Commit ·
a49566a
1
Parent(s): ad84922
End of training
Browse files- README.md +6 -3
- all_results.json +9 -9
- eval_results.json +5 -5
- train_results.json +4 -4
- trainer_state.json +304 -304
README.md
CHANGED
|
@@ -1,5 +1,8 @@
|
|
| 1 |
---
|
|
|
|
| 2 |
tags:
|
|
|
|
|
|
|
| 3 |
- generated_from_trainer
|
| 4 |
datasets:
|
| 5 |
- timit_asr
|
|
@@ -12,11 +15,11 @@ model-index:
|
|
| 12 |
name: Automatic Speech Recognition
|
| 13 |
type: automatic-speech-recognition
|
| 14 |
dataset:
|
| 15 |
-
name:
|
| 16 |
type: timit_asr
|
| 17 |
config: clean
|
| 18 |
split: test
|
| 19 |
-
args:
|
| 20 |
metrics:
|
| 21 |
- name: Wer
|
| 22 |
type: wer
|
|
@@ -28,7 +31,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 28 |
|
| 29 |
# wav2vec2-base-hyperVQ-timit-fine-tuned
|
| 30 |
|
| 31 |
-
This model
|
| 32 |
It achieves the following results on the evaluation set:
|
| 33 |
- Loss: 3.3628
|
| 34 |
- Wer: 0.9993
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model: wav2vec2-pretrained-base-hyperVQ
|
| 3 |
tags:
|
| 4 |
+
- automatic-speech-recognition
|
| 5 |
+
- timit_asr
|
| 6 |
- generated_from_trainer
|
| 7 |
datasets:
|
| 8 |
- timit_asr
|
|
|
|
| 15 |
name: Automatic Speech Recognition
|
| 16 |
type: automatic-speech-recognition
|
| 17 |
dataset:
|
| 18 |
+
name: TIMIT_ASR - NA
|
| 19 |
type: timit_asr
|
| 20 |
config: clean
|
| 21 |
split: test
|
| 22 |
+
args: 'Config: na, Training split: train, Eval split: test'
|
| 23 |
metrics:
|
| 24 |
- name: Wer
|
| 25 |
type: wer
|
|
|
|
| 31 |
|
| 32 |
# wav2vec2-base-hyperVQ-timit-fine-tuned
|
| 33 |
|
| 34 |
+
This model is a fine-tuned version of [wav2vec2-pretrained-base-hyperVQ](https://huggingface.co/wav2vec2-pretrained-base-hyperVQ) on the TIMIT_ASR - NA dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
- Loss: 3.3628
|
| 37 |
- Wer: 0.9993
|
all_results.json
CHANGED
|
@@ -1,14 +1,14 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
-
"eval_loss":
|
| 4 |
-
"eval_runtime": 8.
|
| 5 |
"eval_samples": 1680,
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 6.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
-
"train_loss":
|
| 10 |
-
"train_runtime":
|
| 11 |
"train_samples": 4620,
|
| 12 |
-
"train_samples_per_second":
|
| 13 |
-
"train_steps_per_second":
|
| 14 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
+
"eval_loss": 3.362812042236328,
|
| 4 |
+
"eval_runtime": 8.2056,
|
| 5 |
"eval_samples": 1680,
|
| 6 |
+
"eval_samples_per_second": 204.737,
|
| 7 |
+
"eval_steps_per_second": 6.459,
|
| 8 |
+
"eval_wer": 0.9993108676176694,
|
| 9 |
+
"train_loss": 3.8477164847275307,
|
| 10 |
+
"train_runtime": 343.8146,
|
| 11 |
"train_samples": 4620,
|
| 12 |
+
"train_samples_per_second": 268.749,
|
| 13 |
+
"train_steps_per_second": 8.435
|
| 14 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
-
"eval_loss":
|
| 4 |
-
"eval_runtime": 8.
|
| 5 |
"eval_samples": 1680,
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 6.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
+
"eval_loss": 3.362812042236328,
|
| 4 |
+
"eval_runtime": 8.2056,
|
| 5 |
"eval_samples": 1680,
|
| 6 |
+
"eval_samples_per_second": 204.737,
|
| 7 |
+
"eval_steps_per_second": 6.459,
|
| 8 |
+
"eval_wer": 0.9993108676176694
|
| 9 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 4620,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.8477164847275307,
|
| 4 |
+
"train_runtime": 343.8146,
|
| 5 |
"train_samples": 4620,
|
| 6 |
+
"train_samples_per_second": 268.749,
|
| 7 |
+
"train_steps_per_second": 8.435
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -11,1769 +11,1769 @@
|
|
| 11 |
{
|
| 12 |
"epoch": 0.07,
|
| 13 |
"learning_rate": 1.0000000000000002e-06,
|
| 14 |
-
"loss": 8.
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.14,
|
| 19 |
"learning_rate": 2.0000000000000003e-06,
|
| 20 |
-
"loss": 8.
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.21,
|
| 25 |
"learning_rate": 3e-06,
|
| 26 |
-
"loss": 7.
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.28,
|
| 31 |
"learning_rate": 4.000000000000001e-06,
|
| 32 |
-
"loss": 8.
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.34,
|
| 37 |
"learning_rate": 5e-06,
|
| 38 |
-
"loss": 8.
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.41,
|
| 43 |
"learning_rate": 6e-06,
|
| 44 |
-
"loss":
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.48,
|
| 49 |
"learning_rate": 7.000000000000001e-06,
|
| 50 |
-
"loss": 7.
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.55,
|
| 55 |
"learning_rate": 8.000000000000001e-06,
|
| 56 |
-
"loss":
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.62,
|
| 61 |
"learning_rate": 9e-06,
|
| 62 |
-
"loss":
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.69,
|
| 67 |
"learning_rate": 1e-05,
|
| 68 |
-
"loss":
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.76,
|
| 73 |
"learning_rate": 1.1000000000000001e-05,
|
| 74 |
-
"loss":
|
| 75 |
"step": 110
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.83,
|
| 79 |
"learning_rate": 1.2e-05,
|
| 80 |
-
"loss":
|
| 81 |
"step": 120
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.9,
|
| 85 |
"learning_rate": 1.3000000000000001e-05,
|
| 86 |
-
"loss":
|
| 87 |
"step": 130
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.97,
|
| 91 |
"learning_rate": 1.4000000000000001e-05,
|
| 92 |
-
"loss":
|
| 93 |
"step": 140
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 1.03,
|
| 97 |
"learning_rate": 1.5e-05,
|
| 98 |
-
"loss":
|
| 99 |
"step": 150
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 1.1,
|
| 103 |
"learning_rate": 1.6000000000000003e-05,
|
| 104 |
-
"loss":
|
| 105 |
"step": 160
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 1.17,
|
| 109 |
"learning_rate": 1.7000000000000003e-05,
|
| 110 |
-
"loss":
|
| 111 |
"step": 170
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 1.24,
|
| 115 |
"learning_rate": 1.8e-05,
|
| 116 |
-
"loss":
|
| 117 |
"step": 180
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 1.31,
|
| 121 |
"learning_rate": 1.9e-05,
|
| 122 |
-
"loss":
|
| 123 |
"step": 190
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 1.38,
|
| 127 |
"learning_rate": 2e-05,
|
| 128 |
-
"loss":
|
| 129 |
"step": 200
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 1.45,
|
| 133 |
"learning_rate": 2.1e-05,
|
| 134 |
-
"loss":
|
| 135 |
"step": 210
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 1.52,
|
| 139 |
"learning_rate": 2.2000000000000003e-05,
|
| 140 |
-
"loss":
|
| 141 |
"step": 220
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 1.59,
|
| 145 |
"learning_rate": 2.3000000000000003e-05,
|
| 146 |
-
"loss":
|
| 147 |
"step": 230
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 1.66,
|
| 151 |
"learning_rate": 2.4e-05,
|
| 152 |
-
"loss":
|
| 153 |
"step": 240
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 1.72,
|
| 157 |
"learning_rate": 2.5e-05,
|
| 158 |
-
"loss":
|
| 159 |
"step": 250
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 1.79,
|
| 163 |
"learning_rate": 2.6000000000000002e-05,
|
| 164 |
-
"loss":
|
| 165 |
"step": 260
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 1.86,
|
| 169 |
"learning_rate": 2.7000000000000002e-05,
|
| 170 |
-
"loss":
|
| 171 |
"step": 270
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 1.93,
|
| 175 |
"learning_rate": 2.8000000000000003e-05,
|
| 176 |
-
"loss":
|
| 177 |
"step": 280
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 2.0,
|
| 181 |
"learning_rate": 2.9e-05,
|
| 182 |
-
"loss":
|
| 183 |
"step": 290
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 2.07,
|
| 187 |
"learning_rate": 3e-05,
|
| 188 |
-
"loss":
|
| 189 |
"step": 300
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 2.14,
|
| 193 |
"learning_rate": 3.1e-05,
|
| 194 |
-
"loss":
|
| 195 |
"step": 310
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 2.21,
|
| 199 |
"learning_rate": 3.2000000000000005e-05,
|
| 200 |
-
"loss":
|
| 201 |
"step": 320
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 2.28,
|
| 205 |
"learning_rate": 3.3e-05,
|
| 206 |
-
"loss":
|
| 207 |
"step": 330
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 2.34,
|
| 211 |
"learning_rate": 3.4000000000000007e-05,
|
| 212 |
-
"loss":
|
| 213 |
"step": 340
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 2.41,
|
| 217 |
"learning_rate": 3.5e-05,
|
| 218 |
-
"loss":
|
| 219 |
"step": 350
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 2.48,
|
| 223 |
"learning_rate": 3.6e-05,
|
| 224 |
-
"loss":
|
| 225 |
"step": 360
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 2.55,
|
| 229 |
"learning_rate": 3.7e-05,
|
| 230 |
-
"loss":
|
| 231 |
"step": 370
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 2.62,
|
| 235 |
"learning_rate": 3.8e-05,
|
| 236 |
-
"loss":
|
| 237 |
"step": 380
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 2.69,
|
| 241 |
"learning_rate": 3.9000000000000006e-05,
|
| 242 |
-
"loss":
|
| 243 |
"step": 390
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 2.76,
|
| 247 |
"learning_rate": 4e-05,
|
| 248 |
-
"loss":
|
| 249 |
"step": 400
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 2.83,
|
| 253 |
"learning_rate": 4.1e-05,
|
| 254 |
-
"loss":
|
| 255 |
"step": 410
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 2.9,
|
| 259 |
"learning_rate": 4.2e-05,
|
| 260 |
-
"loss":
|
| 261 |
"step": 420
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 2.97,
|
| 265 |
"learning_rate": 4.3e-05,
|
| 266 |
-
"loss":
|
| 267 |
"step": 430
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 3.03,
|
| 271 |
"learning_rate": 4.4000000000000006e-05,
|
| 272 |
-
"loss":
|
| 273 |
"step": 440
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 3.1,
|
| 277 |
"learning_rate": 4.5e-05,
|
| 278 |
-
"loss":
|
| 279 |
"step": 450
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 3.17,
|
| 283 |
"learning_rate": 4.600000000000001e-05,
|
| 284 |
-
"loss":
|
| 285 |
"step": 460
|
| 286 |
},
|
| 287 |
{
|
| 288 |
"epoch": 3.24,
|
| 289 |
"learning_rate": 4.7e-05,
|
| 290 |
-
"loss":
|
| 291 |
"step": 470
|
| 292 |
},
|
| 293 |
{
|
| 294 |
"epoch": 3.31,
|
| 295 |
"learning_rate": 4.8e-05,
|
| 296 |
-
"loss":
|
| 297 |
"step": 480
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 3.38,
|
| 301 |
"learning_rate": 4.9e-05,
|
| 302 |
-
"loss":
|
| 303 |
"step": 490
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"epoch": 3.45,
|
| 307 |
"learning_rate": 5e-05,
|
| 308 |
-
"loss":
|
| 309 |
"step": 500
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 3.52,
|
| 313 |
"learning_rate": 5.1000000000000006e-05,
|
| 314 |
-
"loss":
|
| 315 |
"step": 510
|
| 316 |
},
|
| 317 |
{
|
| 318 |
"epoch": 3.59,
|
| 319 |
"learning_rate": 5.2000000000000004e-05,
|
| 320 |
-
"loss":
|
| 321 |
"step": 520
|
| 322 |
},
|
| 323 |
{
|
| 324 |
"epoch": 3.66,
|
| 325 |
"learning_rate": 5.300000000000001e-05,
|
| 326 |
-
"loss":
|
| 327 |
"step": 530
|
| 328 |
},
|
| 329 |
{
|
| 330 |
"epoch": 3.72,
|
| 331 |
"learning_rate": 5.4000000000000005e-05,
|
| 332 |
-
"loss":
|
| 333 |
"step": 540
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 3.79,
|
| 337 |
"learning_rate": 5.500000000000001e-05,
|
| 338 |
-
"loss":
|
| 339 |
"step": 550
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 3.86,
|
| 343 |
"learning_rate": 5.6000000000000006e-05,
|
| 344 |
-
"loss":
|
| 345 |
"step": 560
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"epoch": 3.93,
|
| 349 |
"learning_rate": 5.6999999999999996e-05,
|
| 350 |
-
"loss":
|
| 351 |
"step": 570
|
| 352 |
},
|
| 353 |
{
|
| 354 |
"epoch": 4.0,
|
| 355 |
"learning_rate": 5.8e-05,
|
| 356 |
-
"loss":
|
| 357 |
