Commit
•
5df743c
1
Parent(s):
7d002ce
Training in progress, step 7000, checkpoint
Browse files- last-checkpoint/model-00001-of-00002.safetensors +1 -1
- last-checkpoint/model-00002-of-00002.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 +1092 -345
- last-checkpoint/training_args.bin +1 -1
last-checkpoint/model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4993448880
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6cdd883a5460a3d6fef3ba86dc48a5c7818f0b59f546569efce6e0c6c05644fb
|
| 3 |
size 4993448880
|
last-checkpoint/model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1180663192
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3c3361154434454a2c82e489b524924f18144e43f09f9bc443707201e953d6cb
|
| 3 |
size 1180663192
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3095446256
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8225f7d2b350449309cf6cb95eacece99044be12a3a7aa483bdf0783e22ec40f
|
| 3 |
size 3095446256
|
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:da289394bd41e2cdd34e1fa9b7ba6ae9221802035398fe0e37a4ddc28b5db9f6
|
| 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 sha256:380aebeeba070e703dcd469862cff7b4aace5b115dadb0d7385847d018530611
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,1016 +1,1763 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "warmestman/whisper-large-v3-mn-cv-fleurs/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 1000,
|
| 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": 0.15,
|
| 13 |
-
"learning_rate": 5e-
|
| 14 |
-
"loss": 1.
|
| 15 |
"step": 25
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.3,
|
| 19 |
-
"learning_rate": 1e-
|
| 20 |
-
"loss":
|
| 21 |
"step": 50
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.45,
|
| 25 |
-
"learning_rate": 1.5e-
|
| 26 |
-
"loss":
|
| 27 |
"step": 75
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.6,
|
| 31 |
-
"learning_rate": 2e-
|
| 32 |
-
"loss":
|
| 33 |
"step": 100
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.75,
|
| 37 |
-
"learning_rate": 2.5e-
|
| 38 |
-
"loss":
|
| 39 |
"step": 125
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.9,
|
| 43 |
-
"learning_rate": 3e-
|
| 44 |
-
"loss":
|
| 45 |
"step": 150
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.05,
|
| 49 |
-
"learning_rate": 3.5e-
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 175
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 1.2,
|
| 55 |
-
"learning_rate": 4e-
|
| 56 |
-
"loss": 0.
|
| 57 |
"step": 200
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 1.35,
|
| 61 |
-
"learning_rate": 4.5e-
|
| 62 |
-
"loss": 0.
|
| 63 |
"step": 225
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 1.5,
|
| 67 |
-
"learning_rate": 5e-
|
| 68 |
-
"loss": 0.
|
| 69 |
"step": 250
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 1.65,
|
| 73 |
-
"learning_rate": 5.
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 275
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8,
|
| 79 |
-
"learning_rate": 6e-
|
| 80 |
-
"loss": 0.
|
| 81 |
"step": 300
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 1.95,
|
| 85 |
-
"learning_rate": 6.
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 325
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.1,
|
| 91 |
-
"learning_rate": 7e-
|
| 92 |
-
"loss": 0.
|
| 93 |
"step": 350
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 2.25,
|
| 97 |
-
"learning_rate": 7.
|
| 98 |
-
"loss": 0.
|
| 99 |
"step": 375
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 2.4,
|
| 103 |
-
"learning_rate": 8e-
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 400
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.54,
|
| 109 |
-
"learning_rate": 8.
|
| 110 |
-
"loss": 0.
|
| 111 |
"step": 425
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 2.69,
|
| 115 |
-
"learning_rate": 9e-
|
| 116 |
-
"loss": 0.
|
| 117 |
"step": 450
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 2.84,
|
| 121 |
-
"learning_rate": 9.
|
| 122 |
-
"loss": 0.
|
| 123 |
"step": 475
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 2.99,
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss": 0.
|
| 129 |
"step": 500
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 3.14,
|
| 133 |
-
"learning_rate": 9.
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 525
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.29,
|
| 139 |
-
"learning_rate": 9.
|
| 140 |
-
"loss": 0.
|
| 141 |
"step": 550
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 3.44,
|
| 145 |
-
"learning_rate": 9.
|
| 146 |
-
"loss": 0.
|
| 147 |
"step": 575
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 3.59,
|
| 151 |
-
"learning_rate": 9.
|
| 152 |
-
"loss": 0.
|
| 153 |
"step": 600
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 3.74,
|
| 157 |
-
"learning_rate": 9.
|
| 158 |
-
"loss": 0.
|
| 159 |
"step": 625
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 3.89,
|
| 163 |
-
"learning_rate": 9.
|
| 164 |
-
"loss": 0.
|
| 165 |
"step": 650
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 4.04,
|
| 169 |
-
"learning_rate": 9.
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 675
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 4.19,
|
| 175 |
-
"learning_rate": 9.
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 700
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 4.34,
|
| 181 |
-
"learning_rate": 9.
|
| 182 |
-
"loss": 0.
|
| 183 |
"step": 725
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 4.49,
|
| 187 |
-
"learning_rate": 9.
|
| 188 |
-
"loss": 0.
|
| 189 |
"step": 750
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 4.64,
|
| 193 |
-
"learning_rate": 9.
|
| 194 |
-
"loss": 0.
|
| 195 |
"step": 775
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 4.79,
|
| 199 |
-
"learning_rate": 9.
|
| 200 |
-
"loss": 0.
|
| 201 |
"step": 800
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 4.94,
|
| 205 |
-
"learning_rate": 9.
|
| 206 |
-
"loss": 0.
|
| 207 |
"step": 825
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 5.09,
|
| 211 |
-
"learning_rate": 9.
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 850
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 5.24,
|
| 217 |
-
"learning_rate": 9.
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 875
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 5.39,
|
| 223 |
-
"learning_rate": 9.
|
| 224 |
-
"loss": 0.
|
| 225 |
"step": 900
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 5.54,
|
| 229 |
-
"learning_rate": 9.
|
| 230 |
-
"loss": 0.
|
| 231 |
"step": 925
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 5.69,
|
| 235 |
-
"learning_rate": 9.
|
| 236 |
-
"loss": 0.
|
| 237 |
"step": 950
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 5.84,
|
| 241 |
-
"learning_rate": 9.
|
| 242 |
-
"loss": 0.
|
| 243 |
"step": 975
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 5.99,
|
| 247 |
-
"learning_rate": 9.
|
| 248 |
-
"loss": 0.
|
| 249 |
"step": 1000
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 5.99,
|
| 253 |
-
"eval_loss": 0.
|
| 254 |
-
"eval_runtime":
|
| 255 |
-
"eval_samples_per_second": 0.
|
| 256 |
"eval_steps_per_second": 0.091,
|
| 257 |
-
"eval_wer":
|
| 258 |
"step": 1000
|
| 259 |
},
|
| 260 |
{
|
| 261 |
"epoch": 6.14,
|
| 262 |
-
"learning_rate": 9.
|
| 263 |
-
"loss": 0.
|
| 264 |
"step": 1025
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 6.29,
|
| 268 |
-
"learning_rate": 9.
|
| 269 |
-
"loss": 0.
|
| 270 |
"step": 1050
|
| 271 |
},
|
| 272 |
{
|
| 273 |
"epoch": 6.44,
|
| 274 |
-
"learning_rate": 9.
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 1075
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 6.59,
|
| 280 |
-
"learning_rate": 9.
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 1100
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 6.74,
|
| 286 |
-
"learning_rate": 9.
|
| 287 |
-
"loss": 0.
|
| 288 |
"step": 1125
|
| 289 |
},
|
| 290 |
{
|
| 291 |
"epoch": 6.89,
|
| 292 |
-
"learning_rate": 9.
|
| 293 |
-
"loss": 0.
|
| 294 |
"step": 1150
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 7.04,
|
| 298 |
-
"learning_rate": 9.
|
| 299 |
-
"loss": 0.
|
| 300 |
"step": 1175
|
| 301 |
},
|
| 302 |
{
|
| 303 |
"epoch": 7.19,
|
| 304 |
-
"learning_rate": 9.
|
| 305 |
-
"loss": 0.
|
| 306 |
"step": 1200
|
| 307 |
},
|
| 308 |
{
|
| 309 |
"epoch": 7.34,
|
| 310 |
-
"learning_rate": 9.
|
| 311 |
-
"loss": 0.
|
| 312 |
"step": 1225
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 7.49,
|
| 316 |
-
"learning_rate": 9.
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 1250
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 7.63,
|
| 322 |
-
"learning_rate": 9.
|
| 323 |
-
"loss": 0.
|
| 324 |
"step": 1275
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 7.78,
|
| 328 |
-
"learning_rate": 9.
|
| 329 |
-
"loss": 0.
|
| 330 |
"step": 1300
|
| 331 |
},
|
| 332 |
{
|
| 333 |
"epoch": 7.93,
|
| 334 |
-
"learning_rate": 9.
|
| 335 |
-
"loss": 0.
|
| 336 |
"step": 1325
|
| 337 |
},
|
| 338 |
{
|
| 339 |
"epoch": 8.08,
|
| 340 |
-
"learning_rate": 9.
|
| 341 |
-
"loss": 0.
|
| 342 |
"step": 1350
|
| 343 |
},
|
| 344 |
{
|
| 345 |
"epoch": 8.23,
|
| 346 |
-
"learning_rate": 9.
|
| 347 |
-
"loss": 0.
|
| 348 |
"step": 1375
|
| 349 |
},
|
| 350 |
{
|
| 351 |
"epoch": 8.38,
|
| 352 |
-
"learning_rate": 9.
|
| 353 |
-
"loss": 0.
|
| 354 |
"step": 1400
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 8.53,
|
| 358 |
-
"learning_rate": 9.
