Commit
·
07c24cc
1
Parent(s):
75539f4
End of training
Browse files- all_results.json +10 -10
- eval_results.json +6 -6
- pytorch_model.bin +1 -1
- train_results.json +5 -5
- trainer_state.json +1008 -273
all_results.json
CHANGED
|
@@ -1,14 +1,14 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 3038,
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
-
"train_loss": 1.
|
| 10 |
-
"train_runtime":
|
| 11 |
"train_samples": 6769,
|
| 12 |
-
"train_samples_per_second":
|
| 13 |
-
"train_steps_per_second": 0.
|
| 14 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.2759484350681305,
|
| 4 |
+
"eval_runtime": 93.8442,
|
| 5 |
"eval_samples": 3038,
|
| 6 |
+
"eval_samples_per_second": 32.373,
|
| 7 |
+
"eval_steps_per_second": 1.012,
|
| 8 |
+
"eval_wer": 0.3256073722423904,
|
| 9 |
+
"train_loss": 1.235663890838623,
|
| 10 |
+
"train_runtime": 30132.3883,
|
| 11 |
"train_samples": 6769,
|
| 12 |
+
"train_samples_per_second": 22.464,
|
| 13 |
+
"train_steps_per_second": 0.704
|
| 14 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 3038,
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.2759484350681305,
|
| 4 |
+
"eval_runtime": 93.8442,
|
| 5 |
"eval_samples": 3038,
|
| 6 |
+
"eval_samples_per_second": 32.373,
|
| 7 |
+
"eval_steps_per_second": 1.012,
|
| 8 |
+
"eval_wer": 0.3256073722423904
|
| 9 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1262157361
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e6baa59b6dc263be91cec1bb3d88fc58c0fdba4311b736a9b4b0f11fb378cc36
|
| 3 |
size 1262157361
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 1.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 6769,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"train_loss": 1.235663890838623,
|
| 4 |
+
"train_runtime": 30132.3883,
|
| 5 |
"train_samples": 6769,
|
| 6 |
+
"train_samples_per_second": 22.464,
|
| 7 |
+
"train_steps_per_second": 0.704
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,751 +1,1486 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.47,
|
| 12 |
-
"learning_rate":
|
| 13 |
-
"loss":
|
| 14 |
"step": 100
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.94,
|
| 18 |
-
"learning_rate":
|
| 19 |
-
"loss":
|
| 20 |
"step": 200
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 1.42,
|
| 24 |
-
"learning_rate":
|
| 25 |
-
"loss":
|
| 26 |
"step": 300
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 1.89,
|
| 30 |
-
"learning_rate":
|
| 31 |
-
"loss": 4.
|
| 32 |
"step": 400
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 2.36,
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss":
|
| 38 |
"step": 500
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 2.83,
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss": 3.
|
| 44 |
"step": 600
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 3.3,
|
| 48 |
-
"learning_rate":
|
| 49 |
-
"loss": 3.
|
| 50 |
"step": 700
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 3.77,
|
| 54 |
-
"learning_rate":
|
| 55 |
-
"loss":
|
| 56 |
"step": 800
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 4.25,
|
| 60 |
-
"learning_rate":
|
| 61 |
-
"loss":
|
| 62 |
"step": 900
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 4.72,
|
| 66 |
-
"learning_rate":
|
| 67 |
-
"loss":
|
| 68 |
"step": 1000
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 4.72,
|
| 72 |
-
"eval_loss":
|
| 73 |
-
"eval_runtime":
|
| 74 |
-
"eval_samples_per_second":
|
| 75 |
-
"eval_steps_per_second": 0.
|
| 76 |
"eval_wer": 1.0,
|
| 77 |
"step": 1000
|
| 78 |
},
|
| 79 |
{
|
| 80 |
"epoch": 5.19,
|
| 81 |
-
"learning_rate":
|
| 82 |
-
"loss": 2.
|
| 83 |
"step": 1100
|
| 84 |
},
|
| 85 |
{
|
| 86 |
"epoch": 5.66,
|
| 87 |
-
"learning_rate":
|
| 88 |
-
"loss": 2.
|
| 89 |
"step": 1200
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 6.13,
|
| 93 |
-
"learning_rate":
|
| 94 |
-
"loss": 2.
|
| 95 |
"step": 1300
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 6.6,
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"loss": 2.
|
| 101 |
"step": 1400
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 7.08,
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss":
|
| 107 |
"step": 1500
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 7.55,
|
| 111 |
-
"learning_rate":
|
| 112 |
-
"loss":
|
| 113 |
"step": 1600
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 8.02,
|
| 117 |
-
"learning_rate":
|
| 118 |
-
"loss":
|
| 119 |
"step": 1700
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 8.49,
|
| 123 |
-
"learning_rate":
|
| 124 |
-
"loss":
|
| 125 |
"step": 1800
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 8.96,
|
| 129 |
-
"learning_rate":
|
| 130 |
-
"loss":
|
| 131 |
"step": 1900
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 9.43,
|
| 135 |
-
"learning_rate":
|
| 136 |
-
"loss": 1.
|
| 137 |
"step": 2000
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 9.43,
|
| 141 |
-
"eval_loss": 0.
|
| 142 |
-
"eval_runtime":
|
| 143 |
-
"eval_samples_per_second":
|
| 144 |
-
"eval_steps_per_second": 0.
|
| 145 |
-
"eval_wer": 0.
|
| 146 |
"step": 2000
|
| 147 |
},
|
| 148 |
{
|
| 149 |
"epoch": 9.91,
|
| 150 |
-
"learning_rate":
|
| 151 |
-
"loss": 1.
|
| 152 |
"step": 2100
|
| 153 |
},
|
| 154 |
{
|
| 155 |
"epoch": 10.38,
|
| 156 |
-
"learning_rate":
|
| 157 |
-
"loss": 1.
|
| 158 |
"step": 2200
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 10.85,
|
| 162 |
-
"learning_rate":
|
| 163 |
-
"loss": 1.
|
| 164 |
"step": 2300
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 11.32,
|
| 168 |
-
"learning_rate":
|
| 169 |
-
"loss": 1.
|
| 170 |
"step": 2400
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 11.79,
|
| 174 |
-
"learning_rate":
|
| 175 |
-
"loss": 1.
|
| 176 |
"step": 2500
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 12.26,
|
| 180 |
-
"learning_rate":
|
| 181 |
-
"loss": 1.
|
| 182 |
"step": 2600
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 12.74,
|
| 186 |
-
"learning_rate":
|
| 187 |
-
"loss": 1.
|
| 188 |
"step": 2700
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 13.21,
|
| 192 |
-
"learning_rate":
|
| 193 |
-
"loss": 1.
|
| 194 |
"step": 2800
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 13.68,
|
| 198 |
-
"learning_rate":
|
| 199 |
-
"loss": 1.
|
| 200 |
"step": 2900
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 14.15,
|
| 204 |
-
"learning_rate":
|
| 205 |
-
"loss": 1.
|
| 206 |
"step": 3000
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 14.15,
|
| 210 |
-
"eval_loss": 0.
|
| 211 |
-
"eval_runtime":
|
| 212 |
-
"eval_samples_per_second":
|
| 213 |
-
"eval_steps_per_second": 0.
|
| 214 |
-
"eval_wer": 0.
|
| 215 |
"step": 3000
|
| 216 |
},
|
| 217 |
{
|
| 218 |
"epoch": 14.62,
|
| 219 |
-
"learning_rate":
|
| 220 |
-
"loss": 1.
|
| 221 |
"step": 3100
|
| 222 |
},
|
| 223 |
{
|
| 224 |
"epoch": 15.09,
|
| 225 |
-
"learning_rate":
|
| 226 |
-
"loss": 1.
|
| 227 |
"step": 3200
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 15.57,
|
| 231 |
-
"learning_rate": 5.
|
| 232 |
-
"loss": 1.
|
| 233 |
"step": 3300
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 16.04,
|
| 237 |
-
"learning_rate": 5.
|
| 238 |
-
"loss": 1.
|
| 239 |
"step": 3400
|
| 240 |
},
|
| 241 |
{
|
| 242 |
"epoch": 16.51,
|
| 243 |
-
"learning_rate":
|
| 244 |
-
"loss": 1.
|
| 245 |
"step": 3500
|
| 246 |
},
|
| 247 |
{
|
| 248 |
"epoch": 16.98,
|
| 249 |
-
"learning_rate":
|
| 250 |
-
"loss": 1.
|
| 251 |
"step": 3600
|
| 252 |
},
|
| 253 |
{
|
| 254 |
"epoch": 17.45,
|
| 255 |
-
"learning_rate":
|
| 256 |
-
"loss": 1.
|
| 257 |
"step": 3700
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 17.92,
|
| 261 |
-
"learning_rate":
|
| 262 |
-
"loss": 1.
|
| 263 |
"step": 3800
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 18.4,
|
| 267 |
-
"learning_rate":
|
| 268 |
-
"loss": 1.
|
| 269 |
"step": 3900
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 18.87,
|
| 273 |
-
"learning_rate":
|
| 274 |
-
"loss": 1.
|
| 275 |
"step": 4000
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 18.87,
|
| 279 |
-
"eval_loss": 0.
|
| 280 |
-
"eval_runtime":
|
| 281 |
-
"eval_samples_per_second":
|
| 282 |
-
"eval_steps_per_second": 0.
|
| 283 |
-
"eval_wer": 0.
|
| 284 |
"step": 4000
|
| 285 |
},
|
| 286 |
{
|
| 287 |
"epoch": 19.34,
|
| 288 |
-
"learning_rate":
|
| 289 |
-
"loss": 1.
