End of training
Browse files- README.md +9 -6
- all_results.json +11 -10
- eval_results.json +6 -6
- train_results.json +6 -5
- trainer_state.json +1398 -1061
README.md
CHANGED
|
@@ -1,22 +1,25 @@
|
|
| 1 |
---
|
| 2 |
library_name: transformers
|
|
|
|
|
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: openai/whisper-medium
|
| 5 |
tags:
|
|
|
|
| 6 |
- generated_from_trainer
|
| 7 |
datasets:
|
| 8 |
-
- common_voice_17_0
|
| 9 |
metrics:
|
| 10 |
- wer
|
| 11 |
model-index:
|
| 12 |
-
- name:
|
| 13 |
results:
|
| 14 |
- task:
|
| 15 |
name: Automatic Speech Recognition
|
| 16 |
type: automatic-speech-recognition
|
| 17 |
dataset:
|
| 18 |
-
name: common_voice_17_0
|
| 19 |
-
type: common_voice_17_0
|
| 20 |
config: eu
|
| 21 |
split: test
|
| 22 |
args: eu
|
|
@@ -29,9 +32,9 @@ model-index:
|
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 30 |
should probably proofread and complete it, then remove this comment. -->
|
| 31 |
|
| 32 |
-
#
|
| 33 |
|
| 34 |
-
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the common_voice_17_0 dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
- Loss: 0.1787
|
| 37 |
- Wer: 8.8021
|
|
|
|
| 1 |
---
|
| 2 |
library_name: transformers
|
| 3 |
+
language:
|
| 4 |
+
- eu
|
| 5 |
license: apache-2.0
|
| 6 |
base_model: openai/whisper-medium
|
| 7 |
tags:
|
| 8 |
+
- whisper-event
|
| 9 |
- generated_from_trainer
|
| 10 |
datasets:
|
| 11 |
+
- mozilla-foundation/common_voice_17_0
|
| 12 |
metrics:
|
| 13 |
- wer
|
| 14 |
model-index:
|
| 15 |
+
- name: Whisper Medium Basque
|
| 16 |
results:
|
| 17 |
- task:
|
| 18 |
name: Automatic Speech Recognition
|
| 19 |
type: automatic-speech-recognition
|
| 20 |
dataset:
|
| 21 |
+
name: mozilla-foundation/common_voice_17_0 eu
|
| 22 |
+
type: mozilla-foundation/common_voice_17_0
|
| 23 |
config: eu
|
| 24 |
split: test
|
| 25 |
args: eu
|
|
|
|
| 32 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 33 |
should probably proofread and complete it, then remove this comment. -->
|
| 34 |
|
| 35 |
+
# Whisper Medium Basque
|
| 36 |
|
| 37 |
+
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the mozilla-foundation/common_voice_17_0 eu dataset.
|
| 38 |
It achieves the following results on the evaluation set:
|
| 39 |
- Loss: 0.1787
|
| 40 |
- Wer: 8.8021
|
all_results.json
CHANGED
|
@@ -1,12 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
-
"eval_wer":
|
| 8 |
-
"
|
| 9 |
-
"
|
| 10 |
-
"
|
| 11 |
-
"
|
|
|
|
| 12 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.148,
|
| 3 |
+
"eval_loss": 0.17873740196228027,
|
| 4 |
+
"eval_runtime": 2417.8154,
|
| 5 |
+
"eval_samples_per_second": 5.637,
|
| 6 |
+
"eval_steps_per_second": 0.705,
|
| 7 |
+
"eval_wer": 8.8020814247499,
|
| 8 |
+
"total_flos": 1.3062109848403968e+20,
|
| 9 |
+
"train_loss": 0.11351343880966305,
|
| 10 |
+
"train_runtime": 52381.6915,
|
| 11 |
+
"train_samples_per_second": 2.444,
|
| 12 |
+
"train_steps_per_second": 0.153
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
-
"eval_wer":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.148,
|
| 3 |
+
"eval_loss": 0.17873740196228027,
|
| 4 |
+
"eval_runtime": 2417.8154,
|
| 5 |
+
"eval_samples_per_second": 5.637,
|
| 6 |
+
"eval_steps_per_second": 0.705,
|
| 7 |
+
"eval_wer": 8.8020814247499
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,7 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
|
|
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.148,
|
| 3 |
+
"total_flos": 1.3062109848403968e+20,
|
| 4 |
+
"train_loss": 0.11351343880966305,
|
| 5 |
+
"train_runtime": 52381.6915,
|
| 6 |
+
"train_samples_per_second": 2.444,
|
| 7 |
+
"train_steps_per_second": 0.153
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,2089 +1,2426 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
"best_model_checkpoint": "./checkpoint-8000",
|
| 4 |
-
"epoch":
|
|
|
|
| 5 |
"global_step": 8000,
|
| 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.