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 1425
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 8.68,
|
| 364 |
-
"learning_rate":
|
| 365 |
-
"loss": 0.
|
| 366 |
"step": 1450
|
| 367 |
},
|
| 368 |
{
|
| 369 |
"epoch": 8.83,
|
| 370 |
-
"learning_rate":
|
| 371 |
-
"loss": 0.
|
| 372 |
"step": 1475
|
| 373 |
},
|
| 374 |
{
|
| 375 |
"epoch": 8.98,
|
| 376 |
-
"learning_rate":
|
| 377 |
-
"loss": 0.
|
| 378 |
"step": 1500
|
| 379 |
},
|
| 380 |
{
|
| 381 |
"epoch": 9.13,
|
| 382 |
-
"learning_rate":
|
| 383 |
-
"loss": 0.
|
| 384 |
"step": 1525
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 9.28,
|
| 388 |
-
"learning_rate":
|
| 389 |
-
"loss": 0.
|
| 390 |
"step": 1550
|
| 391 |
},
|
| 392 |
{
|
| 393 |
"epoch": 9.43,
|
| 394 |
-
"learning_rate":
|
| 395 |
-
"loss": 0.
|
| 396 |
"step": 1575
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 9.58,
|
| 400 |
-
"learning_rate":
|
| 401 |
-
"loss": 0.
|
| 402 |
"step": 1600
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 9.73,
|
| 406 |
-
"learning_rate":
|
| 407 |
-
"loss": 0.
|
| 408 |
"step": 1625
|
| 409 |
},
|
| 410 |
{
|
| 411 |
"epoch": 9.88,
|
| 412 |
-
"learning_rate":
|
| 413 |
-
"loss": 0.
|
| 414 |
"step": 1650
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 10.03,
|
| 418 |
-
"learning_rate":
|
| 419 |
-
"loss": 0.
|
| 420 |
"step": 1675
|
| 421 |
},
|
| 422 |
{
|
| 423 |
"epoch": 10.18,
|
| 424 |
-
"learning_rate":
|
| 425 |
-
"loss": 0.
|
| 426 |
"step": 1700
|
| 427 |
},
|
| 428 |
{
|
| 429 |
"epoch": 10.33,
|
| 430 |
-
"learning_rate":
|
| 431 |
-
"loss": 0.
|
| 432 |
"step": 1725
|
| 433 |
},
|
| 434 |
{
|
| 435 |
"epoch": 10.48,
|
| 436 |
-
"learning_rate":
|
| 437 |
-
"loss": 0.
|
| 438 |
"step": 1750
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 10.63,
|
| 442 |
-
"learning_rate":
|
| 443 |
-
"loss": 0.
|
| 444 |
"step": 1775
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 10.78,
|
| 448 |
-
"learning_rate":
|
| 449 |
-
"loss": 0.
|
| 450 |
"step": 1800
|
| 451 |
},
|
| 452 |
{
|
| 453 |
"epoch": 10.93,
|
| 454 |
-
"learning_rate":
|
| 455 |
-
"loss": 0.
|
| 456 |
"step": 1825
|
| 457 |
},
|
| 458 |
{
|
| 459 |
"epoch": 11.08,
|
| 460 |
-
"learning_rate":
|
| 461 |
-
"loss": 0.
|
| 462 |
"step": 1850
|
| 463 |
},
|
| 464 |
{
|
| 465 |
"epoch": 11.23,
|
| 466 |
-
"learning_rate":
|
| 467 |
-
"loss": 0.
|
| 468 |
"step": 1875
|
| 469 |
},
|
| 470 |
{
|
| 471 |
"epoch": 11.38,
|
| 472 |
-
"learning_rate":
|
| 473 |
-
"loss": 0.
|
| 474 |
"step": 1900
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 11.53,
|
| 478 |
-
"learning_rate":
|
| 479 |
-
"loss": 0.
|
| 480 |
"step": 1925
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 11.68,
|
| 484 |
-
"learning_rate":
|
| 485 |
-
"loss": 0.
|
| 486 |
"step": 1950
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 11.83,
|
| 490 |
-
"learning_rate":
|
| 491 |
-
"loss": 0.
|
| 492 |
"step": 1975
|
| 493 |
},
|
| 494 |
{
|
| 495 |
"epoch": 11.98,
|
| 496 |
-
"learning_rate":
|
| 497 |
-
"loss": 0.
|
| 498 |
"step": 2000
|
| 499 |
},
|
| 500 |
{
|
| 501 |
"epoch": 11.98,
|
| 502 |
-
"eval_loss": 0.
|
| 503 |
-
"eval_runtime":
|
| 504 |
-
"eval_samples_per_second": 0.
|
| 505 |
-
"eval_steps_per_second": 0.
|
| 506 |
-
"eval_wer":
|
| 507 |
"step": 2000
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 12.13,
|
| 511 |
-
"learning_rate":
|
| 512 |
-
"loss": 0.
|
| 513 |
"step": 2025
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 12.28,
|
| 517 |
-
"learning_rate":
|
| 518 |
-
"loss": 0.
|
| 519 |
"step": 2050
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 12.43,
|
| 523 |
-
"learning_rate":
|
| 524 |
-
"loss": 0.
|
| 525 |
"step": 2075
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 12.57,
|
| 529 |
-
"learning_rate":
|
| 530 |
-
"loss": 0.
|
| 531 |
"step": 2100
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 12.72,
|
| 535 |
-
"learning_rate":
|
| 536 |
-
"loss": 0.
|
| 537 |
"step": 2125
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 12.87,
|
| 541 |
-
"learning_rate":
|
| 542 |
-
"loss": 0.
|
| 543 |
"step": 2150
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 13.02,
|
| 547 |
-
"learning_rate":
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 2175
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 13.17,
|
| 553 |
-
"learning_rate":
|
| 554 |
-
"loss": 0.
|
| 555 |
"step": 2200
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 13.32,
|
| 559 |
-
"learning_rate":
|
| 560 |
-
"loss": 0.
|
| 561 |
"step": 2225
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 13.47,
|
| 565 |
-
"learning_rate":
|
| 566 |
-
"loss": 0.
|
| 567 |
"step": 2250
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 13.62,
|
| 571 |
-
"learning_rate":
|
| 572 |
-
"loss": 0.
|
| 573 |
"step": 2275
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 13.77,
|
| 577 |
-
"learning_rate":
|
| 578 |
-
"loss": 0.
|
| 579 |
"step": 2300
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 13.92,
|
| 583 |
-
"learning_rate":
|
| 584 |
-
"loss": 0.
|
| 585 |
"step": 2325
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 14.07,
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 2350
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 14.22,
|
| 595 |
-
"learning_rate":
|
| 596 |
-
"loss": 0.
|
| 597 |
"step": 2375
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 14.37,
|
| 601 |
-
"learning_rate":
|
| 602 |
-
"loss": 0.
|
| 603 |
"step": 2400
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 14.52,
|
| 607 |
-
"learning_rate":
|
| 608 |
-
"loss": 0.
|
| 609 |
"step": 2425
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 14.67,
|
| 613 |
-
"learning_rate":
|
| 614 |
-
"loss": 0.
|
| 615 |
"step": 2450
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 14.82,
|
| 619 |
-
"learning_rate":
|
| 620 |
-
"loss": 0.
|
| 621 |
"step": 2475
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 14.97,
|
| 625 |
-
"learning_rate":
|
| 626 |
-
"loss": 0.
|
| 627 |
"step": 2500
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 15.12,
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 2525
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 15.27,
|
| 637 |
-
"learning_rate":
|
| 638 |
-
"loss": 0.
|
| 639 |
"step": 2550
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 15.42,
|
| 643 |
-
"learning_rate":
|
| 644 |
-
"loss": 0.
|
| 645 |
"step": 2575
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 15.57,
|
| 649 |
-
"learning_rate":
|
| 650 |
-
"loss": 0.
|
| 651 |
"step": 2600
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 15.72,
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 0.
|
| 657 |
"step": 2625
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 15.87,
|
| 661 |
-
"learning_rate":
|
| 662 |
-
"loss": 0.
|
| 663 |
"step": 2650
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 16.02,
|
| 667 |
-
"learning_rate":
|
| 668 |
-
"loss": 0.
|
| 669 |
"step": 2675
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 16.17,
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 2700
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 16.32,
|
| 679 |
-
"learning_rate":
|
| 680 |
-
"loss": 0.
|
| 681 |
"step": 2725
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 16.47,
|
| 685 |
-
"learning_rate":
|
| 686 |
-
"loss": 0.
|
| 687 |
"step": 2750
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 16.62,
|
| 691 |
-
"learning_rate":
|
| 692 |
-
"loss": 0.
|
| 693 |
"step": 2775
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 16.77,
|
| 697 |
-
"learning_rate":
|
| 698 |
-
"loss": 0.
|
| 699 |
"step": 2800
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 16.92,
|
| 703 |
-
"learning_rate":
|
| 704 |
-
"loss": 0.
|
| 705 |
"step": 2825
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 17.07,
|
| 709 |
-
"learning_rate":
|
| 710 |
-
"loss": 0.
|
| 711 |
"step": 2850
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 17.22,
|
| 715 |
-
"learning_rate":
|
| 716 |
-
"loss": 0.
|
| 717 |
"step": 2875
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 17.37,
|
| 721 |
-
"learning_rate":
|
| 722 |
-
"loss": 0.
|
| 723 |
"step": 2900
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 17.51,
|
| 727 |
-
"learning_rate":
|
| 728 |
-
"loss": 0.
|
| 729 |
"step": 2925
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 17.66,
|
| 733 |
-
"learning_rate":
|
| 734 |
-
"loss": 0.
|
| 735 |
"step": 2950
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 17.81,
|
| 739 |
-
"learning_rate":
|
| 740 |
-
"loss": 0.
|
| 741 |
"step": 2975
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 17.96,
|
| 745 |
-
"learning_rate":
|
| 746 |
-
"loss": 0.
|
| 747 |
"step": 3000
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 17.96,
|
| 751 |
-
"eval_loss": 0.
|
| 752 |
-
"eval_runtime":
|
| 753 |
-
"eval_samples_per_second": 0.
|
| 754 |
-
"eval_steps_per_second": 0.
|
| 755 |
-
"eval_wer":
|
| 756 |
"step": 3000
|
| 757 |
},
|
| 758 |
{
|
| 759 |
"epoch": 18.11,
|
| 760 |
-
"learning_rate":
|
| 761 |
-
"loss": 0.
|
| 762 |
"step": 3025
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 18.26,
|
| 766 |
-
"learning_rate":
|
| 767 |
-
"loss": 0.
|
| 768 |
"step": 3050
|
| 769 |
},
|
| 770 |
{
|
| 771 |
"epoch": 18.41,
|
| 772 |
-
"learning_rate":
|
| 773 |
-
"loss": 0.
|
| 774 |
"step": 3075
|
| 775 |
},
|
| 776 |
{
|
| 777 |
"epoch": 18.56,
|
| 778 |
-
"learning_rate":
|
| 779 |
-
"loss": 0.
|
| 780 |
"step": 3100
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 18.71,
|
| 784 |
-
"learning_rate":
|
| 785 |
-
"loss": 0.
|
| 786 |
"step": 3125
|
| 787 |
},
|
| 788 |
{
|
| 789 |
"epoch": 18.86,
|
| 790 |
-
"learning_rate":
|
| 791 |
-
"loss": 0.
|
| 792 |
"step": 3150
|
| 793 |
},
|
| 794 |
{
|
| 795 |
"epoch": 19.01,
|
| 796 |
-
"learning_rate":
|
| 797 |
-
"loss": 0.
|
| 798 |
"step": 3175
|
| 799 |
},
|
| 800 |
{
|
| 801 |
"epoch": 19.16,
|
| 802 |
-
"learning_rate":
|
| 803 |
-
"loss": 0.
|
| 804 |
"step": 3200
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 19.31,
|
| 808 |
-
"learning_rate":
|
| 809 |
-
"loss": 0.
|
| 810 |
"step": 3225
|
| 811 |
},
|
| 812 |
{
|
| 813 |
"epoch": 19.46,
|
| 814 |
-
"learning_rate":
|
| 815 |
-
"loss": 0.
|
| 816 |
"step": 3250
|
| 817 |
},
|
| 818 |
{
|
| 819 |
"epoch": 19.61,
|
| 820 |
-
"learning_rate":
|
| 821 |
-
"loss": 0.