|
| 290 |
"step": 4100
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 19.81,
|
| 294 |
-
"learning_rate":
|
| 295 |
-
"loss": 1.
|
| 296 |
"step": 4200
|
| 297 |
},
|
| 298 |
{
|
| 299 |
"epoch": 20.28,
|
| 300 |
-
"learning_rate":
|
| 301 |
-
"loss": 1.
|
| 302 |
"step": 4300
|
| 303 |
},
|
| 304 |
{
|
| 305 |
"epoch": 20.75,
|
| 306 |
-
"learning_rate":
|
| 307 |
-
"loss": 1.
|
| 308 |
"step": 4400
|
| 309 |
},
|
| 310 |
{
|
| 311 |
"epoch": 21.23,
|
| 312 |
-
"learning_rate":
|
| 313 |
-
"loss": 1.
|
| 314 |
"step": 4500
|
| 315 |
},
|
| 316 |
{
|
| 317 |
"epoch": 21.7,
|
| 318 |
-
"learning_rate":
|
| 319 |
-
"loss": 1.
|
| 320 |
"step": 4600
|
| 321 |
},
|
| 322 |
{
|
| 323 |
"epoch": 22.17,
|
| 324 |
-
"learning_rate":
|
| 325 |
-
"loss": 1.
|
| 326 |
"step": 4700
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 22.64,
|
| 330 |
-
"learning_rate":
|
| 331 |
-
"loss": 1.
|
| 332 |
"step": 4800
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 23.11,
|
| 336 |
-
"learning_rate":
|
| 337 |
-
"loss": 1.
|
| 338 |
"step": 4900
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 23.58,
|
| 342 |
-
"learning_rate":
|
| 343 |
-
"loss": 1.
|
| 344 |
"step": 5000
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 23.58,
|
| 348 |
-
"eval_loss": 0.
|
| 349 |
-
"eval_runtime":
|
| 350 |
-
"eval_samples_per_second":
|
| 351 |
-
"eval_steps_per_second": 0.
|
| 352 |
-
"eval_wer": 0.
|
| 353 |
"step": 5000
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 24.06,
|
| 357 |
-
"learning_rate":
|
| 358 |
-
"loss": 1.
|
| 359 |
"step": 5100
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"epoch": 24.53,
|
| 363 |
-
"learning_rate":
|
| 364 |
-
"loss": 1.
|
| 365 |
"step": 5200
|
| 366 |
},
|
| 367 |
{
|
| 368 |
"epoch": 25.0,
|
| 369 |
-
"learning_rate":
|
| 370 |
-
"loss": 1.
|
| 371 |
"step": 5300
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 25.47,
|
| 375 |
-
"learning_rate":
|
| 376 |
-
"loss": 1.
|
| 377 |
"step": 5400
|
| 378 |
},
|
| 379 |
{
|
| 380 |
"epoch": 25.94,
|
| 381 |
-
"learning_rate":
|
| 382 |
-
"loss": 1.
|
| 383 |
"step": 5500
|
| 384 |
},
|
| 385 |
{
|
| 386 |
"epoch": 26.42,
|
| 387 |
-
"learning_rate":
|
| 388 |
-
"loss": 1.
|
| 389 |
"step": 5600
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 26.89,
|
| 393 |
-
"learning_rate":
|
| 394 |
-
"loss": 1.
|
| 395 |
"step": 5700
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 27.36,
|
| 399 |
-
"learning_rate":
|
| 400 |
-
"loss": 1.
|
| 401 |
"step": 5800
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 27.83,
|
| 405 |
-
"learning_rate":
|
| 406 |
-
"loss": 1.
|
| 407 |
"step": 5900
|
| 408 |
},
|
| 409 |
{
|
| 410 |
"epoch": 28.3,
|
| 411 |
-
"learning_rate":
|
| 412 |
-
"loss": 1.
|
| 413 |
"step": 6000
|
| 414 |
},
|
| 415 |
{
|
| 416 |
"epoch": 28.3,
|
| 417 |
-
"eval_loss": 0.
|
| 418 |
-
"eval_runtime":
|
| 419 |
-
"eval_samples_per_second":
|
| 420 |
-
"eval_steps_per_second":
|
| 421 |
-
"eval_wer": 0.
|
| 422 |
"step": 6000
|
| 423 |
},
|
| 424 |
{
|
| 425 |
"epoch": 28.77,
|
| 426 |
-
"learning_rate":
|
| 427 |
-
"loss": 1.
|
| 428 |
"step": 6100
|
| 429 |
},
|
| 430 |
{
|
| 431 |
"epoch": 29.25,
|
| 432 |
-
"learning_rate":
|
| 433 |
-
"loss":
|
| 434 |
"step": 6200
|
| 435 |
},
|
| 436 |
{
|
| 437 |
"epoch": 29.72,
|
| 438 |
-
"learning_rate":
|
| 439 |
-
"loss":
|
| 440 |
"step": 6300
|
| 441 |
},
|
| 442 |
{
|
| 443 |
"epoch": 30.19,
|
| 444 |
-
"learning_rate":
|
| 445 |
-
"loss":
|
| 446 |
"step": 6400
|
| 447 |
},
|
| 448 |
{
|
| 449 |
"epoch": 30.66,
|
| 450 |
-
"learning_rate":
|
| 451 |
-
"loss":
|
| 452 |
"step": 6500
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 31.13,
|
| 456 |
-
"learning_rate":
|
| 457 |
-
"loss":
|
| 458 |
"step": 6600
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 31.6,
|
| 462 |
-
"learning_rate":
|
| 463 |
-
"loss":
|
| 464 |
"step": 6700
|
| 465 |
},
|
| 466 |
{
|
| 467 |
"epoch": 32.08,
|
| 468 |
-
"learning_rate":
|
| 469 |
-
"loss":
|
| 470 |
"step": 6800
|
| 471 |
},
|
| 472 |
{
|
| 473 |
"epoch": 32.55,
|
| 474 |
-
"learning_rate":
|
| 475 |
-
"loss":
|
| 476 |
"step": 6900
|
| 477 |
},
|
| 478 |
{
|
| 479 |
"epoch": 33.02,
|
| 480 |
-
"learning_rate":
|
| 481 |
-
"loss":
|
| 482 |
"step": 7000
|
| 483 |
},
|
| 484 |
{
|
| 485 |
"epoch": 33.02,
|
| 486 |
-
"eval_loss": 0.
|
| 487 |
-
"eval_runtime":
|
| 488 |
-
"eval_samples_per_second":
|
| 489 |
-
"eval_steps_per_second":
|
| 490 |
-
"eval_wer": 0.
|
| 491 |
"step": 7000
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 33.49,
|
| 495 |
-
"learning_rate":
|
| 496 |
-
"loss":
|
| 497 |
"step": 7100
|
| 498 |
},
|
| 499 |
{
|
| 500 |
"epoch": 33.96,
|
| 501 |
-
"learning_rate":
|
| 502 |
-
"loss":
|
| 503 |
"step": 7200
|
| 504 |
},
|
| 505 |
{
|
| 506 |
"epoch": 34.43,
|
| 507 |
-
"learning_rate":
|
| 508 |
-
"loss":
|
| 509 |
"step": 7300
|
| 510 |
},
|
| 511 |
{
|
| 512 |
"epoch": 34.91,
|
| 513 |
-
"learning_rate":
|
| 514 |
-
"loss":
|
| 515 |
"step": 7400
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 35.38,
|
| 519 |
-
"learning_rate":
|
| 520 |
-
"loss":
|
| 521 |
"step": 7500
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 35.85,
|
| 525 |
-
"learning_rate":
|
| 526 |
-
"loss":
|
| 527 |
"step": 7600
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"epoch": 36.32,
|
| 531 |
-
"learning_rate":
|
| 532 |
-
"loss":
|
| 533 |
"step": 7700
|
| 534 |
},
|
| 535 |
{
|
| 536 |
"epoch": 36.79,
|
| 537 |
-
"learning_rate":
|
| 538 |
-
"loss":
|
| 539 |
"step": 7800
|
| 540 |
},
|
| 541 |
{
|
| 542 |
"epoch": 37.26,
|
| 543 |
-
"learning_rate":
|
| 544 |
-
"loss": 0.
|
| 545 |
"step": 7900
|
| 546 |
},
|
| 547 |
{
|
| 548 |
"epoch": 37.74,
|
| 549 |
-
"learning_rate":
|
| 550 |
-
"loss": 0.
|
| 551 |
"step": 8000
|
| 552 |
},
|
| 553 |
{
|
| 554 |
"epoch": 37.74,
|
| 555 |
-
"eval_loss": 0.
|
| 556 |
-
"eval_runtime":
|
| 557 |
-
"eval_samples_per_second":
|
| 558 |
-
"eval_steps_per_second":
|
| 559 |
-
"eval_wer": 0.
|
| 560 |
"step": 8000
|
| 561 |
},
|
| 562 |
{
|
| 563 |
"epoch": 38.21,
|
| 564 |
-
"learning_rate":
|
| 565 |
-
"loss": 0.
|
| 566 |
"step": 8100
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 38.68,
|
| 570 |
-
"learning_rate":
|
| 571 |
-
"loss": 0.
|
| 572 |
"step": 8200
|
| 573 |
},
|
| 574 |
{
|
| 575 |
"epoch": 39.15,
|
| 576 |
-
"learning_rate":
|
| 577 |
-
"loss":
|
| 578 |
"step": 8300
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 39.62,
|
| 582 |
-
"learning_rate":
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 8400
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 40.09,
|
| 588 |
-
"learning_rate":
|
| 589 |
-
"loss": 0.
