 38000
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 0.56,
|
| 469 |
+
"learning_rate": 7.756796628456121e-05,
|
| 470 |
+
"loss": 2.4871,
|
| 471 |
"step": 38500
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"epoch": 0.57,
|
| 475 |
+
"learning_rate": 7.750422266883222e-05,
|
| 476 |
+
"loss": 2.413,
|
| 477 |
"step": 39000
|
| 478 |
},
|
| 479 |
{
|
| 480 |
"epoch": 0.58,
|
| 481 |
+
"learning_rate": 7.743968134754806e-05,
|
| 482 |
+
"loss": 2.3566,
|
| 483 |
"step": 39500
|
| 484 |
},
|
| 485 |
{
|
| 486 |
"epoch": 0.59,
|
| 487 |
+
"learning_rate": 7.737434369348664e-05,
|
| 488 |
+
"loss": 2.4936,
|
| 489 |
"step": 40000
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.59,
|
| 493 |
+
"learning_rate": 7.730821109636379e-05,
|
| 494 |
+
"loss": 2.453,
|
| 495 |
"step": 40500
|
| 496 |
},
|
| 497 |
{
|
| 498 |
"epoch": 0.6,
|
| 499 |
+
"learning_rate": 7.724128496280346e-05,
|
| 500 |
+
"loss": 2.4851,
|
| 501 |
"step": 41000
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.61,
|
| 505 |
+
"learning_rate": 7.717356671630802e-05,
|
| 506 |
+
"loss": 2.4564,
|
| 507 |
"step": 41500
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 0.62,
|
| 511 |
+
"learning_rate": 7.710505779722786e-05,
|
| 512 |
+
"loss": 2.5133,
|
| 513 |
"step": 42000
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 0.62,
|
| 517 |
+
"learning_rate": 7.703575966273073e-05,
|
| 518 |
+
"loss": 2.4546,
|
| 519 |
"step": 42500
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.63,
|
| 523 |
+
"learning_rate": 7.696567378677089e-05,
|
| 524 |
+
"loss": 2.5157,
|
| 525 |
"step": 43000
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 0.64,
|
| 529 |
+
"learning_rate": 7.689480166005756e-05,
|
| 530 |
+
"loss": 2.4248,
|
| 531 |
"step": 43500
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 0.64,
|
| 535 |
+
"learning_rate": 7.682314479002344e-05,
|
| 536 |
+
"loss": 2.4853,
|
| 537 |
"step": 44000
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 0.65,
|
| 541 |
+
"learning_rate": 7.67507047007924e-05,
|
| 542 |
+
"loss": 2.4615,
|
| 543 |
"step": 44500
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.66,
|
| 547 |
+
"learning_rate": 7.667748293314729e-05,
|
| 548 |
+
"loss": 2.5391,
|
| 549 |
"step": 45000
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.67,
|
| 553 |
+
"learning_rate": 7.6603481044497e-05,
|
| 554 |
+
"loss": 2.4464,
|
| 555 |
"step": 45500
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 0.67,
|
| 559 |
+
"learning_rate": 7.652870060884345e-05,
|
| 560 |
+
"loss": 2.4941,
|
| 561 |
"step": 46000
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 0.68,
|
| 565 |
+
"learning_rate": 7.645314321674803e-05,
|
| 566 |
+
"loss": 2.4708,
|
| 567 |
"step": 46500
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 0.69,
|
| 571 |
+
"learning_rate": 7.637681047529781e-05,
|
| 572 |
+
"loss": 2.5972,
|
| 573 |
"step": 47000
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 0.7,
|
| 577 |
+
"learning_rate": 7.629970400807136e-05,
|
| 578 |
+
"loss": 2.5369,
|
| 579 |
"step": 47500
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.7,
|
| 583 |
+
"learning_rate": 7.622182545510419e-05,
|
| 584 |
+
"loss": 2.4348,
|
| 585 |
"step": 48000
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 0.71,
|
| 589 |
+
"learning_rate": 7.61431764728539e-05,
|
| 590 |
+
"loss": 2.4546,
|
| 591 |
"step": 48500
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 0.72,
|
| 595 |
+
"learning_rate": 7.606375873416491e-05,
|
| 596 |
+
"loss": 2.4378,
|
| 597 |
"step": 49000