|
| 822 |
"step": 3275
|
| 823 |
},
|
| 824 |
{
|
| 825 |
"epoch": 19.76,
|
| 826 |
-
"learning_rate":
|
| 827 |
-
"loss": 0.
|
| 828 |
"step": 3300
|
| 829 |
},
|
| 830 |
{
|
| 831 |
"epoch": 19.91,
|
| 832 |
-
"learning_rate":
|
| 833 |
-
"loss": 0.
|
| 834 |
"step": 3325
|
| 835 |
},
|
| 836 |
{
|
| 837 |
"epoch": 20.06,
|
| 838 |
-
"learning_rate":
|
| 839 |
-
"loss": 0.
|
| 840 |
"step": 3350
|
| 841 |
},
|
| 842 |
{
|
| 843 |
"epoch": 20.21,
|
| 844 |
-
"learning_rate":
|
| 845 |
"loss": 0.0075,
|
| 846 |
"step": 3375
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 20.36,
|
| 850 |
-
"learning_rate":
|
| 851 |
-
"loss": 0.
|
| 852 |
"step": 3400
|
| 853 |
},
|
| 854 |
{
|
| 855 |
"epoch": 20.51,
|
| 856 |
-
"learning_rate":
|
| 857 |
-
"loss": 0.
|
| 858 |
"step": 3425
|
| 859 |
},
|
| 860 |
{
|
| 861 |
"epoch": 20.66,
|
| 862 |
-
"learning_rate":
|
| 863 |
-
"loss": 0.
|
| 864 |
"step": 3450
|
| 865 |
},
|
| 866 |
{
|
| 867 |
"epoch": 20.81,
|
| 868 |
-
"learning_rate":
|
| 869 |
-
"loss": 0.
|
| 870 |
"step": 3475
|
| 871 |
},
|
| 872 |
{
|
| 873 |
"epoch": 20.96,
|
| 874 |
-
"learning_rate":
|
| 875 |
-
"loss": 0.
|
| 876 |
"step": 3500
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 21.11,
|
| 880 |
-
"learning_rate":
|
| 881 |
-
"loss": 0.
|
| 882 |
"step": 3525
|
| 883 |
},
|
| 884 |
{
|
| 885 |
"epoch": 21.26,
|
| 886 |
-
"learning_rate":
|
| 887 |
-
"loss": 0.
|
| 888 |
"step": 3550
|
| 889 |
},
|
| 890 |
{
|
| 891 |
"epoch": 21.41,
|
| 892 |
-
"learning_rate":
|
| 893 |
-
"loss": 0.
|
| 894 |
"step": 3575
|
| 895 |
},
|
| 896 |
{
|
| 897 |
"epoch": 21.56,
|
| 898 |
-
"learning_rate":
|
| 899 |
-
"loss": 0.
|
| 900 |
"step": 3600
|
| 901 |
},
|
| 902 |
{
|
| 903 |
"epoch": 21.71,
|
| 904 |
-
"learning_rate":
|
| 905 |
-
"loss": 0.
|
| 906 |
"step": 3625
|
| 907 |
},
|
| 908 |
{
|
| 909 |
"epoch": 21.86,
|
| 910 |
-
"learning_rate":
|
| 911 |
-
"loss": 0.
|
| 912 |
"step": 3650
|
| 913 |
},
|
| 914 |
{
|
| 915 |
"epoch": 22.01,
|
| 916 |
-
"learning_rate":
|
| 917 |
-
"loss": 0.
|
| 918 |
"step": 3675
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 22.16,
|
| 922 |
-
"learning_rate":
|
| 923 |
-
"loss": 0.
|
| 924 |
"step": 3700
|
| 925 |
},
|
| 926 |
{
|
| 927 |
"epoch": 22.31,
|
| 928 |
-
"learning_rate":
|
| 929 |
-
"loss": 0.
|
| 930 |
"step": 3725
|
| 931 |
},
|
| 932 |
{
|
| 933 |
"epoch": 22.46,
|
| 934 |
-
"learning_rate":
|
| 935 |
-
"loss": 0.
|
| 936 |
"step": 3750
|
| 937 |
},
|
| 938 |
{
|
| 939 |
"epoch": 22.6,
|
| 940 |
-
"learning_rate":
|
| 941 |
-
"loss": 0.
|
| 942 |
"step": 3775
|
| 943 |
},
|
| 944 |
{
|
| 945 |
"epoch": 22.75,
|
| 946 |
-
"learning_rate":
|
| 947 |
-
"loss": 0.
|
| 948 |
"step": 3800
|
| 949 |
},
|
| 950 |
{
|
| 951 |
"epoch": 22.9,
|
| 952 |
-
"learning_rate":
|
| 953 |
-
"loss": 0.
|
| 954 |
"step": 3825
|
| 955 |
},
|
| 956 |
{
|
| 957 |
"epoch": 23.05,
|
| 958 |
-
"learning_rate":
|
| 959 |
-
"loss": 0.
|
| 960 |
"step": 3850
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 23.2,
|
| 964 |
-
"learning_rate":
|
| 965 |
-
"loss": 0.
|
| 966 |
"step": 3875
|
| 967 |
},
|
| 968 |
{
|
| 969 |
"epoch": 23.35,
|
| 970 |
-
"learning_rate":
|
| 971 |
-
"loss": 0.
|
| 972 |
"step": 3900
|
| 973 |
},
|
| 974 |
{
|
| 975 |
"epoch": 23.5,
|
| 976 |
-
"learning_rate":
|
| 977 |
-
"loss": 0.
|
| 978 |
"step": 3925
|
| 979 |
},
|
| 980 |
{
|
| 981 |
"epoch": 23.65,
|
| 982 |
-
"learning_rate":
|
| 983 |
-
"loss": 0.
|
| 984 |
"step": 3950
|
| 985 |
},
|
| 986 |
{
|
| 987 |
"epoch": 23.8,
|
| 988 |
-
"learning_rate":
|
| 989 |
-
"loss": 0.
|
| 990 |
"step": 3975
|
| 991 |
},
|
| 992 |
{
|
| 993 |
"epoch": 23.95,
|
| 994 |
-
"learning_rate":
|
| 995 |
-
"loss": 0.
|
| 996 |
"step": 4000
|
| 997 |
},
|
| 998 |
{
|
| 999 |
"epoch": 23.95,
|
| 1000 |
-
"eval_loss": 0.
|
| 1001 |
-
"eval_runtime":
|
| 1002 |
-
"eval_samples_per_second": 0.
|
| 1003 |
-
"eval_steps_per_second": 0.
|
| 1004 |
-
"eval_wer":
|
| 1005 |
"step": 4000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1006 |
}
|
| 1007 |
],
|
| 1008 |
"logging_steps": 25,
|
| 1009 |
-
"max_steps":
|
| 1010 |
"num_input_tokens_seen": 0,
|
| 1011 |
-
"num_train_epochs":
|
| 1012 |
"save_steps": 1000,
|
| 1013 |
-
"total_flos":
|
| 1014 |
"train_batch_size": 16,
|
| 1015 |
"trial_name": null,
|
| 1016 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 41.91738628238271,
|
| 3 |
+
"best_model_checkpoint": "warmestman/whisper-large-v3-mn-cv-fleurs/checkpoint-6000",
|
| 4 |
+
"epoch": 41.91616766467066,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 7000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.15,
|
| 13 |
+
"learning_rate": 5e-08,
|
| 14 |
+
"loss": 1.4341,
|
| 15 |
"step": 25
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.3,
|
| 19 |
+
"learning_rate": 1e-07,
|
| 20 |
+
"loss": 1.3997,
|
| 21 |
"step": 50
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.45,
|
| 25 |
+
"learning_rate": 1.5e-07,
|
| 26 |
+
"loss": 1.33,
|
| 27 |
"step": 75
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.6,
|
| 31 |
+
"learning_rate": 2e-07,
|
| 32 |
+
"loss": 1.2112,
|
| 33 |
"step": 100
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.75,
|
| 37 |
+
"learning_rate": 2.5e-07,
|
| 38 |
+
"loss": 1.1128,
|
| 39 |
"step": 125
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.9,
|
| 43 |
+
"learning_rate": 3e-07,
|
| 44 |
+
"loss": 1.0554,
|
| 45 |
"step": 150
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 1.05,
|
| 49 |
+
"learning_rate": 3.5e-07,
|
| 50 |
+
"loss": 0.9914,
|
| 51 |
"step": 175
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 1.2,
|
| 55 |
+
"learning_rate": 4e-07,
|
| 56 |
+
"loss": 0.9151,
|
| 57 |
"step": 200
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 1.35,
|
| 61 |
+
"learning_rate": 4.5e-07,
|
| 62 |
+
"loss": 0.8636,
|
| 63 |
"step": 225
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 1.5,
|
| 67 |
+
"learning_rate": 5e-07,
|
| 68 |
+
"loss": 0.7988,
|
| 69 |
"step": 250
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 1.65,
|
| 73 |
+
"learning_rate": 5.5e-07,
|
| 74 |
+
"loss": 0.755,
|
| 75 |
"step": 275
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8,
|
| 79 |
+
"learning_rate": 6e-07,
|
| 80 |
+
"loss": 0.7075,
|
| 81 |
"step": 300
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 1.95,
|
| 85 |
+
"learning_rate": 6.5e-07,
|
| 86 |
+
"loss": 0.6772,
|
| 87 |
"step": 325
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 2.1,
|
| 91 |
+
"learning_rate": 7e-07,
|
| 92 |
+
"loss": 0.6182,
|
| 93 |
"step": 350
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 2.25,
|
| 97 |
+
"learning_rate": 7.5e-07,
|
| 98 |
+
"loss": 0.5912,
|
| 99 |
"step": 375
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 2.4,
|
| 103 |
+
"learning_rate": 8e-07,
|
| 104 |
+
"loss": 0.5273,
|
| 105 |
"step": 400
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.54,
|
| 109 |
+
"learning_rate": 8.499999999999999e-07,
|
| 110 |
+
"loss": 0.5237,
|
| 111 |
"step": 425
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 2.69,
|
| 115 |
+
"learning_rate": 9e-07,
|
| 116 |
+
"loss": 0.4872,
|
| 117 |
"step": 450
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 2.84,
|
| 121 |
+
"learning_rate": 9.499999999999999e-07,
|
| 122 |
+
"loss": 0.4826,
|
| 123 |
"step": 475
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 2.99,
|
| 127 |
+
"learning_rate": 1e-06,
|
| 128 |
+
"loss": 0.4555,
|
| 129 |
"step": 500
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 3.14,
|
| 133 |
+
"learning_rate": 9.987179487179487e-07,
|
| 134 |
+
"loss": 0.4109,
|
| 135 |
"step": 525
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.29,
|
| 139 |
+
"learning_rate": 9.974358974358974e-07,
|
| 140 |
+
"loss": 0.3968,
|
| 141 |
"step": 550
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 3.44,
|
| 145 |
+
"learning_rate": 9.961538461538461e-07,
|
| 146 |
+
"loss": 0.3976,
|
| 147 |
"step": 575
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 3.59,
|
| 151 |
+
"learning_rate": 9.948717948717949e-07,
|
| 152 |
+
"loss": 0.3741,
|
| 153 |
"step": 600
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 3.74,
|
| 157 |
+
"learning_rate": 9.935897435897436e-07,
|
| 158 |
+
"loss": 0.3633,
|
| 159 |
"step": 625
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 3.89,
|
| 163 |
+
"learning_rate": 9.923076923076923e-07,
|
| 164 |
+
"loss": 0.3697,
|
| 165 |
"step": 650
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 4.04,
|
| 169 |
+
"learning_rate": 9.91025641025641e-07,
|
| 170 |
+
"loss": 0.3382,
|
| 171 |
"step": 675
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 4.19,
|
| 175 |
+
"learning_rate": 9.897435897435898e-07,
|
| 176 |
+
"loss": 0.3004,
|
| 177 |
"step": 700