|
| 590 |
"step": 8500
|
| 591 |
},
|
| 592 |
{
|
| 593 |
"epoch": 40.57,
|
| 594 |
-
"learning_rate":
|
| 595 |
-
"loss": 0.
|
| 596 |
"step": 8600
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 41.04,
|
| 600 |
-
"learning_rate":
|
| 601 |
-
"loss": 0.
|
| 602 |
"step": 8700
|
| 603 |
},
|
| 604 |
{
|
| 605 |
"epoch": 41.51,
|
| 606 |
-
"learning_rate":
|
| 607 |
-
"loss": 0.
|
| 608 |
"step": 8800
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 41.98,
|
| 612 |
-
"learning_rate":
|
| 613 |
-
"loss": 0.
|
| 614 |
"step": 8900
|
| 615 |
},
|
| 616 |
{
|
| 617 |
"epoch": 42.45,
|
| 618 |
-
"learning_rate":
|
| 619 |
-
"loss": 0.
|
| 620 |
"step": 9000
|
| 621 |
},
|
| 622 |
{
|
| 623 |
"epoch": 42.45,
|
| 624 |
-
"eval_loss": 0.
|
| 625 |
-
"eval_runtime":
|
| 626 |
-
"eval_samples_per_second":
|
| 627 |
-
"eval_steps_per_second":
|
| 628 |
-
"eval_wer": 0.
|
| 629 |
"step": 9000
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 42.92,
|
| 633 |
-
"learning_rate":
|
| 634 |
-
"loss": 0.
|
| 635 |
"step": 9100
|
| 636 |
},
|
| 637 |
{
|
| 638 |
"epoch": 43.4,
|
| 639 |
-
"learning_rate":
|
| 640 |
-
"loss": 0.
|
| 641 |
"step": 9200
|
| 642 |
},
|
| 643 |
{
|
| 644 |
"epoch": 43.87,
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 9300
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 44.34,
|
| 651 |
-
"learning_rate":
|
| 652 |
-
"loss": 0.
|
| 653 |
"step": 9400
|
| 654 |
},
|
| 655 |
{
|
| 656 |
"epoch": 44.81,
|
| 657 |
-
"learning_rate":
|
| 658 |
-
"loss": 0.
|
| 659 |
"step": 9500
|
| 660 |
},
|
| 661 |
{
|
| 662 |
"epoch": 45.28,
|
| 663 |
-
"learning_rate":
|
| 664 |
-
"loss": 0.
|
| 665 |
"step": 9600
|
| 666 |
},
|
| 667 |
{
|
| 668 |
"epoch": 45.75,
|
| 669 |
-
"learning_rate":
|
| 670 |
-
"loss": 0.
|
| 671 |
"step": 9700
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 46.23,
|
| 675 |
-
"learning_rate":
|
| 676 |
-
"loss": 0.
|
| 677 |
"step": 9800
|
| 678 |
},
|
| 679 |
{
|
| 680 |
"epoch": 46.7,
|
| 681 |
-
"learning_rate":
|
| 682 |
-
"loss": 0.
|
| 683 |
"step": 9900
|
| 684 |
},
|
| 685 |
{
|
| 686 |
"epoch": 47.17,
|
| 687 |
-
"learning_rate": 4.
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 10000
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 47.17,
|
| 693 |
-
"eval_loss": 0.
|
| 694 |
-
"eval_runtime":
|
| 695 |
-
"eval_samples_per_second":
|
| 696 |
-
"eval_steps_per_second":
|
| 697 |
-
"eval_wer": 0.
|
| 698 |
"step": 10000
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 47.64,
|
| 702 |
-
"learning_rate": 4.
|
| 703 |
-
"loss": 0.
|
| 704 |
"step": 10100
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 48.11,
|
| 708 |
-
"learning_rate":
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 10200
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 48.58,
|
| 714 |
-
"learning_rate":
|
| 715 |
-
"loss": 0.
|
| 716 |
"step": 10300
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 49.06,
|
| 720 |
-
"learning_rate":
|
| 721 |
-
"loss": 0.
|
| 722 |
"step": 10400
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 49.53,
|
| 726 |
-
"learning_rate":
|
| 727 |
-
"loss": 0.
|
| 728 |
"step": 10500
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 50.0,
|
| 732 |
-
"learning_rate":
|
| 733 |
-
"loss": 0.
|
| 734 |
"step": 10600
|
| 735 |
},
|
| 736 |
{
|
| 737 |
-
"epoch": 50.
|
| 738 |
-
"
|
| 739 |
-
"
|
| 740 |
-
"
|
| 741 |
-
|
| 742 |
-
|
| 743 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 744 |
}
|
| 745 |
],
|
| 746 |
-
"max_steps":
|
| 747 |
-
"num_train_epochs":
|
| 748 |
-
"total_flos":
|
| 749 |
"trial_name": null,
|
| 750 |
"trial_params": null
|
| 751 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"global_step": 21200,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.47,
|
| 12 |
+
"learning_rate": 1.7149999999999999e-06,
|
| 13 |
+
"loss": 12.1703,
|
| 14 |
"step": 100
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.94,
|
| 18 |
+
"learning_rate": 3.465e-06,
|
| 19 |
+
"loss": 9.0128,
|
| 20 |
"step": 200
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 1.42,
|
| 24 |
+
"learning_rate": 5.214999999999999e-06,
|
| 25 |
+
"loss": 5.5165,
|
| 26 |
"step": 300
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 1.89,
|
| 30 |
+
"learning_rate": 6.964999999999999e-06,
|
| 31 |
+
"loss": 4.6455,
|
| 32 |
"step": 400
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 2.36,
|
| 36 |
+
"learning_rate": 8.714999999999999e-06,
|
| 37 |
+
"loss": 4.1839,
|
| 38 |
"step": 500
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 2.83,
|
| 42 |
+
"learning_rate": 1.0464999999999999e-05,
|
| 43 |
+
"loss": 3.8664,
|
| 44 |
"step": 600
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 3.3,
|
| 48 |
+
"learning_rate": 1.2214999999999997e-05,
|
| 49 |
+
"loss": 3.5691,
|
| 50 |
"step": 700
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 3.77,
|
| 54 |
+
"learning_rate": 1.3965e-05,
|
| 55 |
+
"loss": 3.3224,
|
| 56 |
"step": 800
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 4.25,
|
| 60 |
+
"learning_rate": 1.5714999999999998e-05,
|
| 61 |
+
"loss": 3.1431,
|
| 62 |
"step": 900
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 4.72,
|
| 66 |
+
"learning_rate": 1.7465e-05,
|
| 67 |
+
"loss": 3.0387,
|
| 68 |
"step": 1000
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 4.72,
|
| 72 |
+
"eval_loss": 3.0891902446746826,
|
| 73 |
+
"eval_runtime": 96.3286,
|
| 74 |
+
"eval_samples_per_second": 31.538,
|
| 75 |
+
"eval_steps_per_second": 0.986,
|
| 76 |
"eval_wer": 1.0,
|
| 77 |
"step": 1000
|
| 78 |
},
|
| 79 |
{
|
| 80 |
"epoch": 5.19,
|
| 81 |
+
"learning_rate": 1.9215e-05,
|
| 82 |
+
"loss": 2.9827,
|
| 83 |
"step": 1100
|
| 84 |
},
|
| 85 |
{
|
| 86 |
"epoch": 5.66,
|
| 87 |
+
"learning_rate": 2.0964999999999997e-05,
|
| 88 |
+
"loss": 2.8891,
|
| 89 |
"step": 1200
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 6.13,
|
| 93 |
+
"learning_rate": 2.2715e-05,
|
| 94 |
+
"loss": 2.8309,
|
| 95 |
"step": 1300
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 6.6,
|
| 99 |
+
"learning_rate": 2.4464999999999997e-05,
|
| 100 |
+
"loss": 2.8077,
|
| 101 |
"step": 1400
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 7.08,
|
| 105 |
+
"learning_rate": 2.6215e-05,
|
| 106 |
+
"loss": 2.7824,
|
| 107 |
"step": 1500
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 7.55,
|
| 111 |
+
"learning_rate": 2.7965e-05,
|
| 112 |
+
"loss": 2.7682,
|
| 113 |
"step": 1600
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 8.02,
|
| 117 |
+
"learning_rate": 2.9714999999999996e-05,
|
| 118 |
+
"loss": 2.7261,
|
| 119 |
"step": 1700
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 8.49,
|
| 123 |
+
"learning_rate": 3.1465e-05,
|
| 124 |
+
"loss": 2.4865,
|
| 125 |
"step": 1800
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 8.96,
|
| 129 |
+
"learning_rate": 3.321499999999999e-05,
|
| 130 |
+
"loss": 2.0721,
|
| 131 |
"step": 1900
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 9.43,
|
| 135 |
+
"learning_rate": 3.4964999999999995e-05,
|
| 136 |
+
"loss": 1.7911,
|
| 137 |
"step": 2000
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 9.43,
|
| 141 |
+
"eval_loss": 0.845119059085846,
|
| 142 |
+
"eval_runtime": 95.4637,
|
| 143 |
+
"eval_samples_per_second": 31.824,
|
| 144 |
+
"eval_steps_per_second": 0.995,
|
| 145 |
+
"eval_wer": 0.6702038536721586,
|
| 146 |
"step": 2000
|
| 147 |
},
|
| 148 |
{
|
| 149 |
"epoch": 9.91,
|
| 150 |
+
"learning_rate": 3.6714999999999997e-05,
|
| 151 |
+
"loss": 1.667,
|
| 152 |
"step": 2100