"step": 580
|
| 358 |
},
|
| 359 |
{
|
| 360 |
"epoch": 4.07,
|
| 361 |
"learning_rate": 5.9e-05,
|
| 362 |
-
"loss":
|
| 363 |
"step": 590
|
| 364 |
},
|
| 365 |
{
|
| 366 |
"epoch": 4.14,
|
| 367 |
"learning_rate": 6e-05,
|
| 368 |
-
"loss":
|
| 369 |
"step": 600
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 4.21,
|
| 373 |
"learning_rate": 6.1e-05,
|
| 374 |
-
"loss":
|
| 375 |
"step": 610
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 4.28,
|
| 379 |
"learning_rate": 6.2e-05,
|
| 380 |
-
"loss":
|
| 381 |
"step": 620
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 4.34,
|
| 385 |
"learning_rate": 6.3e-05,
|
| 386 |
-
"loss":
|
| 387 |
"step": 630
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"epoch": 4.41,
|
| 391 |
"learning_rate": 6.400000000000001e-05,
|
| 392 |
-
"loss":
|
| 393 |
"step": 640
|
| 394 |
},
|
| 395 |
{
|
| 396 |
"epoch": 4.48,
|
| 397 |
"learning_rate": 6.500000000000001e-05,
|
| 398 |
-
"loss":
|
| 399 |
"step": 650
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 4.55,
|
| 403 |
"learning_rate": 6.6e-05,
|
| 404 |
-
"loss":
|
| 405 |
"step": 660
|
| 406 |
},
|
| 407 |
{
|
| 408 |
"epoch": 4.62,
|
| 409 |
"learning_rate": 6.7e-05,
|
| 410 |
-
"loss":
|
| 411 |
"step": 670
|
| 412 |
},
|
| 413 |
{
|
| 414 |
"epoch": 4.69,
|
| 415 |
"learning_rate": 6.800000000000001e-05,
|
| 416 |
-
"loss":
|
| 417 |
"step": 680
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 4.76,
|
| 421 |
"learning_rate": 6.9e-05,
|
| 422 |
-
"loss":
|
| 423 |
"step": 690
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 4.83,
|
| 427 |
"learning_rate": 7e-05,
|
| 428 |
-
"loss":
|
| 429 |
"step": 700
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 4.9,
|
| 433 |
"learning_rate": 7.1e-05,
|
| 434 |
-
"loss":
|
| 435 |
"step": 710
|
| 436 |
},
|
| 437 |
{
|
| 438 |
"epoch": 4.97,
|
| 439 |
"learning_rate": 7.2e-05,
|
| 440 |
-
"loss":
|
| 441 |
"step": 720
|
| 442 |
},
|
| 443 |
{
|
| 444 |
"epoch": 5.03,
|
| 445 |
"learning_rate": 7.3e-05,
|
| 446 |
-
"loss":
|
| 447 |
"step": 730
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"epoch": 5.1,
|
| 451 |
"learning_rate": 7.4e-05,
|
| 452 |
-
"loss":
|
| 453 |
"step": 740
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 5.17,
|
| 457 |
"learning_rate": 7.500000000000001e-05,
|
| 458 |
-
"loss":
|
| 459 |
"step": 750
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 5.24,
|
| 463 |
"learning_rate": 7.6e-05,
|
| 464 |
-
"loss":
|
| 465 |
"step": 760
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 5.31,
|
| 469 |
"learning_rate": 7.7e-05,
|
| 470 |
-
"loss":
|
| 471 |
"step": 770
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"epoch": 5.38,
|
| 475 |
"learning_rate": 7.800000000000001e-05,
|
| 476 |
-
"loss":
|
| 477 |
"step": 780
|
| 478 |
},
|
| 479 |
{
|
| 480 |
"epoch": 5.45,
|
| 481 |
"learning_rate": 7.900000000000001e-05,
|
| 482 |
-
"loss":
|
| 483 |
"step": 790
|
| 484 |
},
|
| 485 |
{
|
| 486 |
"epoch": 5.52,
|
| 487 |
"learning_rate": 8e-05,
|
| 488 |
-
"loss":
|
| 489 |
"step": 800
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 5.59,
|
| 493 |
"learning_rate": 8.1e-05,
|
| 494 |
-
"loss":
|
| 495 |
"step": 810
|
| 496 |
},
|
| 497 |
{
|
| 498 |
"epoch": 5.66,
|
| 499 |
"learning_rate": 8.2e-05,
|
| 500 |
-
"loss":
|
| 501 |
"step": 820
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 5.72,
|
| 505 |
"learning_rate": 8.3e-05,
|
| 506 |
-
"loss":
|
| 507 |
"step": 830
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 5.79,
|
| 511 |
"learning_rate": 8.4e-05,
|
| 512 |
-
"loss":
|
| 513 |
"step": 840
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 5.86,
|
| 517 |
"learning_rate": 8.5e-05,
|
| 518 |
-
"loss":
|
| 519 |
"step": 850
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 5.93,
|
| 523 |
"learning_rate": 8.6e-05,
|
| 524 |
-
"loss":
|
| 525 |
"step": 860
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 6.0,
|
| 529 |
"learning_rate": 8.7e-05,
|
| 530 |
-
"loss":
|
| 531 |
"step": 870
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 6.07,
|
| 535 |
"learning_rate": 8.800000000000001e-05,
|
| 536 |
-
"loss":
|
| 537 |
"step": 880
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 6.14,
|
| 541 |
"learning_rate": 8.900000000000001e-05,
|
| 542 |
-
"loss":
|
| 543 |
"step": 890
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 6.21,
|
| 547 |
"learning_rate": 9e-05,
|
| 548 |
-
"loss":
|
| 549 |
"step": 900
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 6.28,
|
| 553 |
"learning_rate": 9.1e-05,
|
| 554 |
-
"loss":
|
| 555 |
"step": 910
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 6.34,
|
| 559 |
"learning_rate": 9.200000000000001e-05,
|
| 560 |
-
"loss":
|
| 561 |
"step": 920
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 6.41,
|
| 565 |
"learning_rate": 9.300000000000001e-05,
|
| 566 |
-
"loss":
|
| 567 |
"step": 930
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 6.48,
|
| 571 |
"learning_rate": 9.4e-05,
|
| 572 |
-
"loss":
|
| 573 |
"step": 940
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 6.55,
|
| 577 |
"learning_rate": 9.5e-05,
|
| 578 |
-
"loss":
|
| 579 |
"step": 950
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 6.62,
|
| 583 |
"learning_rate": 9.6e-05,
|
| 584 |
-
"loss":
|
| 585 |
"step": 960
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 6.69,
|
| 589 |
"learning_rate": 9.7e-05,
|
| 590 |
-
"loss":
|
| 591 |
"step": 970
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 6.76,
|
| 595 |
"learning_rate": 9.8e-05,
|
| 596 |
-
"loss":
|
| 597 |
"step": 980
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 6.83,
|
| 601 |
"learning_rate": 9.900000000000001e-05,
|
| 602 |
-
"loss":
|
| 603 |
"step": 990
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 6.9,
|
| 607 |
"learning_rate": 0.0001,
|
| 608 |
-
"loss":
|
| 609 |
"step": 1000
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 6.97,
|
| 613 |
"learning_rate": 9.947368421052632e-05,
|
| 614 |
-
"loss":
|
| 615 |
"step": 1010
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 7.03,
|
| 619 |
"learning_rate": 9.894736842105263e-05,
|
| 620 |
-
"loss":
|
| 621 |
"step": 1020
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 7.1,
|
| 625 |
"learning_rate": 9.842105263157894e-05,
|
| 626 |
-
"loss":
|
| 627 |
"step": 1030
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 7.17,
|
| 631 |
"learning_rate": 9.789473684210527e-05,
|
| 632 |
-
"loss":
|
| 633 |
"step": 1040
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 7.24,
|
| 637 |
"learning_rate": 9.736842105263158e-05,
|
| 638 |
-
"loss":
|
| 639 |
"step": 1050
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 7.31,
|
| 643 |
"learning_rate": 9.68421052631579e-05,
|
| 644 |
-
"loss":
|
| 645 |
"step": 1060
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 7.38,
|
| 649 |
"learning_rate": 9.631578947368421e-05,
|
| 650 |
-
"loss":
|
| 651 |
"step": 1070
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 7.45,
|
| 655 |
"learning_rate": 9.578947368421052e-05,
|
| 656 |
-
"loss":
|
| 657 |
"step": 1080
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 7.52,
|
| 661 |
"learning_rate": 9.526315789473685e-05,
|
| 662 |
-
"loss":
|
| 663 |
"step": 1090
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 7.59,
|
| 667 |
"learning_rate": 9.473684210526316e-05,
|
| 668 |
-
"loss":
|
| 669 |
"step": 1100
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 7.66,
|
| 673 |
"learning_rate": 9.421052631578949e-05,
|
| 674 |
-
"loss":
|
| 675 |
"step": 1110
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 7.72,
|
| 679 |
"learning_rate": 9.36842105263158e-05,
|
| 680 |
-
"loss":
|
| 681 |
"step": 1120
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 7.79,
|
| 685 |
"learning_rate": 9.315789473684211e-05,
|
| 686 |
-
"loss":
|
| 687 |
"step": 1130
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 7.86,
|
| 691 |
"learning_rate": 9.263157894736843e-05,
|
| 692 |
-
"loss":
|
| 693 |
"step": 1140
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 7.93,
|
| 697 |
"learning_rate": 9.210526315789474e-05,
|
| 698 |
-
"loss":
|
| 699 |
"step": 1150
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 8.0,
|
| 703 |
"learning_rate": 9.157894736842105e-05,
|
| 704 |
-
"loss":
|
| 705 |
"step": 1160
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 8.07,
|
| 709 |
"learning_rate": 9.105263157894738e-05,
|
| 710 |
-
"loss":
|
| 711 |
"step": 1170
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 8.14,
|
| 715 |
"learning_rate": 9.052631578947369e-05,
|
| 716 |
-
"loss":
|
| 717 |
"step": 1180
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 8.21,
|
| 721 |
"learning_rate": 9e-05,
|
| 722 |
-
"loss":
|
| 723 |
"step": 1190
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 8.28,
|
| 727 |
"learning_rate": 8.947368421052632e-05,
|
| 728 |
-
"loss":
|
| 729 |
"step": 1200
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 8.34,
|
| 733 |
"learning_rate": 8.894736842105263e-05,
|
| 734 |
-
"loss":
|
| 735 |
"step": 1210
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 8.41,
|
| 739 |
"learning_rate": 8.842105263157894e-05,
|
| 740 |
-
"loss":
|
| 741 |
"step": 1220
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 8.48,
|
| 745 |
"learning_rate": 8.789473684210526e-05,
|
| 746 |
-
"loss":
|
| 747 |
"step": 1230
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 8.55,
|
| 751 |
"learning_rate": 8.736842105263158e-05,
|
| 752 |
-
"loss":
|
| 753 |
"step": 1240
|
| 754 |
},
|
| 755 |
{
|
| 756 |
"epoch": 8.62,
|
| 757 |
"learning_rate": 8.68421052631579e-05,
|
| 758 |
-
"loss":
|
| 759 |
"step": 1250
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 8.69,
|
| 763 |
"learning_rate": 8.631578947368421e-05,
|
| 764 |
-
"loss":
|
| 765 |
"step": 1260
|
| 766 |
},
|
| 767 |
{
|
| 768 |
"epoch": 8.76,
|
| 769 |
"learning_rate": 8.578947368421054e-05,
|
| 770 |
-
"loss":
|
| 771 |
"step": 1270
|
| 772 |
},
|
| 773 |
{
|
| 774 |
"epoch": 8.83,
|
| 775 |
"learning_rate": 8.526315789473685e-05,
|
| 776 |
-
"loss":
|
| 777 |
"step": 1280
|
| 778 |
},
|
| 779 |
{
|
| 780 |
"epoch": 8.9,
|
| 781 |
"learning_rate": 8.473684210526316e-05,
|
| 782 |
-
"loss":
|
| 783 |
"step": 1290
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 8.97,
|
| 787 |
"learning_rate": 8.421052631578948e-05,
|
| 788 |
-
"loss":
|
| 789 |
"step": 1300
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 9.03,
|
| 793 |
"learning_rate": 8.36842105263158e-05,
|
| 794 |
-
"loss":
|
| 795 |
"step": 1310
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 9.1,
|
| 799 |
"learning_rate": 8.315789473684212e-05,
|
| 800 |
-
"loss":
|
| 801 |
"step": 1320
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 9.17,
|
| 805 |
"learning_rate": 8.263157894736843e-05,
|
| 806 |
-
"loss":
|
| 807 |
"step": 1330
|
| 808 |
},
|
| 809 |
{
|
| 810 |
"epoch": 9.24,
|
| 811 |
"learning_rate": 8.210526315789474e-05,
|
| 812 |
-
"loss":
|
| 813 |
"step": 1340
|
| 814 |
},
|
| 815 |
{
|
| 816 |
"epoch": 9.31,
|
| 817 |
"learning_rate": 8.157894736842105e-05,
|
| 818 |
-
"loss":
|
| 819 |
"step": 1350
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 9.38,
|
| 823 |
"learning_rate": 8.105263157894737e-05,
|
| 824 |
-
"loss":
|
| 825 |
"step": 1360
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 9.45,
|
| 829 |
"learning_rate": 8.052631578947368e-05,
|
| 830 |
-
"loss":
|
| 831 |
"step": 1370
|
| 832 |
},
|
| 833 |
{
|
| 834 |
"epoch": 9.52,
|
| 835 |
"learning_rate": 8e-05,
|
| 836 |
-
"loss":
|
| 837 |
"step": 1380
|
| 838 |
},
|
| 839 |
{
|
| 840 |
"epoch": 9.59,
|
| 841 |