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 4.34,
|
| 181 |
+
"learning_rate": 9.884615384615385e-07,
|
| 182 |
+
"loss": 0.3067,
|
| 183 |
"step": 725
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 4.49,
|
| 187 |
+
"learning_rate": 9.871794871794872e-07,
|
| 188 |
+
"loss": 0.3106,
|
| 189 |
"step": 750
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 4.64,
|
| 193 |
+
"learning_rate": 9.85897435897436e-07,
|
| 194 |
+
"loss": 0.286,
|
| 195 |
"step": 775
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 4.79,
|
| 199 |
+
"learning_rate": 9.846153846153847e-07,
|
| 200 |
+
"loss": 0.2978,
|
| 201 |
"step": 800
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 4.94,
|
| 205 |
+
"learning_rate": 9.833333333333332e-07,
|
| 206 |
+
"loss": 0.2964,
|
| 207 |
"step": 825
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 5.09,
|
| 211 |
+
"learning_rate": 9.820512820512819e-07,
|
| 212 |
+
"loss": 0.2618,
|
| 213 |
"step": 850
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 5.24,
|
| 217 |
+
"learning_rate": 9.807692307692306e-07,
|
| 218 |
+
"loss": 0.2552,
|
| 219 |
"step": 875
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 5.39,
|
| 223 |
+
"learning_rate": 9.794871794871793e-07,
|
| 224 |
+
"loss": 0.2487,
|
| 225 |
"step": 900
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 5.54,
|
| 229 |
+
"learning_rate": 9.78205128205128e-07,
|
| 230 |
+
"loss": 0.2412,
|
| 231 |
"step": 925
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 5.69,
|
| 235 |
+
"learning_rate": 9.769230769230768e-07,
|
| 236 |
+
"loss": 0.2331,
|
| 237 |
"step": 950
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 5.84,
|
| 241 |
+
"learning_rate": 9.756410256410255e-07,
|
| 242 |
+
"loss": 0.2418,
|
| 243 |
"step": 975
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 5.99,
|
| 247 |
+
"learning_rate": 9.743589743589742e-07,
|
| 248 |
+
"loss": 0.24,
|
| 249 |
"step": 1000
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 5.99,
|
| 253 |
+
"eval_loss": 0.4358440637588501,
|
| 254 |
+
"eval_runtime": 584.1166,
|
| 255 |
+
"eval_samples_per_second": 0.717,
|
| 256 |
"eval_steps_per_second": 0.091,
|
| 257 |
+
"eval_wer": 47.676145426544196,
|
| 258 |
"step": 1000
|
| 259 |
},
|
| 260 |
{
|
| 261 |
"epoch": 6.14,
|
| 262 |
+
"learning_rate": 9.73076923076923e-07,
|
| 263 |
+
"loss": 0.2033,
|
| 264 |
"step": 1025
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 6.29,
|
| 268 |
+
"learning_rate": 9.717948717948717e-07,
|
| 269 |
+
"loss": 0.2085,
|
| 270 |
"step": 1050
|
| 271 |
},
|
| 272 |
{
|
| 273 |
"epoch": 6.44,
|
| 274 |
+
"learning_rate": 9.705128205128204e-07,
|
| 275 |
+
"loss": 0.2081,
|
| 276 |
"step": 1075
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 6.59,
|
| 280 |
+
"learning_rate": 9.692307692307691e-07,
|
| 281 |
+
"loss": 0.1931,
|
| 282 |
"step": 1100
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 6.74,
|
| 286 |
+
"learning_rate": 9.679487179487179e-07,
|
| 287 |
+
"loss": 0.2043,
|
| 288 |
"step": 1125
|
| 289 |
},
|
| 290 |
{
|
| 291 |
"epoch": 6.89,
|
| 292 |
+
"learning_rate": 9.666666666666666e-07,
|
| 293 |
+
"loss": 0.1946,
|
| 294 |
"step": 1150
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 7.04,
|
| 298 |
+
"learning_rate": 9.653846153846153e-07,
|
| 299 |
+
"loss": 0.1871,
|
| 300 |
"step": 1175
|
| 301 |
},
|
| 302 |
{
|
| 303 |
"epoch": 7.19,
|
| 304 |
+
"learning_rate": 9.64102564102564e-07,
|
| 305 |
+
"loss": 0.1628,
|
| 306 |
"step": 1200
|
| 307 |
},
|
| 308 |
{
|
| 309 |
"epoch": 7.34,
|
| 310 |
+
"learning_rate": 9.628205128205128e-07,
|
| 311 |
+
"loss": 0.1626,
|
| 312 |
"step": 1225
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 7.49,
|
| 316 |
+
"learning_rate": 9.615384615384615e-07,
|
| 317 |
+
"loss": 0.1659,
|
| 318 |
"step": 1250
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 7.63,
|
| 322 |
+
"learning_rate": 9.602564102564102e-07,
|
| 323 |
+
"loss": 0.1692,
|
| 324 |
"step": 1275
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 7.78,
|
| 328 |
+
"learning_rate": 9.58974358974359e-07,
|
| 329 |
+
"loss": 0.1708,
|
| 330 |
"step": 1300
|
| 331 |
},
|
| 332 |
{
|
| 333 |
"epoch": 7.93,
|
| 334 |
+
"learning_rate": 9.576923076923077e-07,
|
| 335 |
+
"loss": 0.1643,
|
| 336 |
"step": 1325
|
| 337 |
},
|
| 338 |
{
|
| 339 |
"epoch": 8.08,
|
| 340 |
+
"learning_rate": 9.564102564102564e-07,
|
| 341 |
+
"loss": 0.1476,
|
| 342 |
"step": 1350
|
| 343 |
},
|
| 344 |
{
|
| 345 |
"epoch": 8.23,
|
| 346 |
+
"learning_rate": 9.551282051282051e-07,
|
| 347 |
+
"loss": 0.1328,
|
| 348 |
"step": 1375
|
| 349 |
},
|
| 350 |
{
|
| 351 |
"epoch": 8.38,
|
| 352 |
+
"learning_rate": 9.538461538461538e-07,
|
| 353 |
+
"loss": 0.1337,
|
| 354 |
"step": 1400
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 8.53,
|
| 358 |
+
"learning_rate": 9.525641025641025e-07,
|
| 359 |
+
"loss": 0.1484,
|
| 360 |
"step": 1425
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 8.68,
|
| 364 |
+
"learning_rate": 9.512820512820512e-07,
|
| 365 |
+
"loss": 0.1384,
|
| 366 |
"step": 1450
|
| 367 |
},
|
| 368 |
{
|
| 369 |
"epoch": 8.83,
|
| 370 |
+
"learning_rate": 9.499999999999999e-07,
|
| 371 |
+
"loss": 0.1323,
|
| 372 |
"step": 1475
|
| 373 |
},
|
| 374 |
{
|
| 375 |
"epoch": 8.98,
|
| 376 |
+
"learning_rate": 9.487179487179486e-07,
|
| 377 |
+
"loss": 0.1402,
|
| 378 |
"step": 1500
|
| 379 |
},
|
| 380 |
{
|
| 381 |
"epoch": 9.13,
|
| 382 |
+
"learning_rate": 9.474358974358974e-07,
|
| 383 |
+
"loss": 0.1109,
|
| 384 |
"step": 1525
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 9.28,
|
| 388 |
+
"learning_rate": 9.461538461538461e-07,
|
| 389 |
+
"loss": 0.1134,
|
| 390 |
"step": 1550
|
| 391 |
},
|
| 392 |
{
|
| 393 |
"epoch": 9.43,
|
| 394 |
+
"learning_rate": 9.448717948717948e-07,
|
| 395 |
+
"loss": 0.1103,
|
| 396 |
"step": 1575
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 9.58,
|
| 400 |
+
"learning_rate": 9.435897435897435e-07,
|
| 401 |
+
"loss": 0.1112,
|
| 402 |
"step": 1600
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 9.73,
|
| 406 |
+
"learning_rate": 9.423076923076923e-07,
|
| 407 |
+
"loss": 0.1114,
|
| 408 |
"step": 1625
|
| 409 |
},
|
| 410 |
{
|
| 411 |
"epoch": 9.88,
|
| 412 |
+
"learning_rate": 9.41025641025641e-07,
|
| 413 |
+
"loss": 0.1199,
|
| 414 |
"step": 1650
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 10.03,
|
| 418 |
+
"learning_rate": 9.397435897435897e-07,
|
| 419 |
+
"loss": 0.1094,
|
| 420 |
"step": 1675
|
| 421 |
},
|
| 422 |
{
|
| 423 |
"epoch": 10.18,
|
| 424 |
+
"learning_rate": 9.384615384615384e-07,
|
| 425 |
+
"loss": 0.0867,
|
| 426 |
"step": 1700
|
| 427 |
},
|
| 428 |
{
|
| 429 |
"epoch": 10.33,
|
| 430 |
+
"learning_rate": 9.371794871794872e-07,
|
| 431 |
+
"loss": 0.0892,
|
| 432 |
"step": 1725
|
| 433 |
},
|
| 434 |
{
|
| 435 |
"epoch": 10.48,
|
| 436 |
+
"learning_rate": 9.358974358974359e-07,
|
| 437 |
+
"loss": 0.0911,
|
| 438 |
"step": 1750
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 10.63,
|
| 442 |
+
"learning_rate": 9.346153846153846e-07,
|
| 443 |
+
"loss": 0.0944,
|
| 444 |
"step": 1775
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 10.78,
|
| 448 |
+
"learning_rate": 9.333333333333333e-07,
|
| 449 |
+
"loss": 0.091,
|
| 450 |
"step": 1800
|
| 451 |
},
|
| 452 |
{
|
| 453 |
"epoch": 10.93,
|
| 454 |
+
"learning_rate": 9.32051282051282e-07,
|
| 455 |
+
"loss": 0.0931,
|
| 456 |
"step": 1825
|
| 457 |
},
|
| 458 |
{
|
| 459 |
"epoch": 11.08,
|
| 460 |
+
"learning_rate": 9.307692307692308e-07,
|
| 461 |
+
"loss": 0.0798,
|
| 462 |
"step": 1850
|
| 463 |
},
|
| 464 |
{
|
| 465 |
"epoch": 11.23,
|
| 466 |
+
"learning_rate": 9.294871794871795e-07,
|
| 467 |
+
"loss": 0.071,
|
| 468 |
"step": 1875
|
| 469 |
},
|
| 470 |
{
|
| 471 |
"epoch": 11.38,
|
| 472 |
+
"learning_rate": 9.282051282051282e-07,
|
| 473 |
+
"loss": 0.0722,
|
| 474 |
"step": 1900
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 11.53,
|
| 478 |
+
"learning_rate": 9.26923076923077e-07,
|
| 479 |
+
"loss": 0.0762,
|
| 480 |
"step": 1925
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 11.68,
|
| 484 |
+
"learning_rate": 9.256410256410257e-07,
|
| 485 |
+
"loss": 0.0734,
|
| 486 |
"step": 1950
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 11.83,
|
| 490 |
+
"learning_rate": 9.243589743589744e-07,
|
| 491 |
+
"loss": 0.0738,
|
| 492 |
"step": 1975
|
| 493 |
},
|
| 494 |
{
|
| 495 |
"epoch": 11.98,
|
| 496 |
+
"learning_rate": 9.230769230769231e-07,
|
| 497 |
+
"loss": 0.077,
|
| 498 |
"step": 2000
|
| 499 |
},
|
| 500 |
{
|
| 501 |
"epoch": 11.98,
|
| 502 |
+
"eval_loss": 0.4446885585784912,
|
| 503 |
+
"eval_runtime": 586.2687,
|
| 504 |
+
"eval_samples_per_second": 0.715,
|
| 505 |
+
"eval_steps_per_second": 0.09,
|
| 506 |
+