|
| 153 |
},
|
| 154 |
{
|
| 155 |
"epoch": 10.38,
|
| 156 |
+
"learning_rate": 3.8465e-05,
|
| 157 |
+
"loss": 1.5783,
|
| 158 |
"step": 2200
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 10.85,
|
| 162 |
+
"learning_rate": 4.0215e-05,
|
| 163 |
+
"loss": 1.5165,
|
| 164 |
"step": 2300
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 11.32,
|
| 168 |
+
"learning_rate": 4.1964999999999995e-05,
|
| 169 |
+
"loss": 1.4551,
|
| 170 |
"step": 2400
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 11.79,
|
| 174 |
+
"learning_rate": 4.3715e-05,
|
| 175 |
+
"loss": 1.4246,
|
| 176 |
"step": 2500
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 12.26,
|
| 180 |
+
"learning_rate": 4.546499999999999e-05,
|
| 181 |
+
"loss": 1.3976,
|
| 182 |
"step": 2600
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 12.74,
|
| 186 |
+
"learning_rate": 4.7214999999999994e-05,
|
| 187 |
+
"loss": 1.3508,
|
| 188 |
"step": 2700
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 13.21,
|
| 192 |
+
"learning_rate": 4.8964999999999996e-05,
|
| 193 |
+
"loss": 1.3231,
|
| 194 |
"step": 2800
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 13.68,
|
| 198 |
+
"learning_rate": 5.0715e-05,
|
| 199 |
+
"loss": 1.319,
|
| 200 |
"step": 2900
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 14.15,
|
| 204 |
+
"learning_rate": 5.2465e-05,
|
| 205 |
+
"loss": 1.2826,
|
| 206 |
"step": 3000
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 14.15,
|
| 210 |
+
"eval_loss": 0.4210692048072815,
|
| 211 |
+
"eval_runtime": 96.3761,
|
| 212 |
+
"eval_samples_per_second": 31.522,
|
| 213 |
+
"eval_steps_per_second": 0.986,
|
| 214 |
+
"eval_wer": 0.41658754537838594,
|
| 215 |
"step": 3000
|
| 216 |
},
|
| 217 |
{
|
| 218 |
"epoch": 14.62,
|
| 219 |
+
"learning_rate": 5.4214999999999995e-05,
|
| 220 |
+
"loss": 1.279,
|
| 221 |
"step": 3100
|
| 222 |
},
|
| 223 |
{
|
| 224 |
"epoch": 15.09,
|
| 225 |
+
"learning_rate": 5.5965e-05,
|
| 226 |
+
"loss": 1.2713,
|
| 227 |
"step": 3200
|
| 228 |
},
|
| 229 |
{
|
| 230 |
"epoch": 15.57,
|
| 231 |
+
"learning_rate": 5.7715e-05,
|
| 232 |
+
"loss": 1.2416,
|
| 233 |
"step": 3300
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 16.04,
|
| 237 |
+
"learning_rate": 5.9464999999999994e-05,
|
| 238 |
+
"loss": 1.2412,
|
| 239 |
"step": 3400
|
| 240 |
},
|
| 241 |
{
|
| 242 |
"epoch": 16.51,
|
| 243 |
+
"learning_rate": 6.1215e-05,
|
| 244 |
+
"loss": 1.2136,
|
| 245 |
"step": 3500
|
| 246 |
},
|
| 247 |
{
|
| 248 |
"epoch": 16.98,
|
| 249 |
+
"learning_rate": 6.296499999999999e-05,
|
| 250 |
+
"loss": 1.2282,
|
| 251 |
"step": 3600
|
| 252 |
},
|
| 253 |
{
|
| 254 |
"epoch": 17.45,
|
| 255 |
+
"learning_rate": 6.471499999999999e-05,
|
| 256 |
+
"loss": 1.1977,
|
| 257 |
"step": 3700
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 17.92,
|
| 261 |
+
"learning_rate": 6.6465e-05,
|
| 262 |
+
"loss": 1.204,
|
| 263 |
"step": 3800
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 18.4,
|
| 267 |
+
"learning_rate": 6.8215e-05,
|
| 268 |
+
"loss": 1.1807,
|
| 269 |
"step": 3900
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 18.87,
|
| 273 |
+
"learning_rate": 6.9965e-05,
|
| 274 |
+
"loss": 1.1802,
|
| 275 |
"step": 4000
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 18.87,
|
| 279 |
+
"eval_loss": 0.35076606273651123,
|
| 280 |
+
"eval_runtime": 97.3621,
|
| 281 |
+
"eval_samples_per_second": 31.203,
|
| 282 |
+
"eval_steps_per_second": 0.976,
|
| 283 |
+
"eval_wer": 0.468975146607093,
|
| 284 |
"step": 4000
|
| 285 |
},
|
| 286 |
{
|
| 287 |
"epoch": 19.34,
|
| 288 |
+
"learning_rate": 6.960116279069767e-05,
|
| 289 |
+
"loss": 1.1843,
|
| 290 |
"step": 4100
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 19.81,
|
| 294 |
+
"learning_rate": 6.919418604651161e-05,
|
| 295 |
+
"loss": 1.1726,
|
| 296 |
"step": 4200
|
| 297 |
},
|
| 298 |
{
|
| 299 |
"epoch": 20.28,
|
| 300 |
+
"learning_rate": 6.879127906976743e-05,
|
| 301 |
+
"loss": 1.1681,
|
| 302 |
"step": 4300
|
| 303 |
},
|
| 304 |
{
|
| 305 |
"epoch": 20.75,
|
| 306 |
+
"learning_rate": 6.838430232558139e-05,
|
| 307 |
+
"loss": 1.1553,
|
| 308 |
"step": 4400
|
| 309 |
},
|
| 310 |
{
|
| 311 |
"epoch": 21.23,
|
| 312 |
+
"learning_rate": 6.79813953488372e-05,
|
| 313 |
+
"loss": 1.1477,
|
| 314 |
"step": 4500
|
| 315 |
},
|
| 316 |
{
|
| 317 |
"epoch": 21.7,
|
| 318 |
+
"learning_rate": 6.757441860465116e-05,
|
| 319 |
+
"loss": 1.1251,
|
| 320 |
"step": 4600
|
| 321 |
},
|
| 322 |
{
|
| 323 |
"epoch": 22.17,
|
| 324 |
+
"learning_rate": 6.71674418604651e-05,
|
| 325 |
+
"loss": 1.1271,
|
| 326 |
"step": 4700
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 22.64,
|
| 330 |
+
"learning_rate": 6.676046511627906e-05,
|
| 331 |
+
"loss": 1.1251,
|
| 332 |
"step": 4800
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 23.11,
|
| 336 |
+
"learning_rate": 6.635755813953488e-05,
|
| 337 |
+
"loss": 1.1315,
|
| 338 |
"step": 4900
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 23.58,
|
| 342 |
+
"learning_rate": 6.595058139534883e-05,
|
| 343 |
+
"loss": 1.1065,
|
| 344 |
"step": 5000
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 23.58,
|
| 348 |
+
"eval_loss": 0.3318960666656494,
|
| 349 |
+
"eval_runtime": 96.472,
|
| 350 |
+
"eval_samples_per_second": 31.491,
|
| 351 |
+
"eval_steps_per_second": 0.985,
|
| 352 |
+
"eval_wer": 0.46618263055012565,
|
| 353 |
"step": 5000
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 24.06,
|
| 357 |
+
"learning_rate": 6.554360465116279e-05,
|
| 358 |
+
"loss": 1.1061,
|
| 359 |
"step": 5100
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"epoch": 24.53,
|
| 363 |
+
"learning_rate": 6.513662790697673e-05,
|
| 364 |
+
"loss": 1.1054,
|
| 365 |
"step": 5200
|
| 366 |
},
|
| 367 |
{
|
| 368 |
"epoch": 25.0,
|
| 369 |
+
"learning_rate": 6.472965116279069e-05,
|
| 370 |
+
"loss": 1.0993,
|
| 371 |
"step": 5300
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 25.47,
|
| 375 |
+
"learning_rate": 6.432267441860465e-05,
|
| 376 |
+
"loss": 1.1,
|
| 377 |
"step": 5400
|
| 378 |
},
|
| 379 |
{
|
| 380 |
"epoch": 25.94,
|
| 381 |
+
"learning_rate": 6.39156976744186e-05,
|
| 382 |
+
"loss": 1.1019,
|
| 383 |
"step": 5500
|
| 384 |
},
|
| 385 |
{
|
| 386 |
"epoch": 26.42,
|
| 387 |
+
"learning_rate": 6.350872093023256e-05,
|
| 388 |
+
"loss": 1.078,
|
| 389 |
"step": 5600
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 26.89,
|
| 393 |
+
"learning_rate": 6.31017441860465e-05,
|
| 394 |
+
"loss": 1.0847,
|
| 395 |
"step": 5700
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 27.36,
|
| 399 |
+
"learning_rate": 6.269476744186046e-05,
|
| 400 |
+
"loss": 1.0744,
|
| 401 |
"step": 5800
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 27.83,
|
| 405 |
+
"learning_rate": 6.228779069767442e-05,
|
| 406 |
+
"loss": 1.0703,
|
| 407 |
"step": 5900
|
| 408 |
},
|
| 409 |
{
|
| 410 |
"epoch": 28.3,
|
| 411 |
+
"learning_rate": 6.188081395348836e-05,
|
| 412 |
+
"loss": 1.0921,
|
| 413 |
"step": 6000
|
| 414 |
},
|
| 415 |
{
|
| 416 |
"epoch": 28.3,
|
| 417 |
+
"eval_loss": 0.30558520555496216,
|
| 418 |
+
"eval_runtime": 94.8784,
|
| 419 |
+
"eval_samples_per_second": 32.02,
|
| 420 |
+
"eval_steps_per_second": 1.001,
|
| 421 |
+
"eval_wer": 0.3879921809550405,
|
| 422 |
"step": 6000
|
| 423 |
},
|
| 424 |
{
|
| 425 |
"epoch": 28.77,
|
| 426 |
+
"learning_rate": 6.147383720930232e-05,
|
| 427 |
+
"loss": 1.0704,
|
| 428 |
"step": 6100
|
| 429 |
},
|
| 430 |
{
|
| 431 |
"epoch": 29.25,
|
| 432 |
+