"learning_rate": 7.947368421052632e-05,
|
| 842 |
-
"loss":
|
| 843 |
"step": 1390
|
| 844 |
},
|
| 845 |
{
|
| 846 |
"epoch": 9.66,
|
| 847 |
"learning_rate": 7.894736842105263e-05,
|
| 848 |
-
"loss":
|
| 849 |
"step": 1400
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 9.72,
|
| 853 |
"learning_rate": 7.842105263157895e-05,
|
| 854 |
-
"loss":
|
| 855 |
"step": 1410
|
| 856 |
},
|
| 857 |
{
|
| 858 |
"epoch": 9.79,
|
| 859 |
"learning_rate": 7.789473684210526e-05,
|
| 860 |
-
"loss":
|
| 861 |
"step": 1420
|
| 862 |
},
|
| 863 |
{
|
| 864 |
"epoch": 9.86,
|
| 865 |
"learning_rate": 7.736842105263159e-05,
|
| 866 |
-
"loss":
|
| 867 |
"step": 1430
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 9.93,
|
| 871 |
"learning_rate": 7.68421052631579e-05,
|
| 872 |
-
"loss":
|
| 873 |
"step": 1440
|
| 874 |
},
|
| 875 |
{
|
| 876 |
"epoch": 10.0,
|
| 877 |
"learning_rate": 7.631578947368422e-05,
|
| 878 |
-
"loss":
|
| 879 |
"step": 1450
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 10.0,
|
| 883 |
-
"eval_loss":
|
| 884 |
-
"eval_runtime":
|
| 885 |
-
"eval_samples_per_second":
|
| 886 |
-
"eval_steps_per_second":
|
| 887 |
-
"eval_wer":
|
| 888 |
"step": 1450
|
| 889 |
},
|
| 890 |
{
|
| 891 |
"epoch": 10.07,
|
| 892 |
"learning_rate": 7.578947368421054e-05,
|
| 893 |
-
"loss":
|
| 894 |
"step": 1460
|
| 895 |
},
|
| 896 |
{
|
| 897 |
"epoch": 10.14,
|
| 898 |
"learning_rate": 7.526315789473685e-05,
|
| 899 |
-
"loss":
|
| 900 |
"step": 1470
|
| 901 |
},
|
| 902 |
{
|
| 903 |
"epoch": 10.21,
|
| 904 |
"learning_rate": 7.473684210526316e-05,
|
| 905 |
-
"loss":
|
| 906 |
"step": 1480
|
| 907 |
},
|
| 908 |
{
|
| 909 |
"epoch": 10.28,
|
| 910 |
"learning_rate": 7.421052631578948e-05,
|
| 911 |
-
"loss":
|
| 912 |
"step": 1490
|
| 913 |
},
|
| 914 |
{
|
| 915 |
"epoch": 10.34,
|
| 916 |
"learning_rate": 7.368421052631579e-05,
|
| 917 |
-
"loss":
|
| 918 |
"step": 1500
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 10.41,
|
| 922 |
"learning_rate": 7.315789473684212e-05,
|
| 923 |
-
"loss":
|
| 924 |
"step": 1510
|
| 925 |
},
|
| 926 |
{
|
| 927 |
"epoch": 10.48,
|
| 928 |
"learning_rate": 7.263157894736843e-05,
|
| 929 |
-
"loss":
|
| 930 |
"step": 1520
|
| 931 |
},
|
| 932 |
{
|
| 933 |
"epoch": 10.55,
|
| 934 |
"learning_rate": 7.210526315789474e-05,
|
| 935 |
-
"loss":
|
| 936 |
"step": 1530
|
| 937 |
},
|
| 938 |
{
|
| 939 |
"epoch": 10.62,
|
| 940 |
"learning_rate": 7.157894736842105e-05,
|
| 941 |
-
"loss":
|
| 942 |
"step": 1540
|
| 943 |
},
|
| 944 |
{
|
| 945 |
"epoch": 10.69,
|
| 946 |
"learning_rate": 7.105263157894737e-05,
|
| 947 |
-
"loss":
|
| 948 |
"step": 1550
|
| 949 |
},
|
| 950 |
{
|
| 951 |
"epoch": 10.76,
|
| 952 |
"learning_rate": 7.052631578947368e-05,
|
| 953 |
-
"loss":
|
| 954 |
"step": 1560
|
| 955 |
},
|
| 956 |
{
|
| 957 |
"epoch": 10.83,
|
| 958 |
"learning_rate": 7e-05,
|
| 959 |
-
"loss":
|
| 960 |
"step": 1570
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 10.9,
|
| 964 |
"learning_rate": 6.947368421052632e-05,
|
| 965 |
-
"loss":
|
| 966 |
"step": 1580
|
| 967 |
},
|
| 968 |
{
|
| 969 |
"epoch": 10.97,
|
| 970 |
"learning_rate": 6.894736842105263e-05,
|
| 971 |
-
"loss":
|
| 972 |
"step": 1590
|
| 973 |
},
|
| 974 |
{
|
| 975 |
"epoch": 11.03,
|
| 976 |
"learning_rate": 6.842105263157895e-05,
|
| 977 |
-
"loss":
|
| 978 |
"step": 1600
|
| 979 |
},
|
| 980 |
{
|
| 981 |
"epoch": 11.1,
|
| 982 |
"learning_rate": 6.789473684210527e-05,
|
| 983 |
-
"loss":
|
| 984 |
"step": 1610
|
| 985 |
},
|
| 986 |
{
|
| 987 |
"epoch": 11.17,
|
| 988 |
"learning_rate": 6.736842105263159e-05,
|
| 989 |
-
"loss":
|
| 990 |
"step": 1620
|
| 991 |
},
|
| 992 |
{
|
| 993 |
"epoch": 11.24,
|
| 994 |
"learning_rate": 6.68421052631579e-05,
|
| 995 |
-
"loss":
|
| 996 |
"step": 1630
|
| 997 |
},
|
| 998 |
{
|
| 999 |
"epoch": 11.31,
|
| 1000 |
"learning_rate": 6.631578947368421e-05,
|
| 1001 |
-
"loss":
|
| 1002 |
"step": 1640
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 11.38,
|
| 1006 |
"learning_rate": 6.578947368421054e-05,
|
| 1007 |
-
"loss":
|
| 1008 |
"step": 1650
|
| 1009 |
},
|
| 1010 |
{
|
| 1011 |
"epoch": 11.45,
|
| 1012 |
"learning_rate": 6.526315789473685e-05,
|
| 1013 |
-
"loss":
|
| 1014 |
"step": 1660
|
| 1015 |
},
|
| 1016 |
{
|
| 1017 |
"epoch": 11.52,
|
| 1018 |
"learning_rate": 6.473684210526316e-05,
|
| 1019 |
-
"loss":
|
| 1020 |
"step": 1670
|
| 1021 |
},
|
| 1022 |
{
|
| 1023 |
"epoch": 11.59,
|
| 1024 |
"learning_rate": 6.421052631578948e-05,
|
| 1025 |
-
"loss":
|
| 1026 |
"step": 1680
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
"epoch": 11.66,
|
| 1030 |
"learning_rate": 6.368421052631579e-05,
|
| 1031 |
-
"loss":
|
| 1032 |
"step": 1690
|
| 1033 |
},
|
| 1034 |
{
|
| 1035 |
"epoch": 11.72,
|
| 1036 |
"learning_rate": 6.31578947368421e-05,
|
| 1037 |
-
"loss":
|
| 1038 |
"step": 1700
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
"epoch": 11.79,
|
| 1042 |
"learning_rate": 6.263157894736842e-05,
|
| 1043 |
-
"loss":
|
| 1044 |
"step": 1710
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 11.86,
|
| 1048 |
"learning_rate": 6.210526315789474e-05,
|
| 1049 |
-
"loss":
|
| 1050 |
"step": 1720
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
"epoch": 11.93,
|
| 1054 |
"learning_rate": 6.157894736842106e-05,
|
| 1055 |
-
"loss":
|
| 1056 |
"step": 1730
|
| 1057 |
},
|
| 1058 |
{
|
| 1059 |
"epoch": 12.0,
|
| 1060 |
"learning_rate": 6.105263157894737e-05,
|
| 1061 |
-
"loss":
|
| 1062 |
"step": 1740
|
| 1063 |
},
|
| 1064 |
{
|
| 1065 |
"epoch": 12.07,
|
| 1066 |
"learning_rate": 6.052631578947369e-05,
|
| 1067 |
-
"loss":
|
| 1068 |
"step": 1750
|
| 1069 |
},
|
| 1070 |
{
|
| 1071 |
"epoch": 12.14,
|
| 1072 |
"learning_rate": 6e-05,
|
| 1073 |
-
"loss":
|
| 1074 |
"step": 1760
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
"epoch": 12.21,
|
| 1078 |
"learning_rate": 5.9473684210526315e-05,
|
| 1079 |
-
"loss":
|
| 1080 |
"step": 1770
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
"epoch": 12.28,
|
| 1084 |
"learning_rate": 5.894736842105263e-05,
|
| 1085 |
-
"loss":
|
| 1086 |
"step": 1780
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 12.34,
|
| 1090 |
"learning_rate": 5.8421052631578954e-05,
|
| 1091 |
-
"loss":
|
| 1092 |
"step": 1790
|
| 1093 |
},
|
| 1094 |
{
|
| 1095 |
"epoch": 12.41,
|
| 1096 |
"learning_rate": 5.789473684210527e-05,
|
| 1097 |
-
"loss":
|
| 1098 |
"step": 1800
|
| 1099 |
},
|
| 1100 |
{
|
| 1101 |
"epoch": 12.48,
|
| 1102 |
"learning_rate": 5.736842105263158e-05,
|
| 1103 |
-
"loss":
|
| 1104 |
"step": 1810
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
"epoch": 12.55,
|
| 1108 |
"learning_rate": 5.68421052631579e-05,
|
| 1109 |
-
"loss":
|
| 1110 |
"step": 1820
|
| 1111 |
},
|
| 1112 |
{
|
| 1113 |
"epoch": 12.62,
|
| 1114 |
"learning_rate": 5.631578947368421e-05,
|
| 1115 |
-
"loss":
|
| 1116 |
"step": 1830
|
| 1117 |
},
|
| 1118 |
{
|
| 1119 |
"epoch": 12.69,
|
| 1120 |
"learning_rate": 5.5789473684210526e-05,
|
| 1121 |
-
"loss":
|
| 1122 |
"step": 1840
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
"epoch": 12.76,
|
| 1126 |
"learning_rate": 5.526315789473685e-05,
|
| 1127 |
-
"loss":
|
| 1128 |
"step": 1850
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 12.83,
|
| 1132 |
"learning_rate": 5.4736842105263165e-05,
|
| 1133 |
-
"loss":
|
| 1134 |
"step": 1860
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
"epoch": 12.9,
|
| 1138 |
"learning_rate": 5.421052631578948e-05,
|
| 1139 |
-
"loss":
|
| 1140 |
"step": 1870
|
| 1141 |
},
|
| 1142 |
{
|
| 1143 |
"epoch": 12.97,
|
| 1144 |
"learning_rate": 5.368421052631579e-05,
|
| 1145 |
-
"loss":
|
| 1146 |
"step": 1880
|
| 1147 |
},
|
| 1148 |
{
|
| 1149 |
"epoch": 13.03,
|
| 1150 |
"learning_rate": 5.3157894736842104e-05,
|
| 1151 |
-
"loss":
|
| 1152 |
"step": 1890
|
| 1153 |
},
|
| 1154 |
{
|
| 1155 |
"epoch": 13.1,
|
| 1156 |
"learning_rate": 5.2631578947368424e-05,
|
| 1157 |
-
"loss":
|
| 1158 |
"step": 1900
|
| 1159 |
},
|
| 1160 |
{
|
| 1161 |
"epoch": 13.17,
|
| 1162 |
"learning_rate": 5.210526315789474e-05,
|
| 1163 |
-
"loss":
|
| 1164 |
"step": 1910
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
"epoch": 13.24,
|
| 1168 |
"learning_rate": 5.157894736842106e-05,
|
| 1169 |
-
"loss":
|
| 1170 |
"step": 1920
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
"epoch": 13.31,
|
| 1174 |
"learning_rate": 5.1052631578947376e-05,
|
| 1175 |
-
"loss":
|
| 1176 |
"step": 1930
|
| 1177 |
},
|
| 1178 |
{
|
| 1179 |
"epoch": 13.38,
|
| 1180 |
"learning_rate": 5.052631578947369e-05,
|
| 1181 |
-
"loss":
|
| 1182 |
"step": 1940
|
| 1183 |
},
|
| 1184 |
{
|
| 1185 |
"epoch": 13.45,
|
| 1186 |
"learning_rate": 5e-05,
|
| 1187 |
-
"loss":
|
| 1188 |
"step": 1950
|
| 1189 |
},
|
| 1190 |
{
|
| 1191 |
"epoch": 13.52,
|
| 1192 |
"learning_rate": 4.9473684210526315e-05,
|
| 1193 |
-
"loss":
|
| 1194 |
"step": 1960
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
"epoch": 13.59,
|
| 1198 |
"learning_rate": 4.8947368421052635e-05,
|
| 1199 |
-
"loss":
|
| 1200 |
"step": 1970
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 13.66,
|
| 1204 |
"learning_rate": 4.842105263157895e-05,
|
| 1205 |
-
"loss":
|
| 1206 |
"step": 1980
|
| 1207 |
},
|
| 1208 |
{
|
| 1209 |
"epoch": 13.72,
|
| 1210 |
"learning_rate": 4.789473684210526e-05,
|
| 1211 |
-
"loss":
|
| 1212 |
"step": 1990
|
| 1213 |
},
|
| 1214 |
{
|
| 1215 |
"epoch": 13.79,
|
| 1216 |
"learning_rate": 4.736842105263158e-05,
|
| 1217 |
-
"loss":
|
| 1218 |
"step": 2000
|
| 1219 |
},
|
| 1220 |
{
|
| 1221 |
"epoch": 13.86,
|
| 1222 |
"learning_rate": 4.68421052631579e-05,
|
| 1223 |
-
"loss":
|
| 1224 |
"step": 2010
|
| 1225 |
},
|
| 1226 |
{
|
| 1227 |
"epoch": 13.93,
|
| 1228 |
"learning_rate": 4.6315789473684214e-05,
|
| 1229 |
-
"loss":
|
| 1230 |
"step": 2020
|
| 1231 |
},
|
| 1232 |
{
|
| 1233 |
"epoch": 14.0,
|
| 1234 |
"learning_rate": 4.5789473684210527e-05,
|
| 1235 |
-
"loss":
|
| 1236 |
"step": 2030
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
"epoch": 14.07,
|
| 1240 |
"learning_rate": 4.5263157894736846e-05,
|
| 1241 |
-
"loss":
|
| 1242 |
"step": 2040
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 14.14,
|
| 1246 |
"learning_rate": 4.473684210526316e-05,
|
| 1247 |
-
"loss":
|
| 1248 |
"step": 2050
|
| 1249 |
},
|
| 1250 |
{
|
| 1251 |
"epoch": 14.21,
|
| 1252 |
"learning_rate": 4.421052631578947e-05,
|
| 1253 |
-
"loss":
|
| 1254 |
"step": 2060
|
| 1255 |
},
|
| 1256 |
{
|
| 1257 |
"epoch": 14.28,
|
| 1258 |
"learning_rate": 4.368421052631579e-05,
|
| 1259 |
-
"loss":
|
| 1260 |
"step": 2070
|
| 1261 |
},
|
| 1262 |
{
|
| 1263 |
"epoch": 14.34,
|
| 1264 |
"learning_rate": 4.3157894736842105e-05,
|
| 1265 |
-
"loss":
|
| 1266 |
"step": 2080
|
| 1267 |
},
|
| 1268 |
{
|
| 1269 |
"epoch": 14.41,
|
| 1270 |
"learning_rate": 4.2631578947368425e-05,
|
| 1271 |
-
"loss":
|
| 1272 |
"step": 2090
|
| 1273 |
},
|
| 1274 |
{
|
| 1275 |
"epoch": 14.48,
|
| 1276 |
"learning_rate": 4.210526315789474e-05,
|
| 1277 |
-
"loss":
|
| 1278 |
"step": 2100
|
| 1279 |
},
|
| 1280 |
{
|
| 1281 |
"epoch": 14.55,
|
| 1282 |
"learning_rate": 4.157894736842106e-05,
|
| 1283 |
-
"loss":
|
| 1284 |
"step": 2110
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 14.62,
|
| 1288 |
"learning_rate": 4.105263157894737e-05,
|
| 1289 |
-
"loss":
|
| 1290 |
"step": 2120
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
"epoch": 14.69,
|
| 1294 |
"learning_rate": 4.0526315789473684e-05,
|
| 1295 |
-
"loss":
|
| 1296 |
"step": 2130
|
| 1297 |
},
|
| 1298 |
{
|
| 1299 |
"epoch": 14.76,
|
| 1300 |
"learning_rate": 4e-05,
|
| 1301 |
-
"loss":
|
| 1302 |
"step": 2140