"eval_wer": 43.46845608052362,
|
| 507 |
"step": 2000
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 12.13,
|
| 511 |
+
"learning_rate": 9.217948717948717e-07,
|
| 512 |
+
"loss": 0.0593,
|
| 513 |
"step": 2025
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 12.28,
|
| 517 |
+
"learning_rate": 9.205641025641026e-07,
|
| 518 |
+
"loss": 0.0547,
|
| 519 |
"step": 2050
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 12.43,
|
| 523 |
+
"learning_rate": 9.192820512820513e-07,
|
| 524 |
+
"loss": 0.0598,
|
| 525 |
"step": 2075
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 12.57,
|
| 529 |
+
"learning_rate": 9.18e-07,
|
| 530 |
+
"loss": 0.0564,
|
| 531 |
"step": 2100
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 12.72,
|
| 535 |
+
"learning_rate": 9.167179487179488e-07,
|
| 536 |
+
"loss": 0.0574,
|
| 537 |
"step": 2125
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 12.87,
|
| 541 |
+
"learning_rate": 9.154358974358975e-07,
|
| 542 |
+
"loss": 0.0614,
|
| 543 |
"step": 2150
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 13.02,
|
| 547 |
+
"learning_rate": 9.141538461538462e-07,
|
| 548 |
+
"loss": 0.0575,
|
| 549 |
"step": 2175
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 13.17,
|
| 553 |
+
"learning_rate": 9.128717948717947e-07,
|
| 554 |
+
"loss": 0.0474,
|
| 555 |
"step": 2200
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 13.32,
|
| 559 |
+
"learning_rate": 9.115897435897435e-07,
|
| 560 |
+
"loss": 0.0434,
|
| 561 |
"step": 2225
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 13.47,
|
| 565 |
+
"learning_rate": 9.103076923076922e-07,
|
| 566 |
+
"loss": 0.0464,
|
| 567 |
"step": 2250
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 13.62,
|
| 571 |
+
"learning_rate": 9.090256410256409e-07,
|
| 572 |
+
"loss": 0.0437,
|
| 573 |
"step": 2275
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 13.77,
|
| 577 |
+
"learning_rate": 9.077435897435896e-07,
|
| 578 |
+
"loss": 0.0437,
|
| 579 |
"step": 2300
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 13.92,
|
| 583 |
+
"learning_rate": 9.064615384615384e-07,
|
| 584 |
+
"loss": 0.0457,
|
| 585 |
"step": 2325
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 14.07,
|
| 589 |
+
"learning_rate": 9.051794871794871e-07,
|
| 590 |
+
"loss": 0.042,
|
| 591 |
"step": 2350
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 14.22,
|
| 595 |
+
"learning_rate": 9.038974358974358e-07,
|
| 596 |
+
"loss": 0.0339,
|
| 597 |
"step": 2375
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 14.37,
|
| 601 |
+
"learning_rate": 9.026153846153845e-07,
|
| 602 |
+
"loss": 0.0342,
|
| 603 |
"step": 2400
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 14.52,
|
| 607 |
+
"learning_rate": 9.013333333333333e-07,
|
| 608 |
+
"loss": 0.0323,
|
| 609 |
"step": 2425
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 14.67,
|
| 613 |
+
"learning_rate": 9.00051282051282e-07,
|
| 614 |
+
"loss": 0.0348,
|
| 615 |
"step": 2450
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 14.82,
|
| 619 |
+
"learning_rate": 8.987692307692307e-07,
|
| 620 |
+
"loss": 0.0368,
|
| 621 |
"step": 2475
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 14.97,
|
| 625 |
+
"learning_rate": 8.974871794871794e-07,
|
| 626 |
+
"loss": 0.036,
|
| 627 |
"step": 2500
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 15.12,
|
| 631 |
+
"learning_rate": 8.962051282051282e-07,
|
| 632 |
+
"loss": 0.0252,
|
| 633 |
"step": 2525
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 15.27,
|
| 637 |
+
"learning_rate": 8.949230769230769e-07,
|
| 638 |
+
"loss": 0.0287,
|
| 639 |
"step": 2550
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 15.42,
|
| 643 |
+
"learning_rate": 8.936410256410256e-07,
|
| 644 |
+
"loss": 0.0275,
|
| 645 |
"step": 2575
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 15.57,
|
| 649 |
+
"learning_rate": 8.923589743589743e-07,
|
| 650 |
+
"loss": 0.0263,
|
| 651 |
"step": 2600
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 15.72,
|
| 655 |
+
"learning_rate": 8.91076923076923e-07,
|
| 656 |
+
"loss": 0.0268,
|
| 657 |
"step": 2625
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 15.87,
|
| 661 |
+
"learning_rate": 8.897948717948718e-07,
|
| 662 |
+
"loss": 0.0277,
|
| 663 |
"step": 2650
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 16.02,
|
| 667 |
+
"learning_rate": 8.885128205128205e-07,
|
| 668 |
+
"loss": 0.0263,
|
| 669 |
"step": 2675
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 16.17,
|
| 673 |
+
"learning_rate": 8.872307692307692e-07,
|
| 674 |
+
"loss": 0.0182,
|
| 675 |
"step": 2700
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 16.32,
|
| 679 |
+
"learning_rate": 8.85948717948718e-07,
|
| 680 |
+
"loss": 0.0222,
|
| 681 |
"step": 2725
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 16.47,
|
| 685 |
+
"learning_rate": 8.846666666666667e-07,
|
| 686 |
+
"loss": 0.0205,
|
| 687 |
"step": 2750
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 16.62,
|
| 691 |
+
"learning_rate": 8.833846153846154e-07,
|
| 692 |
+
"loss": 0.0216,
|
| 693 |
"step": 2775
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 16.77,
|
| 697 |
+
"learning_rate": 8.821025641025641e-07,
|
| 698 |
+
"loss": 0.0207,
|
| 699 |
"step": 2800
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 16.92,
|
| 703 |
+
"learning_rate": 8.808205128205127e-07,
|
| 704 |
+
"loss": 0.0223,
|
| 705 |
"step": 2825
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 17.07,
|
| 709 |
+
"learning_rate": 8.795384615384615e-07,
|
| 710 |
+
"loss": 0.0173,
|
| 711 |
"step": 2850
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 17.22,
|
| 715 |
+
"learning_rate": 8.782564102564102e-07,
|
| 716 |
+
"loss": 0.0144,
|
| 717 |
"step": 2875
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 17.37,
|
| 721 |
+
"learning_rate": 8.769743589743589e-07,
|
| 722 |
+
"loss": 0.0159,
|
| 723 |
"step": 2900
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 17.51,
|
| 727 |
+
"learning_rate": 8.756923076923076e-07,
|
| 728 |
+
"loss": 0.0146,
|
| 729 |
"step": 2925
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 17.66,
|
| 733 |
+
"learning_rate": 8.744102564102564e-07,
|
| 734 |
+
"loss": 0.0154,
|
| 735 |
"step": 2950
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 17.81,
|
| 739 |
+
"learning_rate": 8.731282051282051e-07,
|
| 740 |
+
"loss": 0.0152,
|
| 741 |
"step": 2975
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 17.96,
|
| 745 |
+
"learning_rate": 8.718461538461538e-07,
|
| 746 |
+
"loss": 0.0188,
|
| 747 |
"step": 3000
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 17.96,
|
| 751 |
+
"eval_loss": 0.5422356724739075,
|
| 752 |
+
"eval_runtime": 586.4584,
|
| 753 |
+
"eval_samples_per_second": 0.714,
|
| 754 |
+
"eval_steps_per_second": 0.09,
|
| 755 |
+
"eval_wer": 43.32544964523404,
|
| 756 |
"step": 3000
|
| 757 |
},
|
| 758 |
{
|
| 759 |
"epoch": 18.11,
|
| 760 |
+
"learning_rate": 8.705641025641025e-07,
|
| 761 |
+
"loss": 0.0133,
|
| 762 |
"step": 3025
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 18.26,
|
| 766 |
+
"learning_rate": 8.692820512820513e-07,
|
| 767 |
+
"loss": 0.0108,
|
| 768 |
"step": 3050
|
| 769 |
},
|
| 770 |
{
|
| 771 |
"epoch": 18.41,
|
| 772 |
+
"learning_rate": 8.68e-07,
|
| 773 |
+
"loss": 0.0131,
|
| 774 |
"step": 3075
|
| 775 |
},
|
| 776 |
{
|
| 777 |
"epoch": 18.56,
|
| 778 |
+
"learning_rate": 8.667179487179487e-07,
|
| 779 |
+
"loss": 0.0126,
|
| 780 |
"step": 3100
|
| 781 |
},
|
| 782 |
{
|
| 783 |
"epoch": 18.71,
|
| 784 |
+
"learning_rate": 8.654358974358974e-07,
|
| 785 |
+
"loss": 0.0128,
|
| 786 |
"step": 3125
|
| 787 |
},
|
| 788 |
{
|
| 789 |
"epoch": 18.86,
|
| 790 |
+
"learning_rate": 8.641538461538462e-07,
|
| 791 |
+
"loss": 0.012,
|
| 792 |
"step": 3150
|
| 793 |
},
|
| 794 |
{
|
| 795 |
"epoch": 19.01,
|
| 796 |
+
"learning_rate": 8.628717948717949e-07,
|
| 797 |
+
"loss": 0.0128,
|
| 798 |
"step": 3175
|
| 799 |
},
|
| 800 |
{
|
| 801 |
"epoch": 19.16,
|
| 802 |
+
"learning_rate": 8.615897435897436e-07,
|
| 803 |
+
"loss": 0.0098,
|
| 804 |
"step": 3200
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 19.31,
|
| 808 |
+
"learning_rate": 8.603076923076923e-07,
|
| 809 |
+
"loss": 0.0092,
|
| 810 |
"step": 3225
|
| 811 |
},
|
| 812 |
{
|
| 813 |
"epoch": 19.46,
|
| 814 |
+
"learning_rate": 8.590256410256411e-07,
|
| 815 |
+
"loss": 0.0092,
|
| 816 |
"step": 3250
|
| 817 |
},
|
| 818 |
{
|
| 819 |
"epoch": 19.61,
|
| 820 |
+
"learning_rate": 8.577435897435897e-07,
|
| 821 |
+
"loss": 0.009,
|
| 822 |
"step": 3275
|
| 823 |
},
|
| 824 |
{
|
| 825 |
"epoch": 19.76,
|
| 826 |
+
"learning_rate": 8.564615384615384e-07,
|
| 827 |
+
"loss": 0.0092,
|
| 828 |
"step": 3300
|
| 829 |
},
|
| 830 |
{
|