"learning_rate": 6.106686046511627e-05,
|
| 433 |
+
"loss": 1.0479,
|
| 434 |
"step": 6200
|
| 435 |
},
|
| 436 |
{
|
| 437 |
"epoch": 29.72,
|
| 438 |
+
"learning_rate": 6.0659883720930225e-05,
|
| 439 |
+
"loss": 1.0583,
|
| 440 |
"step": 6300
|
| 441 |
},
|
| 442 |
{
|
| 443 |
"epoch": 30.19,
|
| 444 |
+
"learning_rate": 6.0252906976744175e-05,
|
| 445 |
+
"loss": 1.0515,
|
| 446 |
"step": 6400
|
| 447 |
},
|
| 448 |
{
|
| 449 |
"epoch": 30.66,
|
| 450 |
+
"learning_rate": 5.984593023255814e-05,
|
| 451 |
+
"loss": 1.0531,
|
| 452 |
"step": 6500
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 31.13,
|
| 456 |
+
"learning_rate": 5.943895348837209e-05,
|
| 457 |
+
"loss": 1.0505,
|
| 458 |
"step": 6600
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 31.6,
|
| 462 |
+
"learning_rate": 5.9031976744186045e-05,
|
| 463 |
+
"loss": 1.0314,
|
| 464 |
"step": 6700
|
| 465 |
},
|
| 466 |
{
|
| 467 |
"epoch": 32.08,
|
| 468 |
+
"learning_rate": 5.8624999999999995e-05,
|
| 469 |
+
"loss": 1.0473,
|
| 470 |
"step": 6800
|
| 471 |
},
|
| 472 |
{
|
| 473 |
"epoch": 32.55,
|
| 474 |
+
"learning_rate": 5.821802325581395e-05,
|
| 475 |
+
"loss": 1.0362,
|
| 476 |
"step": 6900
|
| 477 |
},
|
| 478 |
{
|
| 479 |
"epoch": 33.02,
|
| 480 |
+
"learning_rate": 5.78110465116279e-05,
|
| 481 |
+
"loss": 1.0366,
|
| 482 |
"step": 7000
|
| 483 |
},
|
| 484 |
{
|
| 485 |
"epoch": 33.02,
|
| 486 |
+
"eval_loss": 0.29966893792152405,
|
| 487 |
+
"eval_runtime": 94.2922,
|
| 488 |
+
"eval_samples_per_second": 32.219,
|
| 489 |
+
"eval_steps_per_second": 1.008,
|
| 490 |
+
"eval_wer": 0.36654565763753144,
|
| 491 |
"step": 7000
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 33.49,
|
| 495 |
+
"learning_rate": 5.740406976744185e-05,
|
| 496 |
+
"loss": 1.0361,
|
| 497 |
"step": 7100
|
| 498 |
},
|
| 499 |
{
|
| 500 |
"epoch": 33.96,
|
| 501 |
+
"learning_rate": 5.699709302325581e-05,
|
| 502 |
+
"loss": 1.0298,
|
| 503 |
"step": 7200
|
| 504 |
},
|
| 505 |
{
|
| 506 |
"epoch": 34.43,
|
| 507 |
+
"learning_rate": 5.659011627906976e-05,
|
| 508 |
+
"loss": 1.0185,
|
| 509 |
"step": 7300
|
| 510 |
},
|
| 511 |
{
|
| 512 |
"epoch": 34.91,
|
| 513 |
+
"learning_rate": 5.6183139534883716e-05,
|
| 514 |
+
"loss": 1.0161,
|
| 515 |
"step": 7400
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 35.38,
|
| 519 |
+
"learning_rate": 5.5776162790697666e-05,
|
| 520 |
+
"loss": 1.0134,
|
| 521 |
"step": 7500
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"epoch": 35.85,
|
| 525 |
+
"learning_rate": 5.536918604651163e-05,
|
| 526 |
+
"loss": 1.0138,
|
| 527 |
"step": 7600
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"epoch": 36.32,
|
| 531 |
+
"learning_rate": 5.496220930232558e-05,
|
| 532 |
+
"loss": 1.0076,
|
| 533 |
"step": 7700
|
| 534 |
},
|
| 535 |
{
|
| 536 |
"epoch": 36.79,
|
| 537 |
+
"learning_rate": 5.455523255813953e-05,
|
| 538 |
+
"loss": 1.0134,
|
| 539 |
"step": 7800
|
| 540 |
},
|
| 541 |
{
|
| 542 |
"epoch": 37.26,
|
| 543 |
+
"learning_rate": 5.415232558139534e-05,
|
| 544 |
+
"loss": 0.9983,
|
| 545 |
"step": 7900
|
| 546 |
},
|
| 547 |
{
|
| 548 |
"epoch": 37.74,
|
| 549 |
+
"learning_rate": 5.37453488372093e-05,
|
| 550 |
+
"loss": 0.9988,
|
| 551 |
"step": 8000
|
| 552 |
},
|
| 553 |
{
|
| 554 |
"epoch": 37.74,
|
| 555 |
+
"eval_loss": 0.2971556782722473,
|
| 556 |
+
"eval_runtime": 94.706,
|
| 557 |
+
"eval_samples_per_second": 32.078,
|
| 558 |
+
"eval_steps_per_second": 1.003,
|
| 559 |
+
"eval_wer": 0.36526110025132646,
|
| 560 |
"step": 8000
|
| 561 |
},
|
| 562 |
{
|
| 563 |
"epoch": 38.21,
|
| 564 |
+
"learning_rate": 5.333837209302325e-05,
|
| 565 |
+
"loss": 0.9974,
|
| 566 |
"step": 8100
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 38.68,
|
| 570 |
+
"learning_rate": 5.293139534883721e-05,
|
| 571 |
+
"loss": 0.9906,
|
| 572 |
"step": 8200
|
| 573 |
},
|
| 574 |
{
|
| 575 |
"epoch": 39.15,
|
| 576 |
+
"learning_rate": 5.252441860465116e-05,
|
| 577 |
+
"loss": 1.0178,
|
| 578 |
"step": 8300
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 39.62,
|
| 582 |
+
"learning_rate": 5.211744186046511e-05,
|
| 583 |
+
"loss": 0.99,
|
| 584 |
"step": 8400
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 40.09,
|
| 588 |
+
"learning_rate": 5.171046511627907e-05,
|
| 589 |
+
"loss": 0.9983,
|
| 590 |
"step": 8500
|
| 591 |
},
|
| 592 |
{
|
| 593 |
"epoch": 40.57,
|
| 594 |
+
"learning_rate": 5.130348837209302e-05,
|
| 595 |
+
"loss": 0.9835,
|
| 596 |
"step": 8600
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 41.04,
|
| 600 |
+
"learning_rate": 5.0896511627906974e-05,
|
| 601 |
+
"loss": 0.9902,
|
| 602 |
"step": 8700
|
| 603 |
},
|
| 604 |
{
|
| 605 |
"epoch": 41.51,
|
| 606 |
+
"learning_rate": 5.0489534883720924e-05,
|
| 607 |
+
"loss": 0.9604,
|
| 608 |
"step": 8800
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 41.98,
|
| 612 |
+
"learning_rate": 5.0082558139534874e-05,
|
| 613 |
+
"loss": 0.9836,
|
| 614 |
"step": 8900
|
| 615 |
},
|
| 616 |
{
|
| 617 |
"epoch": 42.45,
|
| 618 |
+
"learning_rate": 4.967558139534883e-05,
|
| 619 |
+
"loss": 0.9864,
|
| 620 |
"step": 9000
|
| 621 |
},
|
| 622 |
{
|
| 623 |
"epoch": 42.45,
|
| 624 |
+
"eval_loss": 0.2696739435195923,
|
| 625 |
+
"eval_runtime": 94.7296,
|
| 626 |
+
"eval_samples_per_second": 32.07,
|
| 627 |
+
"eval_steps_per_second": 1.003,
|
| 628 |
+
"eval_wer": 0.33705668807595646,
|
| 629 |
"step": 9000
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 42.92,
|
| 633 |
+
"learning_rate": 4.926860465116279e-05,
|
| 634 |
+
"loss": 0.9692,
|
| 635 |
"step": 9100
|
| 636 |
},
|
| 637 |
{
|
| 638 |
"epoch": 43.4,
|
| 639 |
+
"learning_rate": 4.8861627906976744e-05,
|
| 640 |
+
"loss": 0.9643,
|
| 641 |
"step": 9200
|
| 642 |
},
|
| 643 |
{
|
| 644 |
"epoch": 43.87,
|
| 645 |
+
"learning_rate": 4.8454651162790694e-05,
|
| 646 |
+
"loss": 0.9682,
|
| 647 |
"step": 9300
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"epoch": 44.34,
|
| 651 |
+
"learning_rate": 4.804767441860465e-05,
|
| 652 |
+
"loss": 0.9651,
|
| 653 |
"step": 9400
|
| 654 |
},
|
| 655 |
{
|
| 656 |
"epoch": 44.81,
|
| 657 |
+
"learning_rate": 4.76406976744186e-05,
|
| 658 |
+
"loss": 0.9409,
|
| 659 |
"step": 9500
|
| 660 |
},
|
| 661 |
{
|
| 662 |
"epoch": 45.28,
|
| 663 |
+
"learning_rate": 4.723372093023255e-05,
|
| 664 |
+
"loss": 0.9792,
|
| 665 |
"step": 9600
|
| 666 |
},
|
| 667 |
{
|
| 668 |
"epoch": 45.75,
|
| 669 |
+
"learning_rate": 4.682674418604651e-05,
|
| 670 |
+
"loss": 0.9405,
|
| 671 |
"step": 9700
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 46.23,
|
| 675 |
+
"learning_rate": 4.641976744186046e-05,
|
| 676 |
+
"loss": 0.9679,
|
| 677 |
"step": 9800
|
| 678 |
},
|
| 679 |
{
|
| 680 |
"epoch": 46.7,
|
| 681 |
+
"learning_rate": 4.6012790697674415e-05,
|
| 682 |
+
"loss": 0.9468,
|
| 683 |
"step": 9900
|
| 684 |
},
|
| 685 |
{
|
| 686 |
"epoch": 47.17,
|
| 687 |
+
"learning_rate": 4.5605813953488365e-05,
|
| 688 |
+
"loss": 0.9558,
|
| 689 |
"step": 10000
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 47.17,
|
| 693 |
+
"eval_loss": 0.27388861775398254,
|
| 694 |
+
"eval_runtime": 94.8478,
|
| 695 |
+
"eval_samples_per_second": 32.03,
|
| 696 |
+
"eval_steps_per_second": 1.002,
|
| 697 |
+
"eval_wer": 0.314102206087685,
|
| 698 |
"step": 10000
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 47.64,
|
| 702 |
+
"learning_rate": 4.519883720930233e-05,
|
| 703 |
+
"loss": 0.95,
|
| 704 |
"step": 10100
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 48.11,
|
| 708 |
+