|
| 1303 |
},
|
| 1304 |
{
|
| 1305 |
"epoch": 14.83,
|
| 1306 |
"learning_rate": 3.9473684210526316e-05,
|
| 1307 |
-
"loss":
|
| 1308 |
"step": 2150
|
| 1309 |
},
|
| 1310 |
{
|
| 1311 |
"epoch": 14.9,
|
| 1312 |
"learning_rate": 3.894736842105263e-05,
|
| 1313 |
-
"loss":
|
| 1314 |
"step": 2160
|
| 1315 |
},
|
| 1316 |
{
|
| 1317 |
"epoch": 14.97,
|
| 1318 |
"learning_rate": 3.842105263157895e-05,
|
| 1319 |
-
"loss":
|
| 1320 |
"step": 2170
|
| 1321 |
},
|
| 1322 |
{
|
| 1323 |
"epoch": 15.03,
|
| 1324 |
"learning_rate": 3.789473684210527e-05,
|
| 1325 |
-
"loss":
|
| 1326 |
"step": 2180
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 15.1,
|
| 1330 |
"learning_rate": 3.736842105263158e-05,
|
| 1331 |
-
"loss":
|
| 1332 |
"step": 2190
|
| 1333 |
},
|
| 1334 |
{
|
| 1335 |
"epoch": 15.17,
|
| 1336 |
"learning_rate": 3.6842105263157895e-05,
|
| 1337 |
-
"loss":
|
| 1338 |
"step": 2200
|
| 1339 |
},
|
| 1340 |
{
|
| 1341 |
"epoch": 15.24,
|
| 1342 |
"learning_rate": 3.6315789473684214e-05,
|
| 1343 |
-
"loss":
|
| 1344 |
"step": 2210
|
| 1345 |
},
|
| 1346 |
{
|
| 1347 |
"epoch": 15.31,
|
| 1348 |
"learning_rate": 3.578947368421053e-05,
|
| 1349 |
-
"loss":
|
| 1350 |
"step": 2220
|
| 1351 |
},
|
| 1352 |
{
|
| 1353 |
"epoch": 15.38,
|
| 1354 |
"learning_rate": 3.526315789473684e-05,
|
| 1355 |
-
"loss":
|
| 1356 |
"step": 2230
|
| 1357 |
},
|
| 1358 |
{
|
| 1359 |
"epoch": 15.45,
|
| 1360 |
"learning_rate": 3.473684210526316e-05,
|
| 1361 |
-
"loss":
|
| 1362 |
"step": 2240
|
| 1363 |
},
|
| 1364 |
{
|
| 1365 |
"epoch": 15.52,
|
| 1366 |
"learning_rate": 3.421052631578947e-05,
|
| 1367 |
-
"loss":
|
| 1368 |
"step": 2250
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 15.59,
|
| 1372 |
"learning_rate": 3.368421052631579e-05,
|
| 1373 |
-
"loss":
|
| 1374 |
"step": 2260
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
"epoch": 15.66,
|
| 1378 |
"learning_rate": 3.3157894736842106e-05,
|
| 1379 |
-
"loss":
|
| 1380 |
"step": 2270
|
| 1381 |
},
|
| 1382 |
{
|
| 1383 |
"epoch": 15.72,
|
| 1384 |
"learning_rate": 3.2631578947368426e-05,
|
| 1385 |
-
"loss":
|
| 1386 |
"step": 2280
|
| 1387 |
},
|
| 1388 |
{
|
| 1389 |
"epoch": 15.79,
|
| 1390 |
"learning_rate": 3.210526315789474e-05,
|
| 1391 |
-
"loss":
|
| 1392 |
"step": 2290
|
| 1393 |
},
|
| 1394 |
{
|
| 1395 |
"epoch": 15.86,
|
| 1396 |
"learning_rate": 3.157894736842105e-05,
|
| 1397 |
-
"loss":
|
| 1398 |
"step": 2300
|
| 1399 |
},
|
| 1400 |
{
|
| 1401 |
"epoch": 15.93,
|
| 1402 |
"learning_rate": 3.105263157894737e-05,
|
| 1403 |
-
"loss":
|
| 1404 |
"step": 2310
|
| 1405 |
},
|
| 1406 |
{
|
| 1407 |
"epoch": 16.0,
|
| 1408 |
"learning_rate": 3.0526315789473684e-05,
|
| 1409 |
-
"loss":
|
| 1410 |
"step": 2320
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 16.07,
|
| 1414 |
"learning_rate": 3e-05,
|
| 1415 |
-
"loss":
|
| 1416 |
"step": 2330
|
| 1417 |
},
|
| 1418 |
{
|
| 1419 |
"epoch": 16.14,
|
| 1420 |
"learning_rate": 2.9473684210526314e-05,
|
| 1421 |
-
"loss":
|
| 1422 |
"step": 2340
|
| 1423 |
},
|
| 1424 |
{
|
| 1425 |
"epoch": 16.21,
|
| 1426 |
"learning_rate": 2.8947368421052634e-05,
|
| 1427 |
-
"loss":
|
| 1428 |
"step": 2350
|
| 1429 |
},
|
| 1430 |
{
|
| 1431 |
"epoch": 16.28,
|
| 1432 |
"learning_rate": 2.842105263157895e-05,
|
| 1433 |
-
"loss":
|
| 1434 |
"step": 2360
|
| 1435 |
},
|
| 1436 |
{
|
| 1437 |
"epoch": 16.34,
|
| 1438 |
"learning_rate": 2.7894736842105263e-05,
|
| 1439 |
-
"loss":
|
| 1440 |
"step": 2370
|
| 1441 |
},
|
| 1442 |
{
|
| 1443 |
"epoch": 16.41,
|
| 1444 |
"learning_rate": 2.7368421052631583e-05,
|
| 1445 |
-
"loss":
|
| 1446 |
"step": 2380
|
| 1447 |
},
|
| 1448 |
{
|
| 1449 |
"epoch": 16.48,
|
| 1450 |
"learning_rate": 2.6842105263157896e-05,
|
| 1451 |
-
"loss":
|
| 1452 |
"step": 2390
|
| 1453 |
},
|
| 1454 |
{
|
| 1455 |
"epoch": 16.55,
|
| 1456 |
"learning_rate": 2.6315789473684212e-05,
|
| 1457 |
-
"loss":
|
| 1458 |
"step": 2400
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
"epoch": 16.62,
|
| 1462 |
"learning_rate": 2.578947368421053e-05,
|
| 1463 |
-
"loss":
|
| 1464 |
"step": 2410
|
| 1465 |
},
|
| 1466 |
{
|
| 1467 |
"epoch": 16.69,
|
| 1468 |
"learning_rate": 2.5263157894736845e-05,
|
| 1469 |
-
"loss":
|
| 1470 |
"step": 2420
|
| 1471 |
},
|
| 1472 |
{
|
| 1473 |
"epoch": 16.76,
|
| 1474 |
"learning_rate": 2.4736842105263158e-05,
|
| 1475 |
-
"loss":
|
| 1476 |
"step": 2430
|
| 1477 |
},
|
| 1478 |
{
|
| 1479 |
"epoch": 16.83,
|
| 1480 |
"learning_rate": 2.4210526315789474e-05,
|
| 1481 |
-
"loss":
|
| 1482 |
"step": 2440
|
| 1483 |
},
|
| 1484 |
{
|
| 1485 |
"epoch": 16.9,
|
| 1486 |
"learning_rate": 2.368421052631579e-05,
|
| 1487 |
-
"loss":
|
| 1488 |
"step": 2450
|
| 1489 |
},
|
| 1490 |
{
|
| 1491 |
"epoch": 16.97,
|
| 1492 |
"learning_rate": 2.3157894736842107e-05,
|
| 1493 |
-
"loss":
|
| 1494 |
"step": 2460
|
| 1495 |
},
|
| 1496 |
{
|
| 1497 |
"epoch": 17.03,
|
| 1498 |
"learning_rate": 2.2631578947368423e-05,
|
| 1499 |
-
"loss":
|
| 1500 |
"step": 2470
|
| 1501 |
},
|
| 1502 |
{
|
| 1503 |
"epoch": 17.1,
|
| 1504 |
"learning_rate": 2.2105263157894736e-05,
|
| 1505 |
-
"loss":
|
| 1506 |
"step": 2480
|
| 1507 |
},
|
| 1508 |
{
|
| 1509 |
"epoch": 17.17,
|
| 1510 |
"learning_rate": 2.1578947368421053e-05,
|
| 1511 |
-
"loss":
|
| 1512 |
"step": 2490
|
| 1513 |
},
|
| 1514 |
{
|
| 1515 |
"epoch": 17.24,
|
| 1516 |
"learning_rate": 2.105263157894737e-05,
|
| 1517 |
-
"loss":
|
| 1518 |
"step": 2500
|
| 1519 |
},
|
| 1520 |
{
|
| 1521 |
"epoch": 17.31,
|
| 1522 |
"learning_rate": 2.0526315789473685e-05,
|
| 1523 |
-
"loss":
|
| 1524 |
"step": 2510
|
| 1525 |
},
|
| 1526 |
{
|
| 1527 |
"epoch": 17.38,
|
| 1528 |
"learning_rate": 2e-05,
|
| 1529 |
-
"loss":
|
| 1530 |
"step": 2520
|
| 1531 |
},
|
| 1532 |
{
|
| 1533 |
"epoch": 17.45,
|
| 1534 |
"learning_rate": 1.9473684210526315e-05,
|
| 1535 |
-
"loss":
|
| 1536 |
"step": 2530
|
| 1537 |
},
|
| 1538 |
{
|
| 1539 |
"epoch": 17.52,
|
| 1540 |
"learning_rate": 1.8947368421052634e-05,
|
| 1541 |
-
"loss":
|
| 1542 |
"step": 2540
|
| 1543 |
},
|
| 1544 |
{
|
| 1545 |
"epoch": 17.59,
|
| 1546 |
"learning_rate": 1.8421052631578947e-05,
|
| 1547 |
-
"loss":
|
| 1548 |
"step": 2550
|
| 1549 |
},
|
| 1550 |
{
|
| 1551 |
"epoch": 17.66,
|
| 1552 |
"learning_rate": 1.7894736842105264e-05,
|
| 1553 |
-
"loss":
|
| 1554 |
"step": 2560
|
| 1555 |
},
|
| 1556 |
{
|
| 1557 |
"epoch": 17.72,
|
| 1558 |
"learning_rate": 1.736842105263158e-05,
|
| 1559 |
-
"loss":
|
| 1560 |
"step": 2570
|
| 1561 |
},
|
| 1562 |
{
|
| 1563 |
"epoch": 17.79,
|
| 1564 |
"learning_rate": 1.6842105263157896e-05,
|
| 1565 |
-
"loss":
|
| 1566 |
"step": 2580
|
| 1567 |
},
|
| 1568 |
{
|
| 1569 |
"epoch": 17.86,
|
| 1570 |
"learning_rate": 1.6315789473684213e-05,
|
| 1571 |
-
"loss":
|
| 1572 |
"step": 2590
|
| 1573 |
},
|
| 1574 |
{
|
| 1575 |
"epoch": 17.93,
|
| 1576 |
"learning_rate": 1.5789473684210526e-05,
|
| 1577 |
-
"loss":
|
| 1578 |
"step": 2600
|
| 1579 |
},
|
| 1580 |
{
|
| 1581 |
"epoch": 18.0,
|
| 1582 |
"learning_rate": 1.5263157894736842e-05,
|
| 1583 |
-
"loss":
|
| 1584 |
"step": 2610
|
| 1585 |
},
|
| 1586 |
{
|
| 1587 |
"epoch": 18.07,
|
| 1588 |
"learning_rate": 1.4736842105263157e-05,
|
| 1589 |
-
"loss":
|
| 1590 |
"step": 2620
|
| 1591 |
},
|
| 1592 |
{
|
| 1593 |
"epoch": 18.14,
|
| 1594 |
"learning_rate": 1.4210526315789475e-05,
|
| 1595 |
-
"loss":
|
| 1596 |
"step": 2630
|
| 1597 |
},
|
| 1598 |
{
|
| 1599 |
"epoch": 18.21,
|
| 1600 |
"learning_rate": 1.3684210526315791e-05,
|
| 1601 |
-
"loss":
|
| 1602 |
"step": 2640
|
| 1603 |
},
|
| 1604 |
{
|
| 1605 |
"epoch": 18.28,
|
| 1606 |
"learning_rate": 1.3157894736842106e-05,
|
| 1607 |
-
"loss":
|
| 1608 |
"step": 2650
|
| 1609 |
},
|
| 1610 |
{
|
| 1611 |
"epoch": 18.34,
|
| 1612 |
"learning_rate": 1.2631578947368422e-05,
|
| 1613 |
-
"loss":
|
| 1614 |
"step": 2660
|
| 1615 |
},
|
| 1616 |
{
|
| 1617 |
"epoch": 18.41,
|
| 1618 |
"learning_rate": 1.2105263157894737e-05,
|
| 1619 |
-
"loss":
|
| 1620 |
"step": 2670
|
| 1621 |
},
|
| 1622 |
{
|
| 1623 |
"epoch": 18.48,
|
| 1624 |
"learning_rate": 1.1578947368421053e-05,
|
| 1625 |
-
"loss":
|
| 1626 |
"step": 2680
|
| 1627 |
},
|
| 1628 |
{
|
| 1629 |
"epoch": 18.55,
|
| 1630 |
"learning_rate": 1.1052631578947368e-05,
|
| 1631 |
-
"loss":
|
| 1632 |
"step": 2690
|
| 1633 |
},
|
| 1634 |
{
|
| 1635 |
"epoch": 18.62,
|
| 1636 |
"learning_rate": 1.0526315789473684e-05,
|
| 1637 |
-
"loss":
|
| 1638 |
"step": 2700
|
| 1639 |
},
|
| 1640 |
{
|
| 1641 |
"epoch": 18.69,
|
| 1642 |
"learning_rate": 1e-05,
|
| 1643 |
-
"loss":
|
| 1644 |
"step": 2710
|
| 1645 |
},
|
| 1646 |
{
|
| 1647 |
"epoch": 18.76,
|
| 1648 |
"learning_rate": 9.473684210526317e-06,
|
| 1649 |
-
"loss":
|
| 1650 |
"step": 2720
|
| 1651 |
},
|
| 1652 |
{
|
| 1653 |
"epoch": 18.83,
|
| 1654 |
"learning_rate": 8.947368421052632e-06,
|
| 1655 |
-
"loss":
|
| 1656 |
"step": 2730
|
| 1657 |
},
|
| 1658 |
{
|
| 1659 |
"epoch": 18.9,
|
| 1660 |
"learning_rate": 8.421052631578948e-06,
|
| 1661 |
-
"loss":
|
| 1662 |
"step": 2740
|
| 1663 |
},
|
| 1664 |
{
|
| 1665 |
"epoch": 18.97,
|
| 1666 |
"learning_rate": 7.894736842105263e-06,
|
| 1667 |
-
"loss":
|
| 1668 |
"step": 2750
|
| 1669 |
},
|
| 1670 |
{
|
| 1671 |
"epoch": 19.03,
|
| 1672 |
"learning_rate": 7.3684210526315784e-06,
|
| 1673 |
-
"loss":
|
| 1674 |
"step": 2760
|
| 1675 |
},
|
| 1676 |
{
|
| 1677 |
"epoch": 19.1,
|
| 1678 |
"learning_rate": 6.842105263157896e-06,
|
| 1679 |
-
"loss":
|
| 1680 |
"step": 2770
|
| 1681 |
},
|
| 1682 |
{
|
| 1683 |
"epoch": 19.17,
|
| 1684 |
"learning_rate": 6.315789473684211e-06,
|
| 1685 |
-
"loss":
|
| 1686 |
"step": 2780
|
| 1687 |
},
|
| 1688 |
{
|
| 1689 |
"epoch": 19.24,
|
| 1690 |
"learning_rate": 5.789473684210527e-06,
|
| 1691 |
-
"loss":
|
| 1692 |
"step": 2790
|
| 1693 |
},
|
| 1694 |
{
|
| 1695 |
"epoch": 19.31,
|
| 1696 |
"learning_rate": 5.263157894736842e-06,
|
| 1697 |
-
"loss":
|
| 1698 |
"step": 2800
|
| 1699 |
},
|
| 1700 |
{
|
| 1701 |
"epoch": 19.38,
|
| 1702 |
"learning_rate": 4.736842105263159e-06,
|
| 1703 |
-
"loss":
|
| 1704 |
"step": 2810
|
| 1705 |
},
|
| 1706 |
{
|
| 1707 |
"epoch": 19.45,
|
| 1708 |
"learning_rate": 4.210526315789474e-06,
|
| 1709 |
-
"loss":
|
| 1710 |
"step": 2820
|
| 1711 |
},
|
| 1712 |
{
|
| 1713 |
"epoch": 19.52,
|
| 1714 |
"learning_rate": 3.6842105263157892e-06,
|
| 1715 |
-
"loss":
|
| 1716 |
"step": 2830
|
| 1717 |
},
|
| 1718 |
{
|
| 1719 |
"epoch": 19.59,
|
| 1720 |
"learning_rate": 3.1578947368421056e-06,
|
| 1721 |
-
"loss":
|
| 1722 |
"step": 2840
|
| 1723 |
},
|
| 1724 |
{
|
| 1725 |
"epoch": 19.66,
|
| 1726 |
"learning_rate": 2.631578947368421e-06,
|
| 1727 |
-
"loss":
|
| 1728 |
"step": 2850
|
| 1729 |
},
|
| 1730 |
{
|
| 1731 |
"epoch": 19.72,
|
| 1732 |
"learning_rate": 2.105263157894737e-06,
|
| 1733 |
-
"loss":
|
| 1734 |
"step": 2860
|
| 1735 |
},
|
| 1736 |
{
|
| 1737 |
"epoch": 19.79,
|
| 1738 |
"learning_rate": 1.5789473684210528e-06,
|
| 1739 |
-
"loss":
|
| 1740 |
"step": 2870
|
| 1741 |
},
|
| 1742 |
{
|
| 1743 |
"epoch": 19.86,
|
| 1744 |
"learning_rate": 1.0526315789473685e-06,
|
| 1745 |
-
"loss":
|
| 1746 |
"step": 2880
|
| 1747 |
},
|
| 1748 |
{
|
| 1749 |
"epoch": 19.93,
|
| 1750 |
"learning_rate": 5.263157894736843e-07,
|
| 1751 |
-
"loss":
|
| 1752 |
"step": 2890
|
| 1753 |
},
|
| 1754 |
{
|
| 1755 |
"epoch": 20.0,
|
| 1756 |
"learning_rate": 0.0,
|
| 1757 |
-
"loss":
|
| 1758 |
"step": 2900
|
| 1759 |
},
|
| 1760 |
{
|
| 1761 |
"epoch": 20.0,
|
| 1762 |
-
"eval_loss":
|
| 1763 |
-
"eval_runtime": 8.
|
| 1764 |
-
"eval_samples_per_second":
|
| 1765 |
-
"eval_steps_per_second": 6.
|
| 1766 |
-
"eval_wer": 0.