| 831 |
"epoch": 19.91,
|
| 832 |
+
"learning_rate": 8.551794871794871e-07,
|
| 833 |
+
"loss": 0.0091,
|
| 834 |
"step": 3325
|
| 835 |
},
|
| 836 |
{
|
| 837 |
"epoch": 20.06,
|
| 838 |
+
"learning_rate": 8.538974358974359e-07,
|
| 839 |
+
"loss": 0.0097,
|
| 840 |
"step": 3350
|
| 841 |
},
|
| 842 |
{
|
| 843 |
"epoch": 20.21,
|
| 844 |
+
"learning_rate": 8.526153846153846e-07,
|
| 845 |
"loss": 0.0075,
|
| 846 |
"step": 3375
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 20.36,
|
| 850 |
+
"learning_rate": 8.513333333333333e-07,
|
| 851 |
+
"loss": 0.0084,
|
| 852 |
"step": 3400
|
| 853 |
},
|
| 854 |
{
|
| 855 |
"epoch": 20.51,
|
| 856 |
+
"learning_rate": 8.500512820512819e-07,
|
| 857 |
+
"loss": 0.0069,
|
| 858 |
"step": 3425
|
| 859 |
},
|
| 860 |
{
|
| 861 |
"epoch": 20.66,
|
| 862 |
+
"learning_rate": 8.487692307692307e-07,
|
| 863 |
+
"loss": 0.0071,
|
| 864 |
"step": 3450
|
| 865 |
},
|
| 866 |
{
|
| 867 |
"epoch": 20.81,
|
| 868 |
+
"learning_rate": 8.474871794871794e-07,
|
| 869 |
+
"loss": 0.0077,
|
| 870 |
"step": 3475
|
| 871 |
},
|
| 872 |
{
|
| 873 |
"epoch": 20.96,
|
| 874 |
+
"learning_rate": 8.462051282051281e-07,
|
| 875 |
+
"loss": 0.0081,
|
| 876 |
"step": 3500
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 21.11,
|
| 880 |
+
"learning_rate": 8.449230769230768e-07,
|
| 881 |
+
"loss": 0.006,
|
| 882 |
"step": 3525
|
| 883 |
},
|
| 884 |
{
|
| 885 |
"epoch": 21.26,
|
| 886 |
+
"learning_rate": 8.436410256410255e-07,
|
| 887 |
+
"loss": 0.0058,
|
| 888 |
"step": 3550
|
| 889 |
},
|
| 890 |
{
|
| 891 |
"epoch": 21.41,
|
| 892 |
+
"learning_rate": 8.423589743589743e-07,
|
| 893 |
+
"loss": 0.0059,
|
| 894 |
"step": 3575
|
| 895 |
},
|
| 896 |
{
|
| 897 |
"epoch": 21.56,
|
| 898 |
+
"learning_rate": 8.41076923076923e-07,
|
| 899 |
+
"loss": 0.0053,
|
| 900 |
"step": 3600
|
| 901 |
},
|
| 902 |
{
|
| 903 |
"epoch": 21.71,
|
| 904 |
+
"learning_rate": 8.397948717948717e-07,
|
| 905 |
+
"loss": 0.0057,
|
| 906 |
"step": 3625
|
| 907 |
},
|
| 908 |
{
|
| 909 |
"epoch": 21.86,
|
| 910 |
+
"learning_rate": 8.385128205128204e-07,
|
| 911 |
+
"loss": 0.0053,
|
| 912 |
"step": 3650
|
| 913 |
},
|
| 914 |
{
|
| 915 |
"epoch": 22.01,
|
| 916 |
+
"learning_rate": 8.372307692307692e-07,
|
| 917 |
+
"loss": 0.0063,
|
| 918 |
"step": 3675
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 22.16,
|
| 922 |
+
"learning_rate": 8.359487179487179e-07,
|
| 923 |
+
"loss": 0.0045,
|
| 924 |
"step": 3700
|
| 925 |
},
|
| 926 |
{
|
| 927 |
"epoch": 22.31,
|
| 928 |
+
"learning_rate": 8.346666666666666e-07,
|
| 929 |
+
"loss": 0.0043,
|
| 930 |
"step": 3725
|
| 931 |
},
|
| 932 |
{
|
| 933 |
"epoch": 22.46,
|
| 934 |
+
"learning_rate": 8.333846153846153e-07,
|
| 935 |
+
"loss": 0.0041,
|
| 936 |
"step": 3750
|
| 937 |
},
|
| 938 |
{
|
| 939 |
"epoch": 22.6,
|
| 940 |
+
"learning_rate": 8.321025641025641e-07,
|
| 941 |
+
"loss": 0.0042,
|
| 942 |
"step": 3775
|
| 943 |
},
|
| 944 |
{
|
| 945 |
"epoch": 22.75,
|
| 946 |
+
"learning_rate": 8.308205128205128e-07,
|
| 947 |
+
"loss": 0.0044,
|
| 948 |
"step": 3800
|
| 949 |
},
|
| 950 |
{
|
| 951 |
"epoch": 22.9,
|
| 952 |
+
"learning_rate": 8.295384615384615e-07,
|
| 953 |
+
"loss": 0.0046,
|
| 954 |
"step": 3825
|
| 955 |
},
|
| 956 |
{
|
| 957 |
"epoch": 23.05,
|
| 958 |
+
"learning_rate": 8.282564102564102e-07,
|
| 959 |
+
"loss": 0.005,
|
| 960 |
"step": 3850
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 23.2,
|
| 964 |
+
"learning_rate": 8.26974358974359e-07,
|
| 965 |
+
"loss": 0.0036,
|
| 966 |
"step": 3875
|
| 967 |
},
|
| 968 |
{
|
| 969 |
"epoch": 23.35,
|
| 970 |
+
"learning_rate": 8.256923076923077e-07,
|
| 971 |
+
"loss": 0.0039,
|
| 972 |
"step": 3900
|
| 973 |
},
|
| 974 |
{
|
| 975 |
"epoch": 23.5,
|
| 976 |
+
"learning_rate": 8.244102564102564e-07,
|
| 977 |
+
"loss": 0.0033,
|
| 978 |
"step": 3925
|
| 979 |
},
|
| 980 |
{
|
| 981 |
"epoch": 23.65,
|
| 982 |
+
"learning_rate": 8.231282051282051e-07,
|
| 983 |
+
"loss": 0.0035,
|
| 984 |
"step": 3950
|
| 985 |
},
|
| 986 |
{
|
| 987 |
"epoch": 23.8,
|
| 988 |
+
"learning_rate": 8.218461538461539e-07,
|
| 989 |
+
"loss": 0.0041,
|
| 990 |
"step": 3975
|
| 991 |
},
|
| 992 |
{
|
| 993 |
"epoch": 23.95,
|
| 994 |
+
"learning_rate": 8.205641025641026e-07,
|
| 995 |
+
"loss": 0.004,
|
| 996 |
"step": 4000
|
| 997 |
},
|
| 998 |
{
|
| 999 |
"epoch": 23.95,
|
| 1000 |
+
"eval_loss": 0.6144538521766663,
|
| 1001 |
+
"eval_runtime": 590.5826,
|
| 1002 |
+
"eval_samples_per_second": 0.709,
|
| 1003 |
+
"eval_steps_per_second": 0.09,
|
| 1004 |
+
"eval_wer": 42.95693306198779,
|
| 1005 |
"step": 4000
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 24.1,
|
| 1009 |
+
"learning_rate": 8.192820512820512e-07,
|
| 1010 |
+
"loss": 0.0037,
|
| 1011 |
+
"step": 4025
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 24.25,
|
| 1015 |
+
"learning_rate": 8.179999999999999e-07,
|
| 1016 |
+
"loss": 0.0031,
|
| 1017 |
+
"step": 4050
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 24.4,
|
| 1021 |
+
"learning_rate": 8.167179487179487e-07,
|
| 1022 |
+
"loss": 0.0028,
|
| 1023 |
+
"step": 4075
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 24.55,
|
| 1027 |
+
"learning_rate": 8.154358974358974e-07,
|
| 1028 |
+
"loss": 0.0034,
|
| 1029 |
+
"step": 4100
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 24.7,
|
| 1033 |
+
"learning_rate": 8.141538461538461e-07,
|
| 1034 |
+
"loss": 0.0032,
|
| 1035 |
+
"step": 4125
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 24.85,
|
| 1039 |
+
"learning_rate": 8.128717948717948e-07,
|
| 1040 |
+
"loss": 0.0031,
|
| 1041 |
+
"step": 4150
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 25.0,
|
| 1045 |
+
"learning_rate": 8.115897435897436e-07,
|
| 1046 |
+
"loss": 0.0031,
|
| 1047 |
+
"step": 4175
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 25.15,
|
| 1051 |
+
"learning_rate": 8.103076923076923e-07,
|
| 1052 |
+
"loss": 0.0023,
|
| 1053 |
+
"step": 4200
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 25.3,
|
| 1057 |
+
"learning_rate": 8.09025641025641e-07,
|
| 1058 |
+
"loss": 0.0024,
|
| 1059 |
+
"step": 4225
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 25.45,
|
| 1063 |
+
"learning_rate": 8.077435897435897e-07,
|
| 1064 |
+
"loss": 0.0025,
|
| 1065 |
+
"step": 4250
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 25.6,
|
| 1069 |
+
"learning_rate": 8.064615384615385e-07,
|
| 1070 |
+
"loss": 0.0021,
|
| 1071 |
+
"step": 4275
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 25.75,
|
| 1075 |
+
"learning_rate": 8.051794871794872e-07,
|
| 1076 |
+
"loss": 0.0022,
|
| 1077 |
+
"step": 4300
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 25.9,
|
| 1081 |
+
"learning_rate": 8.038974358974359e-07,
|
| 1082 |
+
"loss": 0.0022,
|
| 1083 |
+
"step": 4325
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 26.05,
|
| 1087 |
+
"learning_rate": 8.026153846153845e-07,
|
| 1088 |
+
"loss": 0.0024,
|
| 1089 |
+
"step": 4350
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 26.2,
|
| 1093 |
+
"learning_rate": 8.013333333333333e-07,
|
| 1094 |
+
"loss": 0.0018,
|
| 1095 |
+
"step": 4375
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 26.35,
|
| 1099 |
+
"learning_rate": 8.00051282051282e-07,
|
| 1100 |
+
"loss": 0.0016,
|
| 1101 |
+
"step": 4400
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 26.5,
|
| 1105 |
+
"learning_rate": 7.987692307692307e-07,
|
| 1106 |
+
"loss": 0.0019,
|
| 1107 |
+
"step": 4425
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 26.65,
|
| 1111 |
+
"learning_rate": 7.974871794871794e-07,
|
| 1112 |
+
"loss": 0.0021,
|
| 1113 |
+
"step": 4450
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 26.8,
|
| 1117 |
+
"learning_rate": 7.962051282051282e-07,
|
| 1118 |
+
"loss": 0.0019,
|
| 1119 |
+
"step": 4475
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 26.95,
|
| 1123 |
+
"learning_rate": 7.949230769230769e-07,
|
| 1124 |
+
"loss": 0.0019,
|
| 1125 |
+
"step": 4500
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 27.1,
|
| 1129 |
+
"learning_rate": 7.936410256410256e-07,
|
| 1130 |
+
"loss": 0.0021,
|
| 1131 |
+
"step": 4525
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 27.25,
|
| 1135 |
+
"learning_rate": 7.923589743589743e-07,
|
| 1136 |
+
"loss": 0.0019,
|
| 1137 |
+
"step": 4550
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 27.4,
|
| 1141 |
+
"learning_rate": 7.910769230769231e-07,
|
| 1142 |
+
"loss": 0.002,
|
| 1143 |
+
"step": 4575
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 27.54,
|
| 1147 |
+
"learning_rate": 7.897948717948718e-07,
|
| 1148 |
+
"loss": 0.0024,
|
| 1149 |
+
"step": 4600
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 27.69,
|
| 1153 |
+
"learning_rate": 7.885128205128205e-07,
|