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 0.73,
|
| 601 |
+
"learning_rate": 7.598357392823292e-05,
|
| 602 |
+
"loss": 2.471,
|
| 603 |
"step": 49500
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 0.73,
|
| 607 |
+
"learning_rate": 7.590262376056896e-05,
|
| 608 |
+
"loss": 2.4677,
|
| 609 |
"step": 50000
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.74,
|
| 613 |
+
"learning_rate": 7.58209099529631e-05,
|
| 614 |
+
"loss": 2.4245,
|
| 615 |
"step": 50500
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.75,
|
| 619 |
+
"learning_rate": 7.573843424344783e-05,
|
| 620 |
+
"loss": 2.4734,
|
| 621 |
"step": 51000
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 0.75,
|
| 625 |
+
"learning_rate": 7.565519838626113e-05,
|
| 626 |
+
"loss": 2.4158,
|
| 627 |
"step": 51500
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 0.76,
|
| 631 |
+
"learning_rate": 7.557120415180916e-05,
|
| 632 |
+
"loss": 2.5098,
|
| 633 |
"step": 52000
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 0.77,
|
| 637 |
+
"learning_rate": 7.548645332662853e-05,
|
| 638 |
+
"loss": 2.5478,
|
| 639 |
"step": 52500
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.78,
|
| 643 |
+
"learning_rate": 7.540094771334835e-05,
|
| 644 |
+
"loss": 2.4502,
|
| 645 |
"step": 53000
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 0.78,
|
| 649 |
+
"learning_rate": 7.531468913065192e-05,
|
| 650 |
+
"loss": 2.4264,
|
| 651 |
"step": 53500
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 0.79,
|
| 655 |
+
"learning_rate": 7.522767941323798e-05,
|
| 656 |
+
"loss": 2.4903,
|
| 657 |
"step": 54000
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.8,
|
| 661 |
+
"learning_rate": 7.513992041178174e-05,
|
| 662 |
+
"loss": 2.4486,
|
| 663 |
"step": 54500
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 0.81,
|
| 667 |
+
"learning_rate": 7.505141399289549e-05,
|
| 668 |
+
"loss": 2.5171,
|
| 669 |
"step": 55000
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.81,
|
| 673 |
+
"learning_rate": 7.496216203908891e-05,
|
| 674 |
+
"loss": 2.5396,
|
| 675 |
"step": 55500
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 0.82,
|
| 679 |
+
"learning_rate": 7.487216644872901e-05,
|
| 680 |
+
"loss": 2.4514,
|
| 681 |
"step": 56000
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 0.83,
|
| 685 |
+
"learning_rate": 7.478142913599978e-05,
|
| 686 |
+
"loss": 2.4017,
|
| 687 |
"step": 56500
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 0.84,
|
| 691 |
+
"learning_rate": 7.468995203086146e-05,
|
| 692 |
+
"loss": 2.4591,
|
| 693 |
"step": 57000
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 0.84,
|
| 697 |
+
"learning_rate": 7.459773707900946e-05,
|
| 698 |
+
"loss": 2.5764,
|
| 699 |
"step": 57500
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.85,
|
| 703 |
+
"learning_rate": 7.450478624183306e-05,
|
| 704 |
+
"loss": 2.5013,
|
| 705 |
"step": 58000
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 0.86,
|
| 709 |
+
"learning_rate": 7.441110149637363e-05,
|
| 710 |
+
"loss": 2.51,
|
| 711 |
"step": 58500
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 0.86,
|
| 715 |
+
"learning_rate": 7.431668483528254e-05,
|
| 716 |
+
"loss": 2.3992,
|
| 717 |
"step": 59000
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 0.87,
|
| 721 |
+
"learning_rate": 7.422153826677887e-05,
|
| 722 |
+
"loss": 2.4671,
|
| 723 |
"step": 59500
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 0.88,
|
| 727 |
+
"learning_rate": 7.412566381460662e-05,
|
| 728 |
+
"loss": 2.5362,
|
| 729 |
"step": 60000