|
| 1767 |
"step": 2900
|
| 1768 |
},
|
| 1769 |
{
|
| 1770 |
"epoch": 20.0,
|
| 1771 |
"step": 2900,
|
| 1772 |
"total_flos": 2.6569362344615726e+18,
|
| 1773 |
-
"train_loss":
|
| 1774 |
-
"train_runtime":
|
| 1775 |
-
"train_samples_per_second":
|
| 1776 |
-
"train_steps_per_second":
|
| 1777 |
}
|
| 1778 |
],
|
| 1779 |
"logging_steps": 10,
|
|
|
|
| 11 |
{
|
| 12 |
"epoch": 0.07,
|
| 13 |
"learning_rate": 1.0000000000000002e-06,
|
| 14 |
+
"loss": 8.9224,
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.14,
|
| 19 |
"learning_rate": 2.0000000000000003e-06,
|
| 20 |
+
"loss": 8.5406,
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.21,
|
| 25 |
"learning_rate": 3e-06,
|
| 26 |
+
"loss": 7.9663,
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.28,
|
| 31 |
"learning_rate": 4.000000000000001e-06,
|
| 32 |
+
"loss": 8.388,
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.34,
|
| 37 |
"learning_rate": 5e-06,
|
| 38 |
+
"loss": 8.6117,
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.41,
|
| 43 |
"learning_rate": 6e-06,
|
| 44 |
+
"loss": 8.2034,
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.48,
|
| 49 |
"learning_rate": 7.000000000000001e-06,
|
| 50 |
+
"loss": 7.7261,
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.55,
|
| 55 |
"learning_rate": 8.000000000000001e-06,
|
| 56 |
+
"loss": 8.6426,
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.62,
|
| 61 |
"learning_rate": 9e-06,
|
| 62 |
+
"loss": 8.3122,
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.69,
|
| 67 |
"learning_rate": 1e-05,
|
| 68 |
+
"loss": 7.9465,
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.76,
|
| 73 |
"learning_rate": 1.1000000000000001e-05,
|
| 74 |
+
"loss": 7.9357,
|
| 75 |
"step": 110
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.83,
|
| 79 |
"learning_rate": 1.2e-05,
|
| 80 |
+
"loss": 8.333,
|
| 81 |
"step": 120
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.9,
|
| 85 |
"learning_rate": 1.3000000000000001e-05,
|
| 86 |
+
"loss": 8.0391,
|
| 87 |
"step": 130
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.97,
|
| 91 |
"learning_rate": 1.4000000000000001e-05,
|
| 92 |
+
"loss": 7.5063,
|
| 93 |
"step": 140
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 1.03,
|
| 97 |
"learning_rate": 1.5e-05,
|
| 98 |
+
"loss": 7.9733,
|
| 99 |
"step": 150
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 1.1,
|
| 103 |
"learning_rate": 1.6000000000000003e-05,
|
| 104 |
+
"loss": 7.8903,
|
| 105 |
"step": 160
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 1.17,
|
| 109 |
"learning_rate": 1.7000000000000003e-05,
|
| 110 |
+
"loss": 7.6222,
|
| 111 |
"step": 170
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 1.24,
|
| 115 |
"learning_rate": 1.8e-05,
|
| 116 |
+
"loss": 7.1372,
|
| 117 |
"step": 180
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 1.31,
|
| 121 |
"learning_rate": 1.9e-05,
|
| 122 |
+
"loss": 8.0093,
|
| 123 |
"step": 190
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 1.38,
|
| 127 |
"learning_rate": 2e-05,
|
| 128 |
+
"loss": 7.4925,
|
| 129 |
"step": 200
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 1.45,
|
| 133 |
"learning_rate": 2.1e-05,
|
| 134 |
+
"loss": 7.0882,
|
| 135 |
"step": 210
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 1.52,
|
| 139 |
"learning_rate": 2.2000000000000003e-05,
|
| 140 |
+
"loss": 7.3101,
|
| 141 |
"step": 220
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 1.59,
|
| 145 |
"learning_rate": 2.3000000000000003e-05,
|
| 146 |
+
"loss": 7.3497,
|
| 147 |
"step": 230
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 1.66,
|
| 151 |
"learning_rate": 2.4e-05,
|
| 152 |
+
"loss": 6.957,
|
| 153 |
"step": 240
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 1.72,
|
| 157 |
"learning_rate": 2.5e-05,
|
| 158 |
+
"loss": 6.6912,
|
| 159 |
"step": 250
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 1.79,
|
| 163 |
"learning_rate": 2.6000000000000002e-05,
|
| 164 |
+
"loss": 6.9221,
|
| 165 |
"step": 260
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 1.86,
|
| 169 |
"learning_rate": 2.7000000000000002e-05,
|
| 170 |
+
"loss": 6.7065,
|
| 171 |
"step": 270
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 1.93,
|
| 175 |
"learning_rate": 2.8000000000000003e-05,
|
| 176 |
+
"loss": 6.3174,
|
| 177 |
"step": 280
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 2.0,
|
| 181 |
"learning_rate": 2.9e-05,
|
| 182 |
+
"loss": 6.1904,
|
| 183 |
"step": 290
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 2.07,
|
| 187 |
"learning_rate": 3e-05,
|
| 188 |
+
"loss": 6.4006,
|
| 189 |
"step": 300
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 2.14,
|
| 193 |
"learning_rate": 3.1e-05,
|
| 194 |
+
"loss": 6.0932,
|
| 195 |
"step": 310
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 2.21,
|
| 199 |
"learning_rate": 3.2000000000000005e-05,
|
| 200 |
+
"loss": 5.717,
|
| 201 |
"step": 320
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 2.28,
|
| 205 |
"learning_rate": 3.3e-05,
|
| 206 |
+
"loss": 5.8356,
|
| 207 |
"step": 330
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 2.34,
|
| 211 |
"learning_rate": 3.4000000000000007e-05,
|
| 212 |
+
"loss": 5.6073,
|
| 213 |
"step": 340
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 2.41,
|
| 217 |
"learning_rate": 3.5e-05,
|
| 218 |
+
"loss": 5.3597,
|
| 219 |
"step": 350
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 2.48,
|
| 223 |
"learning_rate": 3.6e-05,
|
| 224 |
+
"loss": 5.0843,
|
| 225 |
"step": 360
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 2.55,
|
| 229 |
"learning_rate": 3.7e-05,
|
| 230 |
+
"loss": 5.3473,
|
| 231 |
"step": 370
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 2.62,
|
| 235 |
"learning_rate": 3.8e-05,
|
| 236 |
+
"loss": 5.0528,
|
| 237 |
"step": 380
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 2.69,
|
| 241 |
"learning_rate": 3.9000000000000006e-05,
|
| 242 |
+
"loss": 4.8183,
|
| 243 |
"step": 390
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 2.76,
|
| 247 |
"learning_rate": 4e-05,
|
| 248 |
+
"loss": 4.7668,
|
| 249 |
"step": 400
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 2.83,
|
| 253 |
"learning_rate": 4.1e-05,
|
| 254 |
+
"loss": 4.6995,
|
| 255 |
"step": 410
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 2.9,
|
| 259 |
"learning_rate": 4.2e-05,
|
| 260 |
+
"loss": 4.5144,
|
| 261 |
"step": 420
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 2.97,
|
| 265 |
"learning_rate": 4.3e-05,
|
| 266 |
+
"loss": 4.3133,
|
| 267 |
"step": 430
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 3.03,
|
| 271 |
"learning_rate": 4.4000000000000006e-05,
|
| 272 |
+
"loss": 4.3413,
|
| 273 |
"step": 440
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 3.1,
|
| 277 |
"learning_rate": 4.5e-05,
|
| 278 |
+
"loss": 4.1984,
|
| 279 |
"step": 450
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 3.17,
|
| 283 |
"learning_rate": 4.600000000000001e-05,
|
| 284 |
+
"loss": 4.0813,
|
| 285 |
"step": 460
|
| 286 |
},
|
| 287 |
{
|
| 288 |
"epoch": 3.24,
|
| 289 |
"learning_rate": 4.7e-05,
|
| 290 |
+
"loss": 4.0248,
|
| 291 |
"step": 470
|
| 292 |
},
|
| 293 |
{
|
| 294 |
"epoch": 3.31,
|
| 295 |
"learning_rate": 4.8e-05,
|
| 296 |
+
"loss": 4.0405,
|
| 297 |
"step": 480
|
| 298 |
},
|
| 299 |
{
|
| 300 |
"epoch": 3.38,
|
| 301 |
"learning_rate": 4.9e-05,
|
| 302 |
+
"loss": 3.9165,
|
| 303 |
"step": 490
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"epoch": 3.45,
|
| 307 |
"learning_rate": 5e-05,
|
| 308 |
+
"loss": 3.8314,
|
| 309 |
"step": 500
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 3.52,
|
| 313 |
"learning_rate": 5.1000000000000006e-05,
|
| 314 |
+
"loss": 3.8916,
|
| 315 |
"step": 510
|
| 316 |
},
|
| 317 |
{
|
| 318 |
"epoch": 3.59,
|
| 319 |
"learning_rate": 5.2000000000000004e-05,
|
| 320 |
+
"loss": 3.8167,
|
| 321 |
"step": 520
|
| 322 |
},
|
| 323 |
{
|
| 324 |
"epoch": 3.66,
|
| 325 |
"learning_rate": 5.300000000000001e-05,
|
| 326 |
+
"loss": 3.7126,
|
| 327 |
"step": 530
|
| 328 |
},
|
| 329 |
{
|
| 330 |
"epoch": 3.72,
|
| 331 |
"learning_rate": 5.4000000000000005e-05,
|
| 332 |
+
"loss": 3.6607,
|
| 333 |
"step": 540
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 3.79,
|
| 337 |
"learning_rate": 5.500000000000001e-05,
|
| 338 |
+
"loss": 3.7683,
|
| 339 |
"step": 550
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 3.86,
|
| 343 |
"learning_rate": 5.6000000000000006e-05,
|
| 344 |
+
"loss": 3.6667,
|
| 345 |
"step": 560
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"epoch": 3.93,
|
| 349 |
"learning_rate": 5.6999999999999996e-05,
|
| 350 |
+
"loss": 3.6181,
|
| 351 |
"step": 570
|
| 352 |
},
|
| 353 |
{
|
| 354 |
"epoch": 4.0,
|
| 355 |
"learning_rate": 5.8e-05,
|
| 356 |
+
"loss": 3.6345,
|
| 357 |
"step": 580
|
| 358 |
},
|
| 359 |
{
|
| 360 |
"epoch": 4.07,
|
| 361 |
"learning_rate": 5.9e-05,
|
| 362 |
+
"loss": 3.6936,
|
| 363 |
"step": 590
|
| 364 |
},
|
| 365 |
{
|
| 366 |
"epoch": 4.14,
|
| 367 |
"learning_rate": 6e-05,
|
| 368 |
+
"loss": 3.5988,
|
| 369 |
"step": 600
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 4.21,
|
| 373 |
"learning_rate": 6.1e-05,
|
| 374 |
+
"loss": 3.5657,
|
| 375 |
"step": 610
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 4.28,
|
| 379 |
"learning_rate": 6.2e-05,
|
| 380 |
+
"loss": 3.6574,
|
| 381 |
"step": 620
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"epoch": 4.34,
|
| 385 |
"learning_rate": 6.3e-05,
|
| 386 |
+
"loss": 3.602,
|
| 387 |
"step": 630
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"epoch": 4.41,
|
| 391 |
"learning_rate": 6.400000000000001e-05,
|
| 392 |
+
"loss": 3.5477,
|
| 393 |
"step": 640
|
| 394 |
},
|
| 395 |
{
|
| 396 |
"epoch": 4.48,
|
| 397 |
"learning_rate": 6.500000000000001e-05,
|
| 398 |
+
"loss": 3.5357,
|
| 399 |
"step": 650
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 4.55,
|
| 403 |
"learning_rate": 6.6e-05,
|
| 404 |
+
"loss": 3.6488,
|
| 405 |
"step": 660
|
| 406 |
},
|
| 407 |
{
|
| 408 |
"epoch": 4.62,
|
| 409 |
"learning_rate": 6.7e-05,
|
| 410 |
+
"loss": 3.5509,
|
| 411 |
"step": 670
|
| 412 |
},
|
| 413 |
{
|
| 414 |
"epoch": 4.69,
|
| 415 |
"learning_rate": 6.800000000000001e-05,
|
| 416 |
+
"loss": 3.5384,
|
| 417 |
"step": 680
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 4.76,
|
| 421 |
"learning_rate": 6.9e-05,
|
| 422 |
+
"loss": 3.5553,
|
| 423 |
"step": 690
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"epoch": 4.83,
|
| 427 |
"learning_rate": 7e-05,
|
| 428 |
+
"loss": 3.5934,
|
| 429 |
"step": 700
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 4.9,
|
| 433 |
"learning_rate": 7.1e-05,
|
| 434 |
+
"loss": 3.5065,
|
| 435 |
"step": 710
|
| 436 |
},
|
| 437 |
{
|
| 438 |
"epoch": 4.97,
|
| 439 |
"learning_rate": 7.2e-05,
|
| 440 |
+
"loss": 3.473,
|
| 441 |
"step": 720
|
| 442 |
},
|
| 443 |
{
|
| 444 |
"epoch": 5.03,
|
| 445 |
"learning_rate": 7.3e-05,
|
| 446 |
+
"loss": 3.5895,
|
| 447 |
"step": 730
|
| 448 |
},
|
| 449 |
{
|
| 450 |
"epoch": 5.1,
|
| 451 |
"learning_rate": 7.4e-05,
|
| 452 |
+
"loss": 3.5227,
|
| 453 |
"step": 740
|
| 454 |
},
|
| 455 |
{
|
| 456 |
"epoch": 5.17,
|
| 457 |
"learning_rate": 7.500000000000001e-05,
|
| 458 |
+
"loss": 3.5018,
|
| 459 |
"step": 750
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 5.24,
|
| 463 |
"learning_rate": 7.6e-05,
|
| 464 |
+
"loss": 3.4822,
|
| 465 |
"step": 760
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"epoch": 5.31,
|
| 469 |