| 1154 |
+
"loss": 0.0028,
|
| 1155 |
+
"step": 4625
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 27.84,
|
| 1159 |
+
"learning_rate": 7.872307692307691e-07,
|
| 1160 |
+
"loss": 0.0033,
|
| 1161 |
+
"step": 4650
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 27.99,
|
| 1165 |
+
"learning_rate": 7.859487179487178e-07,
|
| 1166 |
+
"loss": 0.0036,
|
| 1167 |
+
"step": 4675
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 28.14,
|
| 1171 |
+
"learning_rate": 7.846666666666666e-07,
|
| 1172 |
+
"loss": 0.0035,
|
| 1173 |
+
"step": 4700
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 28.29,
|
| 1177 |
+
"learning_rate": 7.833846153846153e-07,
|
| 1178 |
+
"loss": 0.0031,
|
| 1179 |
+
"step": 4725
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 28.44,
|
| 1183 |
+
"learning_rate": 7.82102564102564e-07,
|
| 1184 |
+
"loss": 0.0032,
|
| 1185 |
+
"step": 4750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 28.59,
|
| 1189 |
+
"learning_rate": 7.808205128205127e-07,
|
| 1190 |
+
"loss": 0.0037,
|
| 1191 |
+
"step": 4775
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 28.74,
|
| 1195 |
+
"learning_rate": 7.795384615384615e-07,
|
| 1196 |
+
"loss": 0.0036,
|
| 1197 |
+
"step": 4800
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 28.89,
|
| 1201 |
+
"learning_rate": 7.782564102564102e-07,
|
| 1202 |
+
"loss": 0.0046,
|
| 1203 |
+
"step": 4825
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 29.04,
|
| 1207 |
+
"learning_rate": 7.769743589743589e-07,
|
| 1208 |
+
"loss": 0.0042,
|
| 1209 |
+
"step": 4850
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 29.19,
|
| 1213 |
+
"learning_rate": 7.756923076923076e-07,
|
| 1214 |
+
"loss": 0.0032,
|
| 1215 |
+
"step": 4875
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 29.34,
|
| 1219 |
+
"learning_rate": 7.744102564102564e-07,
|
| 1220 |
+
"loss": 0.0037,
|
| 1221 |
+
"step": 4900
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 29.49,
|
| 1225 |
+
"learning_rate": 7.731282051282051e-07,
|
| 1226 |
+
"loss": 0.0039,
|
| 1227 |
+
"step": 4925
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 29.64,
|
| 1231 |
+
"learning_rate": 7.718461538461538e-07,
|
| 1232 |
+
"loss": 0.0034,
|
| 1233 |
+
"step": 4950
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 29.79,
|
| 1237 |
+
"learning_rate": 7.705641025641025e-07,
|
| 1238 |
+
"loss": 0.0034,
|
| 1239 |
+
"step": 4975
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 29.94,
|
| 1243 |
+
"learning_rate": 7.692820512820513e-07,
|
| 1244 |
+
"loss": 0.0036,
|
| 1245 |
+
"step": 5000
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 29.94,
|
| 1249 |
+
"eval_loss": 0.6325389742851257,
|
| 1250 |
+
"eval_runtime": 588.3841,
|
| 1251 |
+
"eval_samples_per_second": 0.712,
|
| 1252 |
+
"eval_steps_per_second": 0.09,
|
| 1253 |
+
"eval_wer": 42.08789395522798,
|
| 1254 |
+
"step": 5000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 30.09,
|
| 1258 |
+
"learning_rate": 7.68e-07,
|
| 1259 |
+
"loss": 0.0032,
|
| 1260 |
+
"step": 5025
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 30.24,
|
| 1264 |
+
"learning_rate": 7.667179487179487e-07,
|
| 1265 |
+
"loss": 0.003,
|
| 1266 |
+
"step": 5050
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 30.39,
|
| 1270 |
+
"learning_rate": 7.654358974358974e-07,
|
| 1271 |
+
"loss": 0.0026,
|
| 1272 |
+
"step": 5075
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 30.54,
|
| 1276 |
+
"learning_rate": 7.641538461538462e-07,
|
| 1277 |
+
"loss": 0.0026,
|
| 1278 |
+
"step": 5100
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 30.69,
|
| 1282 |
+
"learning_rate": 7.628717948717949e-07,
|
| 1283 |
+
"loss": 0.0026,
|
| 1284 |
+
"step": 5125
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 30.84,
|
| 1288 |
+
"learning_rate": 7.615897435897436e-07,
|
| 1289 |
+
"loss": 0.0027,
|
| 1290 |
+
"step": 5150
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 30.99,
|
| 1294 |
+
"learning_rate": 7.603076923076923e-07,
|
| 1295 |
+
"loss": 0.0029,
|
| 1296 |
+
"step": 5175
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 31.14,
|
| 1300 |
+
"learning_rate": 7.590256410256411e-07,
|
| 1301 |
+
"loss": 0.0024,
|
| 1302 |
+
"step": 5200
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 31.29,
|
| 1306 |
+
"learning_rate": 7.577435897435898e-07,
|
| 1307 |
+
"loss": 0.0019,
|
| 1308 |
+
"step": 5225
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 31.44,
|
| 1312 |
+
"learning_rate": 7.564615384615384e-07,
|
| 1313 |
+
"loss": 0.0023,
|
| 1314 |
+
"step": 5250
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 31.59,
|
| 1318 |
+
"learning_rate": 7.551794871794871e-07,
|
| 1319 |
+
"loss": 0.0021,
|
| 1320 |
+
"step": 5275
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 31.74,
|
| 1324 |
+
"learning_rate": 7.538974358974359e-07,
|
| 1325 |
+
"loss": 0.002,
|
| 1326 |
+
"step": 5300
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 31.89,
|
| 1330 |
+
"learning_rate": 7.526153846153846e-07,
|
| 1331 |
+
"loss": 0.0021,
|
| 1332 |
+
"step": 5325
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 32.04,
|
| 1336 |
+
"learning_rate": 7.513333333333333e-07,
|
| 1337 |
+
"loss": 0.0024,
|
| 1338 |
+
"step": 5350
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 32.19,
|
| 1342 |
+
"learning_rate": 7.50051282051282e-07,
|
| 1343 |
+
"loss": 0.002,
|
| 1344 |
+
"step": 5375
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 32.34,
|
| 1348 |
+
"learning_rate": 7.487692307692308e-07,
|
| 1349 |
+
"loss": 0.0021,
|
| 1350 |
+
"step": 5400
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 32.49,
|
| 1354 |
+
"learning_rate": 7.475384615384615e-07,
|
| 1355 |
+
"loss": 0.0019,
|
| 1356 |
+
"step": 5425
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 32.63,
|
| 1360 |
+
"learning_rate": 7.462564102564101e-07,
|
| 1361 |
+
"loss": 0.0019,
|
| 1362 |
+
"step": 5450
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 32.78,
|
| 1366 |
+
"learning_rate": 7.449743589743588e-07,
|
| 1367 |
+
"loss": 0.0018,
|
| 1368 |
+
"step": 5475
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 32.93,
|
| 1372 |
+
"learning_rate": 7.436923076923076e-07,
|
| 1373 |
+
"loss": 0.0018,
|
| 1374 |
+
"step": 5500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 33.08,
|
| 1378 |
+
"learning_rate": 7.424102564102563e-07,
|
| 1379 |
+
"loss": 0.0016,
|
| 1380 |
+
"step": 5525
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 33.23,
|
| 1384 |
+
"learning_rate": 7.41128205128205e-07,
|
| 1385 |
+
"loss": 0.0019,
|
| 1386 |
+
"step": 5550
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 33.38,
|
| 1390 |
+
"learning_rate": 7.398461538461537e-07,
|
| 1391 |
+
"loss": 0.0017,
|
| 1392 |
+
"step": 5575
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 33.53,
|
| 1396 |
+
"learning_rate": 7.385641025641025e-07,
|
| 1397 |
+
"loss": 0.0019,
|
| 1398 |
+
"step": 5600
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 33.68,
|
| 1402 |
+
"learning_rate": 7.372820512820512e-07,
|
| 1403 |
+
"loss": 0.0015,
|
| 1404 |
+
"step": 5625
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 33.83,
|
| 1408 |
+
"learning_rate": 7.359999999999999e-07,
|
| 1409 |
+
"loss": 0.0016,
|
| 1410 |
+
"step": 5650
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 33.98,
|
| 1414 |
+
"learning_rate": 7.347179487179486e-07,
|
| 1415 |
+
"loss": 0.0016,
|
| 1416 |
+
"step": 5675
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 34.13,
|
| 1420 |
+
"learning_rate": 7.334358974358974e-07,
|
| 1421 |
+
"loss": 0.0013,
|
| 1422 |
+
"step": 5700
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 34.28,
|
| 1426 |
+
"learning_rate": 7.321538461538461e-07,
|
| 1427 |
+
"loss": 0.0018,
|
| 1428 |
+
"step": 5725
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 34.43,
|
| 1432 |
+
"learning_rate": 7.308717948717948e-07,
|
| 1433 |
+
"loss": 0.0015,
|
| 1434 |
+
"step": 5750
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 34.58,
|
| 1438 |
+
"learning_rate": 7.295897435897435e-07,
|
| 1439 |
+
"loss": 0.0017,
|
| 1440 |
+
"step": 5775
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 34.73,
|
| 1444 |
+
"learning_rate": 7.283076923076923e-07,
|
| 1445 |
+
"loss": 0.0017,
|
| 1446 |
+
"step": 5800
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 34.88,
|
| 1450 |
+
"learning_rate": 7.27025641025641e-07,
|
| 1451 |
+
"loss": 0.0018,
|
| 1452 |
+
"step": 5825
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 35.03,
|
| 1456 |
+
"learning_rate": 7.257435897435897e-07,
|
| 1457 |
+
"loss": 0.002,
|
| 1458 |
+
"step": 5850
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 35.18,
|
| 1462 |
+
"learning_rate": 7.244615384615384e-07,
|
| 1463 |
+
"loss": 0.0016,
|