"learning_rate": 4.479186046511628e-05,
|
| 709 |
+
"loss": 0.9574,
|
| 710 |
"step": 10200
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 48.58,
|
| 714 |
+
"learning_rate": 4.438895348837209e-05,
|
| 715 |
+
"loss": 0.9473,
|
| 716 |
"step": 10300
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 49.06,
|
| 720 |
+
"learning_rate": 4.398197674418604e-05,
|
| 721 |
+
"loss": 0.9319,
|
| 722 |
"step": 10400
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 49.53,
|
| 726 |
+
"learning_rate": 4.3575e-05,
|
| 727 |
+
"loss": 0.9413,
|
| 728 |
"step": 10500
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 50.0,
|
| 732 |
+
"learning_rate": 4.316802325581395e-05,
|
| 733 |
+
"loss": 0.9442,
|
| 734 |
"step": 10600
|
| 735 |
},
|
| 736 |
{
|
| 737 |
+
"epoch": 50.47,
|
| 738 |
+
"learning_rate": 4.27610465116279e-05,
|
| 739 |
+
"loss": 0.9315,
|
| 740 |
+
"step": 10700
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 50.94,
|
| 744 |
+
"learning_rate": 4.235406976744186e-05,
|
| 745 |
+
"loss": 0.9334,
|
| 746 |
+
"step": 10800
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 51.42,
|
| 750 |
+
"learning_rate": 4.194709302325581e-05,
|
| 751 |
+
"loss": 0.928,
|
| 752 |
+
"step": 10900
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 51.89,
|
| 756 |
+
"learning_rate": 4.1540116279069766e-05,
|
| 757 |
+
"loss": 0.9094,
|
| 758 |
+
"step": 11000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 51.89,
|
| 762 |
+
"eval_loss": 0.2657303512096405,
|
| 763 |
+
"eval_runtime": 93.9674,
|
| 764 |
+
"eval_samples_per_second": 32.33,
|
| 765 |
+
"eval_steps_per_second": 1.011,
|
| 766 |
+
"eval_wer": 0.35330913152750626,
|
| 767 |
+
"step": 11000
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 52.36,
|
| 771 |
+
"learning_rate": 4.1133139534883716e-05,
|
| 772 |
+
"loss": 0.917,
|
| 773 |
+
"step": 11100
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 52.83,
|
| 777 |
+
"learning_rate": 4.072616279069767e-05,
|
| 778 |
+
"loss": 0.9164,
|
| 779 |
+
"step": 11200
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 53.3,
|
| 783 |
+
"learning_rate": 4.031918604651162e-05,
|
| 784 |
+
"loss": 0.915,
|
| 785 |
+
"step": 11300
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 53.77,
|
| 789 |
+
"learning_rate": 3.991220930232557e-05,
|
| 790 |
+
"loss": 0.905,
|
| 791 |
+
"step": 11400
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 54.25,
|
| 795 |
+
"learning_rate": 3.950523255813953e-05,
|
| 796 |
+
"loss": 0.9144,
|
| 797 |
+
"step": 11500
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 54.72,
|
| 801 |
+
"learning_rate": 3.9098255813953487e-05,
|
| 802 |
+
"loss": 0.8963,
|
| 803 |
+
"step": 11600
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 55.19,
|
| 807 |
+
"learning_rate": 3.8691279069767443e-05,
|
| 808 |
+
"loss": 0.8935,
|
| 809 |
+
"step": 11700
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 55.66,
|
| 813 |
+
"learning_rate": 3.8284302325581393e-05,
|
| 814 |
+
"loss": 0.8964,
|
| 815 |
+
"step": 11800
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 56.13,
|
| 819 |
+
"learning_rate": 3.7877325581395344e-05,
|
| 820 |
+
"loss": 0.9213,
|
| 821 |
+
"step": 11900
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 56.6,
|
| 825 |
+
"learning_rate": 3.74703488372093e-05,
|
| 826 |
+
"loss": 0.9034,
|
| 827 |
+
"step": 12000
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 56.6,
|
| 831 |
+
"eval_loss": 0.2699211537837982,
|
| 832 |
+
"eval_runtime": 95.9143,
|
| 833 |
+
"eval_samples_per_second": 31.674,
|
| 834 |
+
"eval_steps_per_second": 0.99,
|
| 835 |
+
"eval_wer": 0.3396816531695057,
|
| 836 |
+
"step": 12000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 57.08,
|
| 840 |
+
"learning_rate": 3.706337209302325e-05,
|
| 841 |
+
"loss": 0.8992,
|
| 842 |
+
"step": 12100
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 57.55,
|
| 846 |
+
"learning_rate": 3.665639534883721e-05,
|
| 847 |
+
"loss": 0.8911,
|
| 848 |
+
"step": 12200
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 58.02,
|
| 852 |
+
"learning_rate": 3.624941860465116e-05,
|
| 853 |
+
"loss": 0.8907,
|
| 854 |
+
"step": 12300
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 58.49,
|
| 858 |
+
"learning_rate": 3.5842441860465114e-05,
|
| 859 |
+
"loss": 0.8957,
|
| 860 |
+
"step": 12400
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 58.96,
|
| 864 |
+
"learning_rate": 3.5439534883720924e-05,
|
| 865 |
+
"loss": 0.8888,
|
| 866 |
+
"step": 12500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 59.43,
|
| 870 |
+
"learning_rate": 3.503255813953488e-05,
|
| 871 |
+
"loss": 0.8835,
|
| 872 |
+
"step": 12600
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 59.91,
|
| 876 |
+
"learning_rate": 3.462558139534883e-05,
|
| 877 |
+
"loss": 0.8866,
|
| 878 |
+
"step": 12700
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 60.38,
|
| 882 |
+
"learning_rate": 3.421860465116279e-05,
|
| 883 |
+
"loss": 0.8887,
|
| 884 |
+
"step": 12800
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 60.85,
|
| 888 |
+
"learning_rate": 3.3811627906976744e-05,
|
| 889 |
+
"loss": 0.8919,
|
| 890 |
+
"step": 12900
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 61.32,
|
| 894 |
+
"learning_rate": 3.3404651162790694e-05,
|
| 895 |
+
"loss": 0.8907,
|
| 896 |
+
"step": 13000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 61.32,
|
| 900 |
+
"eval_loss": 0.2764558792114258,
|
| 901 |
+
"eval_runtime": 94.1878,
|
| 902 |
+
"eval_samples_per_second": 32.255,
|
| 903 |
+
"eval_steps_per_second": 1.009,
|
| 904 |
+
"eval_wer": 0.3469980452387601,
|
| 905 |
+
"step": 13000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 61.79,
|
| 909 |
+
"learning_rate": 3.2997674418604644e-05,
|
| 910 |
+
"loss": 0.8729,
|
| 911 |
+
"step": 13100
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 62.26,
|
| 915 |
+
"learning_rate": 3.25906976744186e-05,
|
| 916 |
+
"loss": 0.8808,
|
| 917 |
+
"step": 13200
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 62.74,
|
| 921 |
+
"learning_rate": 3.218372093023256e-05,
|
| 922 |
+
"loss": 0.8845,
|
| 923 |
+
"step": 13300
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 63.21,
|
| 927 |
+
"learning_rate": 3.177674418604651e-05,
|
| 928 |
+
"loss": 0.8616,
|
| 929 |
+
"step": 13400
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 63.68,
|
| 933 |
+
"learning_rate": 3.1369767441860465e-05,
|
| 934 |
+
"loss": 0.8755,
|
| 935 |
+
"step": 13500
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 64.15,
|
| 939 |
+
"learning_rate": 3.0962790697674415e-05,
|
| 940 |
+
"loss": 0.8808,
|
| 941 |
+
"step": 13600
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 64.62,
|
| 945 |
+
"learning_rate": 3.055988372093023e-05,
|
| 946 |
+
"loss": 0.8823,
|
| 947 |
+
"step": 13700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 65.09,
|
| 951 |
+
"learning_rate": 3.0152906976744185e-05,
|
| 952 |
+
"loss": 0.8667,
|
| 953 |
+
"step": 13800
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 65.57,
|
| 957 |
+
"learning_rate": 2.974593023255814e-05,
|
| 958 |
+
"loss": 0.8628,
|
| 959 |
+
"step": 13900
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 66.04,
|
| 963 |
+
"learning_rate": 2.933895348837209e-05,
|
| 964 |
+
"loss": 0.8631,
|
| 965 |
+
"step": 14000
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 66.04,
|
| 969 |
+
"eval_loss": 0.27741023898124695,
|
| 970 |
+
"eval_runtime": 93.7633,
|
| 971 |
+
"eval_samples_per_second": 32.401,
|
| 972 |
+