"learning_rate": 7.7e-05,
|
| 470 |
+
"loss": 3.5872,
|
| 471 |
"step": 770
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"epoch": 5.38,
|
| 475 |
"learning_rate": 7.800000000000001e-05,
|
| 476 |
+
"loss": 3.482,
|
| 477 |
"step": 780
|
| 478 |
},
|
| 479 |
{
|
| 480 |
"epoch": 5.45,
|
| 481 |
"learning_rate": 7.900000000000001e-05,
|
| 482 |
+
"loss": 3.4622,
|
| 483 |
"step": 790
|
| 484 |
},
|
| 485 |
{
|
| 486 |
"epoch": 5.52,
|
| 487 |
"learning_rate": 8e-05,
|
| 488 |
+
"loss": 3.4979,
|
| 489 |
"step": 800
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 5.59,
|
| 493 |
"learning_rate": 8.1e-05,
|
| 494 |
+
"loss": 3.5079,
|
| 495 |
"step": 810
|
| 496 |
},
|
| 497 |
{
|
| 498 |
"epoch": 5.66,
|
| 499 |
"learning_rate": 8.2e-05,
|
| 500 |
+
"loss": 3.4341,
|
| 501 |
"step": 820
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 5.72,
|
| 505 |
"learning_rate": 8.3e-05,
|
| 506 |
+
"loss": 3.4151,
|
| 507 |
"step": 830
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"epoch": 5.79,
|
| 511 |
"learning_rate": 8.4e-05,
|
| 512 |
+
"loss": 3.5327,
|
| 513 |
"step": 840
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 5.86,
|
| 517 |
"learning_rate": 8.5e-05,
|
| 518 |
+
"loss": 3.4457,
|
| 519 |
"step": 850
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 5.93,
|
| 523 |
"learning_rate": 8.6e-05,
|
| 524 |
+
"loss": 3.4219,
|
| 525 |
"step": 860
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 6.0,
|
| 529 |
"learning_rate": 8.7e-05,
|
| 530 |
+
"loss": 3.434,
|
| 531 |
"step": 870
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 6.07,
|
| 535 |
"learning_rate": 8.800000000000001e-05,
|
| 536 |
+
"loss": 3.5263,
|
| 537 |
"step": 880
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 6.14,
|
| 541 |
"learning_rate": 8.900000000000001e-05,
|
| 542 |
+
"loss": 3.4327,
|
| 543 |
"step": 890
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 6.21,
|
| 547 |
"learning_rate": 9e-05,
|
| 548 |
+
"loss": 3.3951,
|
| 549 |
"step": 900
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 6.28,
|
| 553 |
"learning_rate": 9.1e-05,
|
| 554 |
+
"loss": 3.4945,
|
| 555 |
"step": 910
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 6.34,
|
| 559 |
"learning_rate": 9.200000000000001e-05,
|
| 560 |
+
"loss": 3.4294,
|
| 561 |
"step": 920
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 6.41,
|
| 565 |
"learning_rate": 9.300000000000001e-05,
|
| 566 |
+
"loss": 3.3794,
|
| 567 |
"step": 930
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 6.48,
|
| 571 |
"learning_rate": 9.4e-05,
|
| 572 |
+
"loss": 3.3755,
|
| 573 |
"step": 940
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 6.55,
|
| 577 |
"learning_rate": 9.5e-05,
|
| 578 |
+
"loss": 3.4776,
|
| 579 |
"step": 950
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 6.62,
|
| 583 |
"learning_rate": 9.6e-05,
|
| 584 |
+
"loss": 3.4096,
|
| 585 |
"step": 960
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 6.69,
|
| 589 |
"learning_rate": 9.7e-05,
|
| 590 |
+
"loss": 3.366,
|
| 591 |
"step": 970
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 6.76,
|
| 595 |
"learning_rate": 9.8e-05,
|
| 596 |
+
"loss": 3.3924,
|
| 597 |
"step": 980
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 6.83,
|
| 601 |
"learning_rate": 9.900000000000001e-05,
|
| 602 |
+
"loss": 3.4268,
|
| 603 |
"step": 990
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 6.9,
|
| 607 |
"learning_rate": 0.0001,
|
| 608 |
+
"loss": 3.3572,
|
| 609 |
"step": 1000
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 6.97,
|
| 613 |
"learning_rate": 9.947368421052632e-05,
|
| 614 |
+
"loss": 3.3574,
|
| 615 |
"step": 1010
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 7.03,
|
| 619 |
"learning_rate": 9.894736842105263e-05,
|
| 620 |
+
"loss": 3.4417,
|
| 621 |
"step": 1020
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 7.1,
|
| 625 |
"learning_rate": 9.842105263157894e-05,
|
| 626 |
+
"loss": 3.3858,
|
| 627 |
"step": 1030
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 7.17,
|
| 631 |
"learning_rate": 9.789473684210527e-05,
|
| 632 |
+
"loss": 3.3541,
|
| 633 |
"step": 1040
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 7.24,
|
| 637 |
"learning_rate": 9.736842105263158e-05,
|
| 638 |
+
"loss": 3.334,
|
| 639 |
"step": 1050
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 7.31,
|
| 643 |
"learning_rate": 9.68421052631579e-05,
|
| 644 |
+
"loss": 3.4488,
|
| 645 |
"step": 1060
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 7.38,
|
| 649 |
"learning_rate": 9.631578947368421e-05,
|
| 650 |
+
"loss": 3.3454,
|
| 651 |
"step": 1070
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 7.45,
|
| 655 |
"learning_rate": 9.578947368421052e-05,
|
| 656 |
+
"loss": 3.3057,
|
| 657 |
"step": 1080
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 7.52,
|
| 661 |
"learning_rate": 9.526315789473685e-05,
|
| 662 |
+
"loss": 3.3948,
|
| 663 |
"step": 1090
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 7.59,
|
| 667 |
"learning_rate": 9.473684210526316e-05,
|
| 668 |
+
"loss": 3.3869,
|
| 669 |
"step": 1100
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 7.66,
|
| 673 |
"learning_rate": 9.421052631578949e-05,
|
| 674 |
+
"loss": 3.3038,
|
| 675 |
"step": 1110
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 7.72,
|
| 679 |
"learning_rate": 9.36842105263158e-05,
|
| 680 |
+
"loss": 3.3054,
|
| 681 |
"step": 1120
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 7.79,
|
| 685 |
"learning_rate": 9.315789473684211e-05,
|
| 686 |
+
"loss": 3.4006,
|
| 687 |
"step": 1130
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 7.86,
|
| 691 |
"learning_rate": 9.263157894736843e-05,
|
| 692 |
+
"loss": 3.3313,
|
| 693 |
"step": 1140
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 7.93,
|
| 697 |
"learning_rate": 9.210526315789474e-05,
|
| 698 |
+
"loss": 3.2982,
|
| 699 |
"step": 1150
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 8.0,
|
| 703 |
"learning_rate": 9.157894736842105e-05,
|
| 704 |
+
"loss": 3.3071,
|
| 705 |
"step": 1160
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 8.07,
|
| 709 |
"learning_rate": 9.105263157894738e-05,
|
| 710 |
+
"loss": 3.3822,
|
| 711 |
"step": 1170
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 8.14,
|
| 715 |
"learning_rate": 9.052631578947369e-05,
|
| 716 |
+
"loss": 3.3196,
|
| 717 |
"step": 1180
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 8.21,
|
| 721 |
"learning_rate": 9e-05,
|
| 722 |
+
"loss": 3.2765,
|
| 723 |
"step": 1190
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 8.28,
|
| 727 |
"learning_rate": 8.947368421052632e-05,
|
| 728 |
+
"loss": 3.3805,
|
| 729 |
"step": 1200
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 8.34,
|
| 733 |
"learning_rate": 8.894736842105263e-05,
|
| 734 |
+
"loss": 3.3321,
|
| 735 |
"step": 1210
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 8.41,
|
| 739 |
"learning_rate": 8.842105263157894e-05,
|
| 740 |
+
"loss": 3.2837,
|
| 741 |
"step": 1220
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 8.48,
|
| 745 |
"learning_rate": 8.789473684210526e-05,
|
| 746 |
+
"loss": 3.2773,
|
| 747 |
"step": 1230
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 8.55,
|
| 751 |
"learning_rate": 8.736842105263158e-05,
|
| 752 |
+
"loss": 3.3566,
|
| 753 |
"step": 1240
|
| 754 |
},
|
| 755 |
{
|
| 756 |
"epoch": 8.62,
|
| 757 |
"learning_rate": 8.68421052631579e-05,
|
| 758 |
+
"loss": 3.2951,
|
| 759 |
"step": 1250
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 8.69,
|
| 763 |
"learning_rate": 8.631578947368421e-05,
|
| 764 |
+
"loss": 3.2632,
|
| 765 |
"step": 1260
|
| 766 |
},
|
| 767 |
{
|
| 768 |
"epoch": 8.76,
|
| 769 |
"learning_rate": 8.578947368421054e-05,
|
| 770 |
+
"loss": 3.3313,
|
| 771 |
"step": 1270
|
| 772 |
},
|
| 773 |
{
|
| 774 |
"epoch": 8.83,
|
| 775 |
"learning_rate": 8.526315789473685e-05,
|
| 776 |
+
"loss": 3.3523,
|
| 777 |
"step": 1280
|
| 778 |
},
|
| 779 |
{
|
| 780 |
"epoch": 8.9,
|
| 781 |
"learning_rate": 8.473684210526316e-05,
|
| 782 |
+
"loss": 3.2767,
|
| 783 |
"step": 1290
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 8.97,
|
| 787 |
"learning_rate": 8.421052631578948e-05,
|
| 788 |
+
"loss": 3.2448,
|
| 789 |
"step": 1300
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 9.03,
|
| 793 |
"learning_rate": 8.36842105263158e-05,
|
| 794 |
+
"loss": 3.3419,
|
| 795 |
"step": 1310
|
| 796 |
},
|
| 797 |
{
|
| 798 |
"epoch": 9.1,
|
| 799 |
"learning_rate": 8.315789473684212e-05,
|
| 800 |
+
"loss": 3.2813,
|
| 801 |
"step": 1320
|
| 802 |
},
|
| 803 |
{
|
| 804 |
"epoch": 9.17,
|
| 805 |
"learning_rate": 8.263157894736843e-05,
|
| 806 |
+
"loss": 3.2558,
|
| 807 |
"step": 1330
|
| 808 |
},
|
| 809 |
{
|
| 810 |
"epoch": 9.24,
|
| 811 |
"learning_rate": 8.210526315789474e-05,
|
| 812 |
+
"loss": 3.2502,
|
| 813 |
"step": 1340
|
| 814 |
},
|
| 815 |
{
|
| 816 |
"epoch": 9.31,
|
| 817 |
"learning_rate": 8.157894736842105e-05,
|
| 818 |
+
"loss": 3.3594,
|
| 819 |
"step": 1350
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 9.38,
|
| 823 |
"learning_rate": 8.105263157894737e-05,
|
| 824 |
+
"loss": 3.2739,
|
| 825 |
"step": 1360
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 9.45,
|
| 829 |
"learning_rate": 8.052631578947368e-05,
|
| 830 |
+
"loss": 3.2404,
|
| 831 |
"step": 1370
|
| 832 |
},
|
| 833 |
{
|
| 834 |
"epoch": 9.52,
|
| 835 |
"learning_rate": 8e-05,
|
| 836 |
+
"loss": 3.3528,
|
| 837 |
"step": 1380
|
| 838 |
},
|
| 839 |
{
|
| 840 |
"epoch": 9.59,
|
| 841 |
"learning_rate": 7.947368421052632e-05,
|
| 842 |
+
"loss": 3.3139,
|
| 843 |
"step": 1390
|
| 844 |
},
|
| 845 |
{
|
| 846 |
"epoch": 9.66,
|
| 847 |
"learning_rate": 7.894736842105263e-05,
|
| 848 |
+
"loss": 3.2333,
|
| 849 |
"step": 1400
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 9.72,
|
| 853 |
"learning_rate": 7.842105263157895e-05,
|
| 854 |
+
"loss": 3.225,
|
| 855 |
"step": 1410
|
| 856 |
},
|
| 857 |
{
|
| 858 |
"epoch": 9.79,
|
| 859 |
"learning_rate": 7.789473684210526e-05,
|
| 860 |
+
"loss": 3.3473,
|
| 861 |
"step": 1420
|
| 862 |
},
|
| 863 |
{
|
| 864 |
"epoch": 9.86,
|
| 865 |
"learning_rate": 7.736842105263159e-05,
|
| 866 |
+
"loss": 3.2565,
|
| 867 |
"step": 1430
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 9.93,
|
| 871 |
"learning_rate": 7.68421052631579e-05,
|
| 872 |
+
"loss": 3.2372,
|
| 873 |
"step": 1440
|
| 874 |
},
|
| 875 |
{
|
| 876 |
"epoch": 10.0,
|
| 877 |
"learning_rate": 7.631578947368422e-05,
|
| 878 |
+
"loss": 3.2725,
|
| 879 |
"step": 1450
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 10.0,
|
| 883 |
+
"eval_loss": 3.4699134826660156,
|
| 884 |
+
"eval_runtime": 8.1516,
|
| 885 |
+
"eval_samples_per_second": 206.093,
|
| 886 |
+
"eval_steps_per_second": 6.502,
|
| 887 |
+
"eval_wer": 1.0005513059058646,
|
| 888 |
"step": 1450
|
| 889 |
},
|
| 890 |
{
|
| 891 |
"epoch": 10.07,
|
| 892 |
"learning_rate": 7.578947368421054e-05,
|
| 893 |
+
"loss": 3.3204,
|
| 894 |
"step": 1460
|
| 895 |
},
|
| 896 |
{
|
| 897 |
"epoch": 10.14,
|
| 898 |
"learning_rate": 7.526315789473685e-05,
|
| 899 |
+
"loss": 3.2301,
|
| 900 |
"step": 1470
|
| 901 |
},
|
| 902 |
{
|
| 903 |
"epoch": 10.21,
|
| 904 |
"learning_rate": 7.473684210526316e-05,
|
| 905 |
+
"loss": 3.2381,
|
| 906 |
"step": 1480
|
| 907 |
},
|
| 908 |
{
|
| 909 |
"epoch": 10.28,
|
| 910 |
"learning_rate": 7.421052631578948e-05,
|
| 911 |
+
"loss": 3.3093,
|
| 912 |
"step": 1490
|
| 913 |
},
|
| 914 |
{
|
| 915 |