| 1464 |
+
"step": 5875
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 35.33,
|
| 1468 |
+
"learning_rate": 7.231794871794872e-07,
|
| 1469 |
+
"loss": 0.0018,
|
| 1470 |
+
"step": 5900
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 35.48,
|
| 1474 |
+
"learning_rate": 7.218974358974359e-07,
|
| 1475 |
+
"loss": 0.0016,
|
| 1476 |
+
"step": 5925
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 35.63,
|
| 1480 |
+
"learning_rate": 7.206153846153846e-07,
|
| 1481 |
+
"loss": 0.0015,
|
| 1482 |
+
"step": 5950
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 35.78,
|
| 1486 |
+
"learning_rate": 7.193333333333333e-07,
|
| 1487 |
+
"loss": 0.0014,
|
| 1488 |
+
"step": 5975
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 35.93,
|
| 1492 |
+
"learning_rate": 7.180512820512821e-07,
|
| 1493 |
+
"loss": 0.0018,
|
| 1494 |
+
"step": 6000
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 35.93,
|
| 1498 |
+
"eval_loss": 0.6662390232086182,
|
| 1499 |
+
"eval_runtime": 594.4694,
|
| 1500 |
+
"eval_samples_per_second": 0.705,
|
| 1501 |
+
"eval_steps_per_second": 0.089,
|
| 1502 |
+
"eval_wer": 41.91738628238271,
|
| 1503 |
+
"step": 6000
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 36.08,
|
| 1507 |
+
"learning_rate": 7.167692307692307e-07,
|
| 1508 |
+
"loss": 0.0017,
|
| 1509 |
+
"step": 6025
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 36.23,
|
| 1513 |
+
"learning_rate": 7.154871794871794e-07,
|
| 1514 |
+
"loss": 0.0019,
|
| 1515 |
+
"step": 6050
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 36.38,
|
| 1519 |
+
"learning_rate": 7.142051282051281e-07,
|
| 1520 |
+
"loss": 0.0015,
|
| 1521 |
+
"step": 6075
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 36.53,
|
| 1525 |
+
"learning_rate": 7.129230769230769e-07,
|
| 1526 |
+
"loss": 0.0025,
|
| 1527 |
+
"step": 6100
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 36.68,
|
| 1531 |
+
"learning_rate": 7.116410256410256e-07,
|
| 1532 |
+
"loss": 0.0017,
|
| 1533 |
+
"step": 6125
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 36.83,
|
| 1537 |
+
"learning_rate": 7.103589743589743e-07,
|
| 1538 |
+
"loss": 0.0014,
|
| 1539 |
+
"step": 6150
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 36.98,
|
| 1543 |
+
"learning_rate": 7.09076923076923e-07,
|
| 1544 |
+
"loss": 0.0015,
|
| 1545 |
+
"step": 6175
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 37.13,
|
| 1549 |
+
"learning_rate": 7.077948717948718e-07,
|
| 1550 |
+
"loss": 0.0011,
|
| 1551 |
+
"step": 6200
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 37.28,
|
| 1555 |
+
"learning_rate": 7.065128205128205e-07,
|
| 1556 |
+
"loss": 0.0012,
|
| 1557 |
+
"step": 6225
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 37.43,
|
| 1561 |
+
"learning_rate": 7.052307692307692e-07,
|
| 1562 |
+
"loss": 0.0017,
|
| 1563 |
+
"step": 6250
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 37.57,
|
| 1567 |
+
"learning_rate": 7.039487179487179e-07,
|
| 1568 |
+
"loss": 0.0015,
|
| 1569 |
+
"step": 6275
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 37.72,
|
| 1573 |
+
"learning_rate": 7.026666666666667e-07,
|
| 1574 |
+
"loss": 0.0016,
|
| 1575 |
+
"step": 6300
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 37.87,
|
| 1579 |
+
"learning_rate": 7.013846153846154e-07,
|
| 1580 |
+
"loss": 0.0018,
|
| 1581 |
+
"step": 6325
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 38.02,
|
| 1585 |
+
"learning_rate": 7.001025641025641e-07,
|
| 1586 |
+
"loss": 0.0013,
|
| 1587 |
+
"step": 6350
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 38.17,
|
| 1591 |
+
"learning_rate": 6.988205128205128e-07,
|
| 1592 |
+
"loss": 0.0012,
|
| 1593 |
+
"step": 6375
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 38.32,
|
| 1597 |
+
"learning_rate": 6.975384615384616e-07,
|
| 1598 |
+
"loss": 0.0013,
|
| 1599 |
+
"step": 6400
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 38.47,
|
| 1603 |
+
"learning_rate": 6.962564102564103e-07,
|
| 1604 |
+
"loss": 0.0009,
|
| 1605 |
+
"step": 6425
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 38.62,
|
| 1609 |
+
"learning_rate": 6.94974358974359e-07,
|
| 1610 |
+
"loss": 0.0012,
|
| 1611 |
+
"step": 6450
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 38.77,
|
| 1615 |
+
"learning_rate": 6.936923076923077e-07,
|
| 1616 |
+
"loss": 0.001,
|
| 1617 |
+
"step": 6475
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 38.92,
|
| 1621 |
+
"learning_rate": 6.924102564102565e-07,
|
| 1622 |
+
"loss": 0.001,
|
| 1623 |
+
"step": 6500
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 39.07,
|
| 1627 |
+
"learning_rate": 6.911282051282052e-07,
|
| 1628 |
+
"loss": 0.001,
|
| 1629 |
+
"step": 6525
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 39.22,
|
| 1633 |
+
"learning_rate": 6.898461538461539e-07,
|
| 1634 |
+
"loss": 0.0012,
|
| 1635 |
+
"step": 6550
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 39.37,
|
| 1639 |
+
"learning_rate": 6.885641025641025e-07,
|
| 1640 |
+
"loss": 0.0012,
|
| 1641 |
+
"step": 6575
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 39.52,
|
| 1645 |
+
"learning_rate": 6.872820512820513e-07,
|
| 1646 |
+
"loss": 0.0014,
|
| 1647 |
+
"step": 6600
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 39.67,
|
| 1651 |
+
"learning_rate": 6.86e-07,
|
| 1652 |
+
"loss": 0.0018,
|
| 1653 |
+
"step": 6625
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 39.82,
|
| 1657 |
+
"learning_rate": 6.847179487179486e-07,
|
| 1658 |
+
"loss": 0.0014,
|
| 1659 |
+
"step": 6650
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 39.97,
|
| 1663 |
+
"learning_rate": 6.834358974358973e-07,
|
| 1664 |
+
"loss": 0.0017,
|
| 1665 |
+
"step": 6675
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 40.12,
|
| 1669 |
+
"learning_rate": 6.82153846153846e-07,
|
| 1670 |
+
"loss": 0.0015,
|
| 1671 |
+
"step": 6700
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 40.27,
|
| 1675 |
+
"learning_rate": 6.808717948717948e-07,
|
| 1676 |
+
"loss": 0.0019,
|
| 1677 |
+
"step": 6725
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 40.42,
|
| 1681 |
+
"learning_rate": 6.795897435897435e-07,
|
| 1682 |
+
"loss": 0.0014,
|
| 1683 |
+
"step": 6750
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 40.57,
|
| 1687 |
+
"learning_rate": 6.783076923076922e-07,
|
| 1688 |
+
"loss": 0.0019,
|
| 1689 |
+
"step": 6775
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 40.72,
|
| 1693 |
+
"learning_rate": 6.770256410256409e-07,
|
| 1694 |
+
"loss": 0.0016,
|
| 1695 |
+
"step": 6800
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 40.87,
|
| 1699 |
+
"learning_rate": 6.757435897435897e-07,
|
| 1700 |
+
"loss": 0.0015,
|
| 1701 |
+
"step": 6825
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 41.02,
|
| 1705 |
+
"learning_rate": 6.744615384615384e-07,
|
| 1706 |
+
"loss": 0.0014,
|
| 1707 |
+
"step": 6850
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 41.17,
|
| 1711 |
+
"learning_rate": 6.731794871794871e-07,
|
| 1712 |
+
"loss": 0.0012,
|
| 1713 |
+
"step": 6875
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 41.32,
|
| 1717 |
+
"learning_rate": 6.718974358974358e-07,
|
| 1718 |
+
"loss": 0.0013,
|
| 1719 |
+
"step": 6900
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 41.47,
|
| 1723 |
+
"learning_rate": 6.706153846153846e-07,
|
| 1724 |
+
"loss": 0.0013,
|
| 1725 |
+
"step": 6925
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 41.62,
|
| 1729 |
+
"learning_rate": 6.693333333333333e-07,
|
| 1730 |
+
"loss": 0.0012,
|
| 1731 |
+
"step": 6950
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 41.77,
|
| 1735 |
+
"learning_rate": 6.68051282051282e-07,
|
| 1736 |
+
"loss": 0.0012,
|
| 1737 |
+
"step": 6975
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 41.92,
|
| 1741 |
+
"learning_rate": 6.667692307692307e-07,
|
| 1742 |
+
"loss": 0.0013,
|
| 1743 |
+
"step": 7000
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 41.92,
|
| 1747 |
+
"eval_loss": 0.6843737363815308,
|
| 1748 |
+
"eval_runtime": 592.3336,
|
| 1749 |
+
"eval_samples_per_second": 0.707,
|
| 1750 |
+
"eval_steps_per_second": 0.089,
|
| 1751 |
+
"eval_wer": 42.35190583576261,
|
| 1752 |
+
"step": 7000
|
| 1753 |
}
|
| 1754 |
],
|
| 1755 |
"logging_steps": 25,
|
| 1756 |
+
"max_steps": 20000,
|
| 1757 |
"num_input_tokens_seen": 0,
|
| 1758 |
+
"num_train_epochs": 120,
|
| 1759 |
"save_steps": 1000,
|
| 1760 |
+
"total_flos": 3.801019180430131e+20,
|
| 1761 |
"train_batch_size": 16,
|
| 1762 |
"trial_name": null,
|
| 1763 |
"trial_params": null
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5112
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0359fd8c0bca8ce29466944ad8b07dcfba0ff1a4788df0f13fa23689a41235a
|
| 3 |
size 5112
|