"eval_steps_per_second": 1.013,
|
| 973 |
+
"eval_wer": 0.33459927394582517,
|
| 974 |
+
"step": 14000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 66.51,
|
| 978 |
+
"learning_rate": 2.8931976744186042e-05,
|
| 979 |
+
"loss": 0.865,
|
| 980 |
+
"step": 14100
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 66.98,
|
| 984 |
+
"learning_rate": 2.8524999999999995e-05,
|
| 985 |
+
"loss": 0.857,
|
| 986 |
+
"step": 14200
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 67.45,
|
| 990 |
+
"learning_rate": 2.8118023255813952e-05,
|
| 991 |
+
"loss": 0.8575,
|
| 992 |
+
"step": 14300
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 67.92,
|
| 996 |
+
"learning_rate": 2.7711046511627906e-05,
|
| 997 |
+
"loss": 0.8497,
|
| 998 |
+
"step": 14400
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 68.4,
|
| 1002 |
+
"learning_rate": 2.730406976744186e-05,
|
| 1003 |
+
"loss": 0.855,
|
| 1004 |
+
"step": 14500
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 68.87,
|
| 1008 |
+
"learning_rate": 2.689709302325581e-05,
|
| 1009 |
+
"loss": 0.845,
|
| 1010 |
+
"step": 14600
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 69.34,
|
| 1014 |
+
"learning_rate": 2.6490116279069763e-05,
|
| 1015 |
+
"loss": 0.8324,
|
| 1016 |
+
"step": 14700
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 69.81,
|
| 1020 |
+
"learning_rate": 2.608313953488372e-05,
|
| 1021 |
+
"loss": 0.844,
|
| 1022 |
+
"step": 14800
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 70.28,
|
| 1026 |
+
"learning_rate": 2.5676162790697673e-05,
|
| 1027 |
+
"loss": 0.8483,
|
| 1028 |
+
"step": 14900
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 70.75,
|
| 1032 |
+
"learning_rate": 2.5269186046511626e-05,
|
| 1033 |
+
"loss": 0.8389,
|
| 1034 |
+
"step": 15000
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 70.75,
|
| 1038 |
+
"eval_loss": 0.2742805778980255,
|
| 1039 |
+
"eval_runtime": 94.8277,
|
| 1040 |
+
"eval_samples_per_second": 32.037,
|
| 1041 |
+
"eval_steps_per_second": 1.002,
|
| 1042 |
+
"eval_wer": 0.33649818486456295,
|
| 1043 |
+
"step": 15000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 71.23,
|
| 1047 |
+
"learning_rate": 2.486220930232558e-05,
|
| 1048 |
+
"loss": 0.8383,
|
| 1049 |
+
"step": 15100
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 71.7,
|
| 1053 |
+
"learning_rate": 2.445523255813953e-05,
|
| 1054 |
+
"loss": 0.8468,
|
| 1055 |
+
"step": 15200
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 72.17,
|
| 1059 |
+
"learning_rate": 2.4048255813953487e-05,
|
| 1060 |
+
"loss": 0.8511,
|
| 1061 |
+
"step": 15300
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 72.64,
|
| 1065 |
+
"learning_rate": 2.364127906976744e-05,
|
| 1066 |
+
"loss": 0.8457,
|
| 1067 |
+
"step": 15400
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 73.11,
|
| 1071 |
+
"learning_rate": 2.3234302325581394e-05,
|
| 1072 |
+
"loss": 0.8314,
|
| 1073 |
+
"step": 15500
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 73.58,
|
| 1077 |
+
"learning_rate": 2.2827325581395347e-05,
|
| 1078 |
+
"loss": 0.8278,
|
| 1079 |
+
"step": 15600
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 74.06,
|
| 1083 |
+
"learning_rate": 2.24203488372093e-05,
|
| 1084 |
+
"loss": 0.8481,
|
| 1085 |
+
"step": 15700
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 74.53,
|
| 1089 |
+
"learning_rate": 2.201337209302325e-05,
|
| 1090 |
+
"loss": 0.8158,
|
| 1091 |
+
"step": 15800
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 75.0,
|
| 1095 |
+
"learning_rate": 2.1606395348837207e-05,
|
| 1096 |
+
"loss": 0.8356,
|
| 1097 |
+
"step": 15900
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 75.47,
|
| 1101 |
+
"learning_rate": 2.119941860465116e-05,
|
| 1102 |
+
"loss": 0.8214,
|
| 1103 |
+
"step": 16000
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 75.47,
|
| 1107 |
+
"eval_loss": 0.2778330445289612,
|
| 1108 |
+
"eval_runtime": 94.0407,
|
| 1109 |
+
"eval_samples_per_second": 32.305,
|
| 1110 |
+
"eval_steps_per_second": 1.01,
|
| 1111 |
+
"eval_wer": 0.32013404077073443,
|
| 1112 |
+
"step": 16000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 75.94,
|
| 1116 |
+
"learning_rate": 2.0792441860465114e-05,
|
| 1117 |
+
"loss": 0.8293,
|
| 1118 |
+
"step": 16100
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 76.42,
|
| 1122 |
+
"learning_rate": 2.0389534883720927e-05,
|
| 1123 |
+
"loss": 0.834,
|
| 1124 |
+
"step": 16200
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 76.89,
|
| 1128 |
+
"learning_rate": 1.9982558139534884e-05,
|
| 1129 |
+
"loss": 0.8308,
|
| 1130 |
+
"step": 16300
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 77.36,
|
| 1134 |
+
"learning_rate": 1.9575581395348838e-05,
|
| 1135 |
+
"loss": 0.8292,
|
| 1136 |
+
"step": 16400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 77.83,
|
| 1140 |
+
"learning_rate": 1.9168604651162788e-05,
|
| 1141 |
+
"loss": 0.8224,
|
| 1142 |
+
"step": 16500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 78.3,
|
| 1146 |
+
"learning_rate": 1.876162790697674e-05,
|
| 1147 |
+
"loss": 0.8172,
|
| 1148 |
+
"step": 16600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 78.77,
|
| 1152 |
+
"learning_rate": 1.8354651162790695e-05,
|
| 1153 |
+
"loss": 0.8202,
|
| 1154 |
+
"step": 16700
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 79.25,
|
| 1158 |
+
"learning_rate": 1.795174418604651e-05,
|
| 1159 |
+
"loss": 0.8222,
|
| 1160 |
+
"step": 16800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 79.72,
|
| 1164 |
+
"learning_rate": 1.7544767441860464e-05,
|
| 1165 |
+
"loss": 0.811,
|
| 1166 |
+
"step": 16900
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 80.19,
|
| 1170 |
+
"learning_rate": 1.7137790697674414e-05,
|
| 1171 |
+
"loss": 0.8195,
|
| 1172 |
+
"step": 17000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 80.19,
|
| 1176 |
+
"eval_loss": 0.2725354731082916,
|
| 1177 |
+
"eval_runtime": 94.0258,
|
| 1178 |
+
"eval_samples_per_second": 32.31,
|
| 1179 |
+
"eval_steps_per_second": 1.01,
|
| 1180 |
+
"eval_wer": 0.32856743926277576,
|
| 1181 |
+
"step": 17000
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 80.66,
|
| 1185 |
+
"learning_rate": 1.673081395348837e-05,
|
| 1186 |
+
"loss": 0.8053,
|
| 1187 |
+
"step": 17100
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 81.13,
|
| 1191 |
+
"learning_rate": 1.6323837209302325e-05,
|
| 1192 |
+
"loss": 0.8152,
|
| 1193 |
+
"step": 17200
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 81.6,
|
| 1197 |
+
"learning_rate": 1.5916860465116278e-05,
|
| 1198 |
+
"loss": 0.8074,
|
| 1199 |
+
"step": 17300
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 82.08,
|
| 1203 |
+
"learning_rate": 1.550988372093023e-05,
|
| 1204 |
+
"loss": 0.8179,
|
| 1205 |
+
"step": 17400
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 82.55,
|
| 1209 |
+
"learning_rate": 1.5102906976744183e-05,
|
| 1210 |
+
"loss": 0.8078,
|
| 1211 |
+
"step": 17500
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 83.02,
|
| 1215 |
+
"learning_rate": 1.4695930232558139e-05,
|
| 1216 |
+
"loss": 0.8078,
|
| 1217 |
+
"step": 17600
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 83.49,
|
| 1221 |
+
"learning_rate": 1.4288953488372092e-05,
|
| 1222 |
+
"loss": 0.8061,
|
| 1223 |
+
"step": 17700
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 83.96,
|
| 1227 |
+
"learning_rate": 1.3881976744186044e-05,
|
| 1228 |
+
"loss": 0.813,
|
| 1229 |
+
"step": 17800
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 84.43,