"epoch": 10.34,
|
| 916 |
"learning_rate": 7.368421052631579e-05,
|
| 917 |
+
"loss": 3.2747,
|
| 918 |
"step": 1500
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 10.41,
|
| 922 |
"learning_rate": 7.315789473684212e-05,
|
| 923 |
+
"loss": 3.2268,
|
| 924 |
"step": 1510
|
| 925 |
},
|
| 926 |
{
|
| 927 |
"epoch": 10.48,
|
| 928 |
"learning_rate": 7.263157894736843e-05,
|
| 929 |
+
"loss": 3.2105,
|
| 930 |
"step": 1520
|
| 931 |
},
|
| 932 |
{
|
| 933 |
"epoch": 10.55,
|
| 934 |
"learning_rate": 7.210526315789474e-05,
|
| 935 |
+
"loss": 3.3174,
|
| 936 |
"step": 1530
|
| 937 |
},
|
| 938 |
{
|
| 939 |
"epoch": 10.62,
|
| 940 |
"learning_rate": 7.157894736842105e-05,
|
| 941 |
+
"loss": 3.2254,
|
| 942 |
"step": 1540
|
| 943 |
},
|
| 944 |
{
|
| 945 |
"epoch": 10.69,
|
| 946 |
"learning_rate": 7.105263157894737e-05,
|
| 947 |
+
"loss": 3.2547,
|
| 948 |
"step": 1550
|
| 949 |
},
|
| 950 |
{
|
| 951 |
"epoch": 10.76,
|
| 952 |
"learning_rate": 7.052631578947368e-05,
|
| 953 |
+
"loss": 3.2704,
|
| 954 |
"step": 1560
|
| 955 |
},
|
| 956 |
{
|
| 957 |
"epoch": 10.83,
|
| 958 |
"learning_rate": 7e-05,
|
| 959 |
+
"loss": 3.3047,
|
| 960 |
"step": 1570
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 10.9,
|
| 964 |
"learning_rate": 6.947368421052632e-05,
|
| 965 |
+
"loss": 3.2407,
|
| 966 |
"step": 1580
|
| 967 |
},
|
| 968 |
{
|
| 969 |
"epoch": 10.97,
|
| 970 |
"learning_rate": 6.894736842105263e-05,
|
| 971 |
+
"loss": 3.1856,
|
| 972 |
"step": 1590
|
| 973 |
},
|
| 974 |
{
|
| 975 |
"epoch": 11.03,
|
| 976 |
"learning_rate": 6.842105263157895e-05,
|
| 977 |
+
"loss": 3.3326,
|
| 978 |
"step": 1600
|
| 979 |
},
|
| 980 |
{
|
| 981 |
"epoch": 11.1,
|
| 982 |
"learning_rate": 6.789473684210527e-05,
|
| 983 |
+
"loss": 3.2289,
|
| 984 |
"step": 1610
|
| 985 |
},
|
| 986 |
{
|
| 987 |
"epoch": 11.17,
|
| 988 |
"learning_rate": 6.736842105263159e-05,
|
| 989 |
+
"loss": 3.229,
|
| 990 |
"step": 1620
|
| 991 |
},
|
| 992 |
{
|
| 993 |
"epoch": 11.24,
|
| 994 |
"learning_rate": 6.68421052631579e-05,
|
| 995 |
+
"loss": 3.2157,
|
| 996 |
"step": 1630
|
| 997 |
},
|
| 998 |
{
|
| 999 |
"epoch": 11.31,
|
| 1000 |
"learning_rate": 6.631578947368421e-05,
|
| 1001 |
+
"loss": 3.3323,
|
| 1002 |
"step": 1640
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 11.38,
|
| 1006 |
"learning_rate": 6.578947368421054e-05,
|
| 1007 |
+
"loss": 3.2229,
|
| 1008 |
"step": 1650
|
| 1009 |
},
|
| 1010 |
{
|
| 1011 |
"epoch": 11.45,
|
| 1012 |
"learning_rate": 6.526315789473685e-05,
|
| 1013 |
+
"loss": 3.1919,
|
| 1014 |
"step": 1660
|
| 1015 |
},
|
| 1016 |
{
|
| 1017 |
"epoch": 11.52,
|
| 1018 |
"learning_rate": 6.473684210526316e-05,
|
| 1019 |
+
"loss": 3.2794,
|
| 1020 |
"step": 1670
|
| 1021 |
},
|
| 1022 |
{
|
| 1023 |
"epoch": 11.59,
|
| 1024 |
"learning_rate": 6.421052631578948e-05,
|
| 1025 |
+
"loss": 3.2455,
|
| 1026 |
"step": 1680
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
"epoch": 11.66,
|
| 1030 |
"learning_rate": 6.368421052631579e-05,
|
| 1031 |
+
"loss": 3.1852,
|
| 1032 |
"step": 1690
|
| 1033 |
},
|
| 1034 |
{
|
| 1035 |
"epoch": 11.72,
|
| 1036 |
"learning_rate": 6.31578947368421e-05,
|
| 1037 |
+
"loss": 3.1948,
|
| 1038 |
"step": 1700
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
"epoch": 11.79,
|
| 1042 |
"learning_rate": 6.263157894736842e-05,
|
| 1043 |
+
"loss": 3.3379,
|
| 1044 |
"step": 1710
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 11.86,
|
| 1048 |
"learning_rate": 6.210526315789474e-05,
|
| 1049 |
+
"loss": 3.2203,
|
| 1050 |
"step": 1720
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
"epoch": 11.93,
|
| 1054 |
"learning_rate": 6.157894736842106e-05,
|
| 1055 |
+
"loss": 3.2083,
|
| 1056 |
"step": 1730
|
| 1057 |
},
|
| 1058 |
{
|
| 1059 |
"epoch": 12.0,
|
| 1060 |
"learning_rate": 6.105263157894737e-05,
|
| 1061 |
+
"loss": 3.2124,
|
| 1062 |
"step": 1740
|
| 1063 |
},
|
| 1064 |
{
|
| 1065 |
"epoch": 12.07,
|
| 1066 |
"learning_rate": 6.052631578947369e-05,
|
| 1067 |
+
"loss": 3.2972,
|
| 1068 |
"step": 1750
|
| 1069 |
},
|
| 1070 |
{
|
| 1071 |
"epoch": 12.14,
|
| 1072 |
"learning_rate": 6e-05,
|
| 1073 |
+
"loss": 3.1723,
|
| 1074 |
"step": 1760
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
"epoch": 12.21,
|
| 1078 |
"learning_rate": 5.9473684210526315e-05,
|
| 1079 |
+
"loss": 3.1838,
|
| 1080 |
"step": 1770
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
"epoch": 12.28,
|
| 1084 |
"learning_rate": 5.894736842105263e-05,
|
| 1085 |
+
"loss": 3.2744,
|
| 1086 |
"step": 1780
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 12.34,
|
| 1090 |
"learning_rate": 5.8421052631578954e-05,
|
| 1091 |
+
"loss": 3.257,
|
| 1092 |
"step": 1790
|
| 1093 |
},
|
| 1094 |
{
|
| 1095 |
"epoch": 12.41,
|
| 1096 |
"learning_rate": 5.789473684210527e-05,
|
| 1097 |
+
"loss": 3.2009,
|
| 1098 |
"step": 1800
|
| 1099 |
},
|
| 1100 |
{
|
| 1101 |
"epoch": 12.48,
|
| 1102 |
"learning_rate": 5.736842105263158e-05,
|
| 1103 |
+
"loss": 3.1939,
|
| 1104 |
"step": 1810
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
"epoch": 12.55,
|
| 1108 |
"learning_rate": 5.68421052631579e-05,
|
| 1109 |
+
"loss": 3.3005,
|
| 1110 |
"step": 1820
|
| 1111 |
},
|
| 1112 |
{
|
| 1113 |
"epoch": 12.62,
|
| 1114 |
"learning_rate": 5.631578947368421e-05,
|
| 1115 |
+
"loss": 3.1852,
|
| 1116 |
"step": 1830
|
| 1117 |
},
|
| 1118 |
{
|
| 1119 |
"epoch": 12.69,
|
| 1120 |
"learning_rate": 5.5789473684210526e-05,
|
| 1121 |
+
"loss": 3.1796,
|
| 1122 |
"step": 1840
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
"epoch": 12.76,
|
| 1126 |
"learning_rate": 5.526315789473685e-05,
|
| 1127 |
+
"loss": 3.2465,
|
| 1128 |
"step": 1850
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 12.83,
|
| 1132 |
"learning_rate": 5.4736842105263165e-05,
|
| 1133 |
+
"loss": 3.2793,
|
| 1134 |
"step": 1860
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
"epoch": 12.9,
|
| 1138 |
"learning_rate": 5.421052631578948e-05,
|
| 1139 |
+
"loss": 3.179,
|
| 1140 |
"step": 1870
|
| 1141 |
},
|
| 1142 |
{
|
| 1143 |
"epoch": 12.97,
|
| 1144 |
"learning_rate": 5.368421052631579e-05,
|
| 1145 |
+
"loss": 3.2026,
|
| 1146 |
"step": 1880
|
| 1147 |
},
|
| 1148 |
{
|
| 1149 |
"epoch": 13.03,
|
| 1150 |
"learning_rate": 5.3157894736842104e-05,
|
| 1151 |
+
"loss": 3.2608,
|
| 1152 |
"step": 1890
|
| 1153 |
},
|
| 1154 |
{
|
| 1155 |
"epoch": 13.1,
|
| 1156 |
"learning_rate": 5.2631578947368424e-05,
|
| 1157 |
+
"loss": 3.2105,
|
| 1158 |
"step": 1900
|
| 1159 |
},
|
| 1160 |
{
|
| 1161 |
"epoch": 13.17,
|
| 1162 |
"learning_rate": 5.210526315789474e-05,
|
| 1163 |
+
"loss": 3.1839,
|
| 1164 |
"step": 1910
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
"epoch": 13.24,
|
| 1168 |
"learning_rate": 5.157894736842106e-05,
|
| 1169 |
+
"loss": 3.2157,
|
| 1170 |
"step": 1920
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
"epoch": 13.31,
|
| 1174 |
"learning_rate": 5.1052631578947376e-05,
|
| 1175 |
+
"loss": 3.2742,
|
| 1176 |
"step": 1930
|
| 1177 |
},
|
| 1178 |
{
|
| 1179 |
"epoch": 13.38,
|
| 1180 |
"learning_rate": 5.052631578947369e-05,
|
| 1181 |
+
"loss": 3.1615,
|
| 1182 |
"step": 1940
|
| 1183 |
},
|
| 1184 |
{
|
| 1185 |
"epoch": 13.45,
|
| 1186 |
"learning_rate": 5e-05,
|
| 1187 |
+
"loss": 3.1736,
|
| 1188 |
"step": 1950
|
| 1189 |
},
|
| 1190 |
{
|
| 1191 |
"epoch": 13.52,
|
| 1192 |
"learning_rate": 4.9473684210526315e-05,
|
| 1193 |
+
"loss": 3.2578,
|
| 1194 |
"step": 1960
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
"epoch": 13.59,
|
| 1198 |
"learning_rate": 4.8947368421052635e-05,
|
| 1199 |
+
"loss": 3.2397,
|
| 1200 |
"step": 1970
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 13.66,
|
| 1204 |
"learning_rate": 4.842105263157895e-05,
|
| 1205 |
+
"loss": 3.1528,
|
| 1206 |
"step": 1980
|
| 1207 |
},
|
| 1208 |
{
|
| 1209 |
"epoch": 13.72,
|
| 1210 |
"learning_rate": 4.789473684210526e-05,
|
| 1211 |
+
"loss": 3.1764,
|
| 1212 |
"step": 1990
|
| 1213 |
},
|
| 1214 |
{
|
| 1215 |
"epoch": 13.79,
|
| 1216 |
"learning_rate": 4.736842105263158e-05,
|
| 1217 |
+
"loss": 3.2888,
|
| 1218 |
"step": 2000
|
| 1219 |
},
|
| 1220 |
{
|
| 1221 |
"epoch": 13.86,
|
| 1222 |
"learning_rate": 4.68421052631579e-05,
|
| 1223 |
+
"loss": 3.1991,
|
| 1224 |
"step": 2010
|
| 1225 |
},
|
| 1226 |
{
|
| 1227 |
"epoch": 13.93,
|
| 1228 |
"learning_rate": 4.6315789473684214e-05,
|
| 1229 |
+
"loss": 3.1734,
|
| 1230 |
"step": 2020
|
| 1231 |
},
|
| 1232 |
{
|
| 1233 |
"epoch": 14.0,
|
| 1234 |
"learning_rate": 4.5789473684210527e-05,
|
| 1235 |
+
"loss": 3.193,
|
| 1236 |
"step": 2030
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
"epoch": 14.07,
|
| 1240 |
"learning_rate": 4.5263157894736846e-05,
|
| 1241 |
+
"loss": 3.2672,
|
| 1242 |
"step": 2040
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 14.14,
|
| 1246 |
"learning_rate": 4.473684210526316e-05,
|
| 1247 |
+
"loss": 3.1536,
|
| 1248 |
"step": 2050
|
| 1249 |
},
|
| 1250 |
{
|
| 1251 |
"epoch": 14.21,
|
| 1252 |
"learning_rate": 4.421052631578947e-05,
|
| 1253 |
+
"loss": 3.1904,
|
| 1254 |
"step": 2060
|
| 1255 |
},
|
| 1256 |
{
|
| 1257 |
"epoch": 14.28,
|
| 1258 |
"learning_rate": 4.368421052631579e-05,
|
| 1259 |
+
"loss": 3.2782,
|
| 1260 |
"step": 2070
|
| 1261 |
},
|
| 1262 |
{
|
| 1263 |
"epoch": 14.34,
|
| 1264 |
"learning_rate": 4.3157894736842105e-05,
|
| 1265 |
+
"loss": 3.2108,
|
| 1266 |
"step": 2080
|
| 1267 |
},
|
| 1268 |
{
|
| 1269 |
"epoch": 14.41,
|
| 1270 |
"learning_rate": 4.2631578947368425e-05,
|
| 1271 |
+
"loss": 3.1988,
|
| 1272 |
"step": 2090
|
| 1273 |
},
|
| 1274 |
{
|
| 1275 |
"epoch": 14.48,
|
| 1276 |
"learning_rate": 4.210526315789474e-05,
|
| 1277 |
+
"loss": 3.1915,
|
| 1278 |
"step": 2100
|
| 1279 |
},
|
| 1280 |
{
|
| 1281 |
"epoch": 14.55,
|
| 1282 |
"learning_rate": 4.157894736842106e-05,
|
| 1283 |
+
"loss": 3.2633,
|
| 1284 |
"step": 2110
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 14.62,
|
| 1288 |
"learning_rate": 4.105263157894737e-05,
|
| 1289 |
+
"loss": 3.1832,
|
| 1290 |
"step": 2120
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
"epoch": 14.69,
|
| 1294 |
"learning_rate": 4.0526315789473684e-05,
|
| 1295 |
+
"loss": 3.1445,
|
| 1296 |
"step": 2130
|
| 1297 |
},
|
| 1298 |
{
|
| 1299 |
"epoch": 14.76,
|
| 1300 |
"learning_rate": 4e-05,
|
| 1301 |
+
"loss": 3.2,
|
| 1302 |
"step": 2140
|
| 1303 |
},
|
| 1304 |
{
|
| 1305 |
"epoch": 14.83,
|
| 1306 |
"learning_rate": 3.9473684210526316e-05,
|
| 1307 |
+
"loss": 3.2517,
|
| 1308 |
"step": 2150
|
| 1309 |
},
|
| 1310 |
{
|
| 1311 |
"epoch": 14.9,
|
| 1312 |
"learning_rate": 3.894736842105263e-05,
|
| 1313 |
+
"loss": 3.1483,
|
| 1314 |
"step": 2160
|
| 1315 |
},
|
| 1316 |
{
|
| 1317 |
"epoch": 14.97,
|
| 1318 |
"learning_rate": 3.842105263157895e-05,
|
| 1319 |
+
"loss": 3.1521,
|
| 1320 |
"step": 2170
|
| 1321 |
},
|
| 1322 |
{
|
| 1323 |
"epoch": 15.03,
|
| 1324 |
"learning_rate": 3.789473684210527e-05,
|
| 1325 |
+
"loss": 3.273,
|
| 1326 |
"step": 2180
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 15.1,
|
| 1330 |
"learning_rate": 3.736842105263158e-05,
|
| 1331 |
+
"loss": 3.2115,
|
| 1332 |
"step": 2190
|
| 1333 |
},
|
| 1334 |
{
|
| 1335 |
"epoch": 15.17,
|
| 1336 |
"learning_rate": 3.6842105263157895e-05,
|
| 1337 |
+
"loss": 3.1547,
|
| 1338 |
"step": 2200
|
| 1339 |
},
|
| 1340 |
{
|
| 1341 |
"epoch": 15.24,
|
| 1342 |
"learning_rate": 3.6315789473684214e-05,
|
| 1343 |
+
"loss": 3.1564,
|
| 1344 |
"step": 2210
|
| 1345 |
},
|
| 1346 |
{
|
| 1347 |
"epoch": 15.31,