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.89,
|
| 733 |
+
"learning_rate": 7.402906351799175e-05,
|
| 734 |
+
"loss": 2.4981,
|
| 735 |
"step": 60500
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 0.89,
|
| 739 |
+
"learning_rate": 7.39317394315987e-05,
|
| 740 |
+
"loss": 2.4574,
|
| 741 |
"step": 61000
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 0.9,
|
| 745 |
+
"learning_rate": 7.383369362548674e-05,
|
| 746 |
+
"loss": 2.4777,
|
| 747 |
"step": 61500
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 0.91,
|
| 751 |
+
"learning_rate": 7.373492818506597e-05,
|
| 752 |
+
"loss": 2.4628,
|
| 753 |
"step": 62000
|
| 754 |
},
|
| 755 |
{
|
| 756 |
"epoch": 0.92,
|
| 757 |
+
"learning_rate": 7.363544521105292e-05,
|
| 758 |
+
"loss": 2.4506,
|
| 759 |
"step": 62500
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.92,
|
| 763 |
+
"learning_rate": 7.353524681942585e-05,
|
| 764 |
+
"loss": 2.5644,
|
| 765 |
"step": 63000
|
| 766 |
},
|
| 767 |
{
|
| 768 |
"epoch": 0.93,
|
| 769 |
+
"learning_rate": 7.343433514137987e-05,
|
| 770 |
+
"loss": 2.5131,
|
| 771 |
"step": 63500
|
| 772 |
},
|
| 773 |
{
|
| 774 |
"epoch": 0.94,
|
| 775 |
+
"learning_rate": 7.333271232328141e-05,
|
| 776 |
+
"loss": 2.5575,
|
| 777 |
"step": 64000
|
| 778 |
},
|
| 779 |
{
|
| 780 |
"epoch": 0.95,
|
| 781 |
+
"learning_rate": 7.32303805266227e-05,
|
| 782 |
+
"loss": 2.5901,
|
| 783 |
"step": 64500
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 0.95,
|
| 787 |
+
"learning_rate": 7.312734192797583e-05,
|
| 788 |
+
"loss": 2.5701,
|
| 789 |
"step": 65000
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 0.96,
|
| 793 |
+
"learning_rate": 7.302359871894635e-05,
|
| 794 |
+
"loss": 2.5201,
|
| 795 |
"step": 65500
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 0.97,
|
| 799 |
+
"learning_rate": 7.291915310612666e-05,
|
| 800 |
+
"loss": 2.5402,
|
| 801 |
"step": 66000
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 0.97,
|
| 805 |
+
"learning_rate": 7.281400731104918e-05,
|
| 806 |
+
"loss": 2.4797,
|
| 807 |
"step": 66500
|
| 808 |
},
|
| 809 |
{
|
| 810 |
"epoch": 0.98,
|
| 811 |
+
"learning_rate": 7.2708163570139e-05,
|
| 812 |
+
"loss": 2.4836,
|
| 813 |
"step": 67000
|
| 814 |
},
|
| 815 |
{
|
| 816 |
"epoch": 0.99,
|
| 817 |
+
"learning_rate": 7.260162413466636e-05,
|
| 818 |
+
"loss": 2.4707,
|
| 819 |
"step": 67500
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 1.0,
|
| 823 |
+
"learning_rate": 7.249439127069873e-05,
|
| 824 |
+
"loss": 2.4334,
|
| 825 |
"step": 68000
|
| 826 |
},
|
| 827 |
{
|
|
|
|
| 829 |
"eval_bleu": 1.0,
|
| 830 |
"eval_brevity_penalty": 1.0,
|
| 831 |
"eval_length_ratio": 1.0,
|
| 832 |
+
"eval_loss": 2.243044853210449,
|
| 833 |
"eval_precisions": [
|
| 834 |
1.0,
|
| 835 |
1.0,
|
|
|
|
| 837 |
1.0
|
| 838 |
],
|
| 839 |
"eval_reference_length": 7761920,
|
| 840 |
+
"eval_runtime": 17988.3779,
|
| 841 |
+
"eval_samples_per_second": 0.843,
|
| 842 |
+
"eval_steps_per_second": 0.421,
|
| 843 |
"eval_translation_length": 7761920,
|
| 844 |
"step": 68219
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 845 |
}
|
| 846 |
],
|
| 847 |
"logging_steps": 500,
|
| 848 |
"max_steps": 341095,
|
| 849 |
"num_train_epochs": 5,
|
| 850 |
+
"save_steps": 5000,
|
| 851 |
+
"total_flos": 3.144579296777994e+17,
|
| 852 |
"trial_name": null,
|
| 853 |
"trial_params": null
|
| 854 |
}
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4728
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dcc6271f614588d827bad076034b6eed6bf1dc0f937c1c713e0d7ec1959bdb0d
|
| 3 |
size 4728
|