|
| 1233 |
+
"learning_rate": 1.3474999999999999e-05,
|
| 1234 |
+
"loss": 0.7891,
|
| 1235 |
+
"step": 17900
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 84.91,
|
| 1239 |
+
"learning_rate": 1.3068023255813952e-05,
|
| 1240 |
+
"loss": 0.7994,
|
| 1241 |
+
"step": 18000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 84.91,
|
| 1245 |
+
"eval_loss": 0.2781510651111603,
|
| 1246 |
+
"eval_runtime": 92.9473,
|
| 1247 |
+
"eval_samples_per_second": 32.685,
|
| 1248 |
+
"eval_steps_per_second": 1.022,
|
| 1249 |
+
"eval_wer": 0.33147165596202177,
|
| 1250 |
+
"step": 18000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 85.38,
|
| 1254 |
+
"learning_rate": 1.2661046511627907e-05,
|
| 1255 |
+
"loss": 0.8011,
|
| 1256 |
+
"step": 18100
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 85.85,
|
| 1260 |
+
"learning_rate": 1.225406976744186e-05,
|
| 1261 |
+
"loss": 0.7886,
|
| 1262 |
+
"step": 18200
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 86.32,
|
| 1266 |
+
"learning_rate": 1.1847093023255813e-05,
|
| 1267 |
+
"loss": 0.8107,
|
| 1268 |
+
"step": 18300
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 86.79,
|
| 1272 |
+
"learning_rate": 1.1440116279069768e-05,
|
| 1273 |
+
"loss": 0.7976,
|
| 1274 |
+
"step": 18400
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 87.26,
|
| 1278 |
+
"learning_rate": 1.103313953488372e-05,
|
| 1279 |
+
"loss": 0.7871,
|
| 1280 |
+
"step": 18500
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 87.74,
|
| 1284 |
+
"learning_rate": 1.0626162790697673e-05,
|
| 1285 |
+
"loss": 0.7922,
|
| 1286 |
+
"step": 18600
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 88.21,
|
| 1290 |
+
"learning_rate": 1.0219186046511628e-05,
|
| 1291 |
+
"loss": 0.7826,
|
| 1292 |
+
"step": 18700
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 88.68,
|
| 1296 |
+
"learning_rate": 9.81220930232558e-06,
|
| 1297 |
+
"loss": 0.7901,
|
| 1298 |
+
"step": 18800
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 89.15,
|
| 1302 |
+
"learning_rate": 9.405232558139535e-06,
|
| 1303 |
+
"loss": 0.793,
|
| 1304 |
+
"step": 18900
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 89.62,
|
| 1308 |
+
"learning_rate": 8.998255813953488e-06,
|
| 1309 |
+
"loss": 0.7816,
|
| 1310 |
+
"step": 19000
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 89.62,
|
| 1314 |
+
"eval_loss": 0.2774805724620819,
|
| 1315 |
+
"eval_runtime": 94.1962,
|
| 1316 |
+
"eval_samples_per_second": 32.252,
|
| 1317 |
+
"eval_steps_per_second": 1.009,
|
| 1318 |
+
"eval_wer": 0.3363306339011449,
|
| 1319 |
+
"step": 19000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 90.09,
|
| 1323 |
+
"learning_rate": 8.591279069767442e-06,
|
| 1324 |
+
"loss": 0.7922,
|
| 1325 |
+
"step": 19100
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 90.57,
|
| 1329 |
+
"learning_rate": 8.188372093023255e-06,
|
| 1330 |
+
"loss": 0.7937,
|
| 1331 |
+
"step": 19200
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 91.04,
|
| 1335 |
+
"learning_rate": 7.781395348837208e-06,
|
| 1336 |
+
"loss": 0.784,
|
| 1337 |
+
"step": 19300
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 91.51,
|
| 1341 |
+
"learning_rate": 7.374418604651162e-06,
|
| 1342 |
+
"loss": 0.7811,
|
| 1343 |
+
"step": 19400
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 91.98,
|
| 1347 |
+
"learning_rate": 6.967441860465116e-06,
|
| 1348 |
+
"loss": 0.7784,
|
| 1349 |
+
"step": 19500
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 92.45,
|
| 1353 |
+
"learning_rate": 6.560465116279069e-06,
|
| 1354 |
+
"loss": 0.789,
|
| 1355 |
+
"step": 19600
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 92.92,
|
| 1359 |
+
"learning_rate": 6.153488372093022e-06,
|
| 1360 |
+
"loss": 0.7785,
|
| 1361 |
+
"step": 19700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 93.4,
|
| 1365 |
+
"learning_rate": 5.7465116279069765e-06,
|
| 1366 |
+
"loss": 0.7799,
|
| 1367 |
+
"step": 19800
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 93.87,
|
| 1371 |
+
"learning_rate": 5.33953488372093e-06,
|
| 1372 |
+
"loss": 0.7784,
|
| 1373 |
+
"step": 19900
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 94.34,
|
| 1377 |
+
"learning_rate": 4.9325581395348825e-06,
|
| 1378 |
+
"loss": 0.7816,
|
| 1379 |
+
"step": 20000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 94.34,
|
| 1383 |
+
"eval_loss": 0.2731296420097351,
|
| 1384 |
+
"eval_runtime": 94.2664,
|
| 1385 |
+
"eval_samples_per_second": 32.228,
|
| 1386 |
+
"eval_steps_per_second": 1.008,
|
| 1387 |
+
"eval_wer": 0.32784138508796423,
|
| 1388 |
+
"step": 20000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 94.81,
|
| 1392 |
+
"learning_rate": 4.525581395348837e-06,
|
| 1393 |
+
"loss": 0.7788,
|
| 1394 |
+
"step": 20100
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 95.28,
|
| 1398 |
+
"learning_rate": 4.11860465116279e-06,
|
| 1399 |
+
"loss": 0.781,
|
| 1400 |
+
"step": 20200
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 95.75,
|
| 1404 |
+
"learning_rate": 3.7116279069767437e-06,
|
| 1405 |
+
"loss": 0.785,
|
| 1406 |
+
"step": 20300
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 96.23,
|
| 1410 |
+
"learning_rate": 3.3046511627906976e-06,
|
| 1411 |
+
"loss": 0.7664,
|
| 1412 |
+
"step": 20400
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 96.7,
|
| 1416 |
+
"learning_rate": 2.8976744186046506e-06,
|
| 1417 |
+
"loss": 0.7771,
|
| 1418 |
+
"step": 20500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 97.17,
|
| 1422 |
+
"learning_rate": 2.4906976744186045e-06,
|
| 1423 |
+
"loss": 0.7848,
|
| 1424 |
+
"step": 20600
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 97.64,
|
| 1428 |
+
"learning_rate": 2.083720930232558e-06,
|
| 1429 |
+
"loss": 0.765,
|
| 1430 |
+
"step": 20700
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 98.11,
|
| 1434 |
+
"learning_rate": 1.6767441860465114e-06,
|
| 1435 |
+
"loss": 0.7877,
|
| 1436 |
+
"step": 20800
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 98.58,
|
| 1440 |
+
"learning_rate": 1.2697674418604648e-06,
|
| 1441 |
+
"loss": 0.7681,
|
| 1442 |
+
"step": 20900
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 99.06,
|
| 1446 |
+
"learning_rate": 8.627906976744186e-07,
|
| 1447 |
+
"loss": 0.7635,
|
| 1448 |
+
"step": 21000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 99.06,
|
| 1452 |
+
"eval_loss": 0.276745080947876,
|
| 1453 |
+
"eval_runtime": 95.4293,
|
| 1454 |
+
"eval_samples_per_second": 31.835,
|
| 1455 |
+
"eval_steps_per_second": 0.996,
|
| 1456 |
+
"eval_wer": 0.32594247416922645,
|
| 1457 |
+
"step": 21000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 99.53,
|
| 1461 |
+
"learning_rate": 4.558139534883721e-07,
|
| 1462 |
+
"loss": 0.7818,
|
| 1463 |
+
"step": 21100
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 100.0,
|
| 1467 |
+
"learning_rate": 4.883720930232558e-08,
|
| 1468 |
+
"loss": 0.7716,
|
| 1469 |
+
"step": 21200
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 100.0,
|
| 1473 |
+
"step": 21200,
|
| 1474 |
+
"total_flos": 7.799992701307658e+19,
|
| 1475 |
+
"train_loss": 1.235663890838623,
|
| 1476 |
+
"train_runtime": 30132.3883,
|
| 1477 |
+
"train_samples_per_second": 22.464,
|
| 1478 |
+
"train_steps_per_second": 0.704
|
| 1479 |
}
|
| 1480 |
],
|
| 1481 |
+
"max_steps": 21200,
|
| 1482 |
+
"num_train_epochs": 100,
|
| 1483 |
+
"total_flos": 7.799992701307658e+19,
|
| 1484 |
"trial_name": null,
|
| 1485 |
"trial_params": null
|
| 1486 |
}
|