|
| 1348 |
"learning_rate": 3.578947368421053e-05,
|
| 1349 |
+
"loss": 3.2829,
|
| 1350 |
"step": 2220
|
| 1351 |
},
|
| 1352 |
{
|
| 1353 |
"epoch": 15.38,
|
| 1354 |
"learning_rate": 3.526315789473684e-05,
|
| 1355 |
+
"loss": 3.1727,
|
| 1356 |
"step": 2230
|
| 1357 |
},
|
| 1358 |
{
|
| 1359 |
"epoch": 15.45,
|
| 1360 |
"learning_rate": 3.473684210526316e-05,
|
| 1361 |
+
"loss": 3.1617,
|
| 1362 |
"step": 2240
|
| 1363 |
},
|
| 1364 |
{
|
| 1365 |
"epoch": 15.52,
|
| 1366 |
"learning_rate": 3.421052631578947e-05,
|
| 1367 |
+
"loss": 3.2143,
|
| 1368 |
"step": 2250
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 15.59,
|
| 1372 |
"learning_rate": 3.368421052631579e-05,
|
| 1373 |
+
"loss": 3.2045,
|
| 1374 |
"step": 2260
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
"epoch": 15.66,
|
| 1378 |
"learning_rate": 3.3157894736842106e-05,
|
| 1379 |
+
"loss": 3.1499,
|
| 1380 |
"step": 2270
|
| 1381 |
},
|
| 1382 |
{
|
| 1383 |
"epoch": 15.72,
|
| 1384 |
"learning_rate": 3.2631578947368426e-05,
|
| 1385 |
+
"loss": 3.1733,
|
| 1386 |
"step": 2280
|
| 1387 |
},
|
| 1388 |
{
|
| 1389 |
"epoch": 15.79,
|
| 1390 |
"learning_rate": 3.210526315789474e-05,
|
| 1391 |
+
"loss": 3.2579,
|
| 1392 |
"step": 2290
|
| 1393 |
},
|
| 1394 |
{
|
| 1395 |
"epoch": 15.86,
|
| 1396 |
"learning_rate": 3.157894736842105e-05,
|
| 1397 |
+
"loss": 3.1808,
|
| 1398 |
"step": 2300
|
| 1399 |
},
|
| 1400 |
{
|
| 1401 |
"epoch": 15.93,
|
| 1402 |
"learning_rate": 3.105263157894737e-05,
|
| 1403 |
+
"loss": 3.1587,
|
| 1404 |
"step": 2310
|
| 1405 |
},
|
| 1406 |
{
|
| 1407 |
"epoch": 16.0,
|
| 1408 |
"learning_rate": 3.0526315789473684e-05,
|
| 1409 |
+
"loss": 3.229,
|
| 1410 |
"step": 2320
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 16.07,
|
| 1414 |
"learning_rate": 3e-05,
|
| 1415 |
+
"loss": 3.2771,
|
| 1416 |
"step": 2330
|
| 1417 |
},
|
| 1418 |
{
|
| 1419 |
"epoch": 16.14,
|
| 1420 |
"learning_rate": 2.9473684210526314e-05,
|
| 1421 |
+
"loss": 3.178,
|
| 1422 |
"step": 2340
|
| 1423 |
},
|
| 1424 |
{
|
| 1425 |
"epoch": 16.21,
|
| 1426 |
"learning_rate": 2.8947368421052634e-05,
|
| 1427 |
+
"loss": 3.1357,
|
| 1428 |
"step": 2350
|
| 1429 |
},
|
| 1430 |
{
|
| 1431 |
"epoch": 16.28,
|
| 1432 |
"learning_rate": 2.842105263157895e-05,
|
| 1433 |
+
"loss": 3.2502,
|
| 1434 |
"step": 2360
|
| 1435 |
},
|
| 1436 |
{
|
| 1437 |
"epoch": 16.34,
|
| 1438 |
"learning_rate": 2.7894736842105263e-05,
|
| 1439 |
+
"loss": 3.1893,
|
| 1440 |
"step": 2370
|
| 1441 |
},
|
| 1442 |
{
|
| 1443 |
"epoch": 16.41,
|
| 1444 |
"learning_rate": 2.7368421052631583e-05,
|
| 1445 |
+
"loss": 3.1526,
|
| 1446 |
"step": 2380
|
| 1447 |
},
|
| 1448 |
{
|
| 1449 |
"epoch": 16.48,
|
| 1450 |
"learning_rate": 2.6842105263157896e-05,
|
| 1451 |
+
"loss": 3.1775,
|
| 1452 |
"step": 2390
|
| 1453 |
},
|
| 1454 |
{
|
| 1455 |
"epoch": 16.55,
|
| 1456 |
"learning_rate": 2.6315789473684212e-05,
|
| 1457 |
+
"loss": 3.266,
|
| 1458 |
"step": 2400
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
"epoch": 16.62,
|
| 1462 |
"learning_rate": 2.578947368421053e-05,
|
| 1463 |
+
"loss": 3.1632,
|
| 1464 |
"step": 2410
|
| 1465 |
},
|
| 1466 |
{
|
| 1467 |
"epoch": 16.69,
|
| 1468 |
"learning_rate": 2.5263157894736845e-05,
|
| 1469 |
+
"loss": 3.1558,
|
| 1470 |
"step": 2420
|
| 1471 |
},
|
| 1472 |
{
|
| 1473 |
"epoch": 16.76,
|
| 1474 |
"learning_rate": 2.4736842105263158e-05,
|
| 1475 |
+
"loss": 3.2264,
|
| 1476 |
"step": 2430
|
| 1477 |
},
|
| 1478 |
{
|
| 1479 |
"epoch": 16.83,
|
| 1480 |
"learning_rate": 2.4210526315789474e-05,
|
| 1481 |
+
"loss": 3.2213,
|
| 1482 |
"step": 2440
|
| 1483 |
},
|
| 1484 |
{
|
| 1485 |
"epoch": 16.9,
|
| 1486 |
"learning_rate": 2.368421052631579e-05,
|
| 1487 |
+
"loss": 3.1527,
|
| 1488 |
"step": 2450
|
| 1489 |
},
|
| 1490 |
{
|
| 1491 |
"epoch": 16.97,
|
| 1492 |
"learning_rate": 2.3157894736842107e-05,
|
| 1493 |
+
"loss": 3.1292,
|
| 1494 |
"step": 2460
|
| 1495 |
},
|
| 1496 |
{
|
| 1497 |
"epoch": 17.03,
|
| 1498 |
"learning_rate": 2.2631578947368423e-05,
|
| 1499 |
+
"loss": 3.251,
|
| 1500 |
"step": 2470
|
| 1501 |
},
|
| 1502 |
{
|
| 1503 |
"epoch": 17.1,
|
| 1504 |
"learning_rate": 2.2105263157894736e-05,
|
| 1505 |
+
"loss": 3.149,
|
| 1506 |
"step": 2480
|
| 1507 |
},
|
| 1508 |
{
|
| 1509 |
"epoch": 17.17,
|
| 1510 |
"learning_rate": 2.1578947368421053e-05,
|
| 1511 |
+
"loss": 3.1467,
|
| 1512 |
"step": 2490
|
| 1513 |
},
|
| 1514 |
{
|
| 1515 |
"epoch": 17.24,
|
| 1516 |
"learning_rate": 2.105263157894737e-05,
|
| 1517 |
+
"loss": 3.1744,
|
| 1518 |
"step": 2500
|
| 1519 |
},
|
| 1520 |
{
|
| 1521 |
"epoch": 17.31,
|
| 1522 |
"learning_rate": 2.0526315789473685e-05,
|
| 1523 |
+
"loss": 3.2617,
|
| 1524 |
"step": 2510
|
| 1525 |
},
|
| 1526 |
{
|
| 1527 |
"epoch": 17.38,
|
| 1528 |
"learning_rate": 2e-05,
|
| 1529 |
+
"loss": 3.1516,
|
| 1530 |
"step": 2520
|
| 1531 |
},
|
| 1532 |
{
|
| 1533 |
"epoch": 17.45,
|
| 1534 |
"learning_rate": 1.9473684210526315e-05,
|
| 1535 |
+
"loss": 3.1366,
|
| 1536 |
"step": 2530
|
| 1537 |
},
|
| 1538 |
{
|
| 1539 |
"epoch": 17.52,
|
| 1540 |
"learning_rate": 1.8947368421052634e-05,
|
| 1541 |
+
"loss": 3.2229,
|
| 1542 |
"step": 2540
|
| 1543 |
},
|
| 1544 |
{
|
| 1545 |
"epoch": 17.59,
|
| 1546 |
"learning_rate": 1.8421052631578947e-05,
|
| 1547 |
+
"loss": 3.195,
|
| 1548 |
"step": 2550
|
| 1549 |
},
|
| 1550 |
{
|
| 1551 |
"epoch": 17.66,
|
| 1552 |
"learning_rate": 1.7894736842105264e-05,
|
| 1553 |
+
"loss": 3.1612,
|
| 1554 |
"step": 2560
|
| 1555 |
},
|
| 1556 |
{
|
| 1557 |
"epoch": 17.72,
|
| 1558 |
"learning_rate": 1.736842105263158e-05,
|
| 1559 |
+
"loss": 3.1497,
|
| 1560 |
"step": 2570
|
| 1561 |
},
|
| 1562 |
{
|
| 1563 |
"epoch": 17.79,
|
| 1564 |
"learning_rate": 1.6842105263157896e-05,
|
| 1565 |
+
"loss": 3.2591,
|
| 1566 |
"step": 2580
|
| 1567 |
},
|
| 1568 |
{
|
| 1569 |
"epoch": 17.86,
|
| 1570 |
"learning_rate": 1.6315789473684213e-05,
|
| 1571 |
+
"loss": 3.1834,
|
| 1572 |
"step": 2590
|
| 1573 |
},
|
| 1574 |
{
|
| 1575 |
"epoch": 17.93,
|
| 1576 |
"learning_rate": 1.5789473684210526e-05,
|
| 1577 |
+
"loss": 3.1799,
|
| 1578 |
"step": 2600
|
| 1579 |
},
|
| 1580 |
{
|
| 1581 |
"epoch": 18.0,
|
| 1582 |
"learning_rate": 1.5263157894736842e-05,
|
| 1583 |
+
"loss": 3.1597,
|
| 1584 |
"step": 2610
|
| 1585 |
},
|
| 1586 |
{
|
| 1587 |
"epoch": 18.07,
|
| 1588 |
"learning_rate": 1.4736842105263157e-05,
|
| 1589 |
+
"loss": 3.2753,
|
| 1590 |
"step": 2620
|
| 1591 |
},
|
| 1592 |
{
|
| 1593 |
"epoch": 18.14,
|
| 1594 |
"learning_rate": 1.4210526315789475e-05,
|
| 1595 |
+
"loss": 3.1269,
|
| 1596 |
"step": 2630
|
| 1597 |
},
|
| 1598 |
{
|
| 1599 |
"epoch": 18.21,
|
| 1600 |
"learning_rate": 1.3684210526315791e-05,
|
| 1601 |
+
"loss": 3.1321,
|
| 1602 |
"step": 2640
|
| 1603 |
},
|
| 1604 |
{
|
| 1605 |
"epoch": 18.28,
|
| 1606 |
"learning_rate": 1.3157894736842106e-05,
|
| 1607 |
+
"loss": 3.2213,
|
| 1608 |
"step": 2650
|
| 1609 |
},
|
| 1610 |
{
|
| 1611 |
"epoch": 18.34,
|
| 1612 |
"learning_rate": 1.2631578947368422e-05,
|
| 1613 |
+
"loss": 3.1916,
|
| 1614 |
"step": 2660
|
| 1615 |
},
|
| 1616 |
{
|
| 1617 |
"epoch": 18.41,
|
| 1618 |
"learning_rate": 1.2105263157894737e-05,
|
| 1619 |
+
"loss": 3.1388,
|
| 1620 |
"step": 2670
|
| 1621 |
},
|
| 1622 |
{
|
| 1623 |
"epoch": 18.48,
|
| 1624 |
"learning_rate": 1.1578947368421053e-05,
|
| 1625 |
+
"loss": 3.1401,
|
| 1626 |
"step": 2680
|
| 1627 |
},
|
| 1628 |
{
|
| 1629 |
"epoch": 18.55,
|
| 1630 |
"learning_rate": 1.1052631578947368e-05,
|
| 1631 |
+
"loss": 3.2428,
|
| 1632 |
"step": 2690
|
| 1633 |
},
|
| 1634 |
{
|
| 1635 |
"epoch": 18.62,
|
| 1636 |
"learning_rate": 1.0526315789473684e-05,
|
| 1637 |
+
"loss": 3.1645,
|
| 1638 |
"step": 2700
|
| 1639 |
},
|
| 1640 |
{
|
| 1641 |
"epoch": 18.69,
|
| 1642 |
"learning_rate": 1e-05,
|
| 1643 |
+
"loss": 3.1607,
|
| 1644 |
"step": 2710
|
| 1645 |
},
|
| 1646 |
{
|
| 1647 |
"epoch": 18.76,
|
| 1648 |
"learning_rate": 9.473684210526317e-06,
|
| 1649 |
+
"loss": 3.2464,
|
| 1650 |
"step": 2720
|
| 1651 |
},
|
| 1652 |
{
|
| 1653 |
"epoch": 18.83,
|
| 1654 |
"learning_rate": 8.947368421052632e-06,
|
| 1655 |
+
"loss": 3.2228,
|
| 1656 |
"step": 2730
|
| 1657 |
},
|
| 1658 |
{
|
| 1659 |
"epoch": 18.9,
|
| 1660 |
"learning_rate": 8.421052631578948e-06,
|
| 1661 |
+
"loss": 3.1488,
|
| 1662 |
"step": 2740
|
| 1663 |
},
|
| 1664 |
{
|
| 1665 |
"epoch": 18.97,
|
| 1666 |
"learning_rate": 7.894736842105263e-06,
|
| 1667 |
+
"loss": 3.162,
|
| 1668 |
"step": 2750
|
| 1669 |
},
|
| 1670 |
{
|
| 1671 |
"epoch": 19.03,
|
| 1672 |
"learning_rate": 7.3684210526315784e-06,
|
| 1673 |
+
"loss": 3.2708,
|
| 1674 |
"step": 2760
|
| 1675 |
},
|
| 1676 |
{
|
| 1677 |
"epoch": 19.1,
|
| 1678 |
"learning_rate": 6.842105263157896e-06,
|
| 1679 |
+
"loss": 3.1619,
|
| 1680 |
"step": 2770
|
| 1681 |
},
|
| 1682 |
{
|
| 1683 |
"epoch": 19.17,
|
| 1684 |
"learning_rate": 6.315789473684211e-06,
|
| 1685 |
+
"loss": 3.1504,
|
| 1686 |
"step": 2780
|
| 1687 |
},
|
| 1688 |
{
|
| 1689 |
"epoch": 19.24,
|
| 1690 |
"learning_rate": 5.789473684210527e-06,
|
| 1691 |
+
"loss": 3.1647,
|
| 1692 |
"step": 2790
|
| 1693 |
},
|
| 1694 |
{
|
| 1695 |
"epoch": 19.31,
|
| 1696 |
"learning_rate": 5.263157894736842e-06,
|
| 1697 |
+
"loss": 3.2402,
|
| 1698 |
"step": 2800
|
| 1699 |
},
|
| 1700 |
{
|
| 1701 |
"epoch": 19.38,
|
| 1702 |
"learning_rate": 4.736842105263159e-06,
|
| 1703 |
+
"loss": 3.1644,
|
| 1704 |
"step": 2810
|
| 1705 |
},
|
| 1706 |
{
|
| 1707 |
"epoch": 19.45,
|
| 1708 |
"learning_rate": 4.210526315789474e-06,
|
| 1709 |
+
"loss": 3.1178,
|
| 1710 |
"step": 2820
|
| 1711 |
},
|
| 1712 |
{
|
| 1713 |
"epoch": 19.52,
|
| 1714 |
"learning_rate": 3.6842105263157892e-06,
|
| 1715 |
+
"loss": 3.2254,
|
| 1716 |
"step": 2830
|
| 1717 |
},
|
| 1718 |
{
|
| 1719 |
"epoch": 19.59,
|
| 1720 |
"learning_rate": 3.1578947368421056e-06,
|
| 1721 |
+
"loss": 3.1963,
|
| 1722 |
"step": 2840
|
| 1723 |
},
|
| 1724 |
{
|
| 1725 |
"epoch": 19.66,
|
| 1726 |
"learning_rate": 2.631578947368421e-06,
|
| 1727 |
+
"loss": 3.1382,
|
| 1728 |
"step": 2850
|
| 1729 |
},
|
| 1730 |
{
|
| 1731 |
"epoch": 19.72,
|
| 1732 |
"learning_rate": 2.105263157894737e-06,
|
| 1733 |
+
"loss": 3.1406,
|
| 1734 |
"step": 2860
|
| 1735 |
},
|
| 1736 |
{
|
| 1737 |
"epoch": 19.79,
|
| 1738 |
"learning_rate": 1.5789473684210528e-06,
|
| 1739 |
+
"loss": 3.2676,
|
| 1740 |
"step": 2870
|
| 1741 |
},
|
| 1742 |
{
|
| 1743 |
"epoch": 19.86,
|
| 1744 |
"learning_rate": 1.0526315789473685e-06,
|
| 1745 |
+
"loss": 3.1524,
|
| 1746 |
"step": 2880
|
| 1747 |
},
|
| 1748 |
{
|
| 1749 |
"epoch": 19.93,
|
| 1750 |
"learning_rate": 5.263157894736843e-07,
|
| 1751 |
+
"loss": 3.1674,
|
| 1752 |
"step": 2890
|
| 1753 |
},
|
| 1754 |
{
|
| 1755 |
"epoch": 20.0,
|
| 1756 |
"learning_rate": 0.0,
|
| 1757 |
+
"loss": 3.1682,
|
| 1758 |
"step": 2900
|
| 1759 |
},
|
| 1760 |
{
|
| 1761 |
"epoch": 20.0,
|
| 1762 |
+
"eval_loss": 3.362812042236328,
|
| 1763 |
+
"eval_runtime": 8.2276,
|
| 1764 |
+
"eval_samples_per_second": 204.191,
|
| 1765 |
+
"eval_steps_per_second": 6.442,
|
| 1766 |
+
"eval_wer": 0.9993108676176694,
|
| 1767 |
"step": 2900
|
| 1768 |
},
|
| 1769 |
{
|
| 1770 |
"epoch": 20.0,
|
| 1771 |
"step": 2900,
|
| 1772 |
"total_flos": 2.6569362344615726e+18,
|
| 1773 |
+
"train_loss": 3.8477164847275307,
|
| 1774 |
+
"train_runtime": 343.8146,
|
| 1775 |
+
"train_samples_per_second": 268.749,
|
| 1776 |
+
"train_steps_per_second": 8.435
|
| 1777 |
}
|
| 1778 |
],
|
| 1779 |
"logging_steps": 10,
|