|
| 12 |
-
"
|
| 13 |
-
"
|
|
|
|
| 14 |
"step": 25
|
| 15 |
},
|
| 16 |
{
|
| 17 |
-
"epoch": 0.
|
| 18 |
-
"
|
| 19 |
-
"
|
|
|
|
| 20 |
"step": 50
|
| 21 |
},
|
| 22 |
{
|
| 23 |
-
"epoch": 0.
|
| 24 |
-
"
|
| 25 |
-
"
|
|
|
|
| 26 |
"step": 75
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
-
"
|
| 31 |
-
"
|
|
|
|
| 32 |
"step": 100
|
| 33 |
},
|
| 34 |
{
|
| 35 |
-
"epoch": 0.
|
| 36 |
-
"
|
| 37 |
-
"
|
|
|
|
| 38 |
"step": 125
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 0.
|
| 42 |
-
"
|
| 43 |
-
"
|
|
|
|
| 44 |
"step": 150
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"
|
| 49 |
-
"
|
|
|
|
| 50 |
"step": 175
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"
|
| 55 |
-
"
|
|
|
|
| 56 |
"step": 200
|
| 57 |
},
|
| 58 |
{
|
| 59 |
-
"epoch": 0.
|
| 60 |
-
"
|
| 61 |
-
"
|
|
|
|
| 62 |
"step": 225
|
| 63 |
},
|
| 64 |
{
|
| 65 |
-
"epoch": 0.
|
| 66 |
-
"
|
| 67 |
-
"
|
|
|
|
| 68 |
"step": 250
|
| 69 |
},
|
| 70 |
{
|
| 71 |
-
"epoch": 0.
|
| 72 |
-
"
|
| 73 |
-
"
|
|
|
|
| 74 |
"step": 275
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch": 0.
|
| 78 |
-
"
|
| 79 |
-
"
|
|
|
|
| 80 |
"step": 300
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch": 0.
|
| 84 |
-
"
|
| 85 |
-
"
|
|
|
|
| 86 |
"step": 325
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"
|
| 91 |
-
"
|
|
|
|
| 92 |
"step": 350
|
| 93 |
},
|
| 94 |
{
|
| 95 |
-
"epoch": 0.
|
| 96 |
-
"
|
| 97 |
-
"
|
|
|
|
| 98 |
"step": 375
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.05,
|
| 102 |
-
"
|
| 103 |
-
"
|
|
|
|
| 104 |
"step": 400
|
| 105 |
},
|
| 106 |
{
|
| 107 |
-
"epoch": 0.
|
| 108 |
-
"
|
| 109 |
-
"
|
|
|
|
| 110 |
"step": 425
|
| 111 |
},
|
| 112 |
{
|
| 113 |
-
"epoch": 0.
|
| 114 |
-
"
|
| 115 |
-
"
|
|
|
|
| 116 |
"step": 450
|
| 117 |
},
|
| 118 |
{
|
| 119 |
-
"epoch": 0.
|
| 120 |
-
"
|
| 121 |
-
"
|
|
|
|
| 122 |
"step": 475
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch": 0.
|
| 126 |
-
"
|
| 127 |
-
"
|
|
|
|
| 128 |
"step": 500
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"eval_loss": 0.
|
| 133 |
-
"eval_runtime":
|
| 134 |
-
"eval_samples_per_second":
|
| 135 |
-
"eval_steps_per_second": 0.
|
| 136 |
-
"eval_wer":
|
| 137 |
"step": 500
|
| 138 |
},
|
| 139 |
{
|
| 140 |
-
"epoch": 0.
|
| 141 |
-
"
|
| 142 |
-
"
|
|
|
|
| 143 |
"step": 525
|
| 144 |
},
|
| 145 |
{
|
| 146 |
-
"epoch": 0.
|
| 147 |
-
"
|
| 148 |
-
"
|
|
|
|
| 149 |
"step": 550
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"
|
| 154 |
-
"
|
|
|
|
| 155 |
"step": 575
|
| 156 |
},
|
| 157 |
{
|
| 158 |
-
"epoch": 0.
|
| 159 |
-
"
|
| 160 |
-
"
|
|
|
|
| 161 |
"step": 600
|
| 162 |
},
|
| 163 |
{
|
| 164 |
-
"epoch": 0.
|
| 165 |
-
"
|
| 166 |
-
"
|
|
|
|
| 167 |
"step": 625
|
| 168 |
},
|
| 169 |
{
|
| 170 |
-
"epoch": 0.
|
| 171 |
-
"
|
| 172 |
-
"
|
|
|
|
| 173 |
"step": 650
|
| 174 |
},
|
| 175 |
{
|
| 176 |
-
"epoch": 0.
|
| 177 |
-
"
|
| 178 |
-
"
|
|
|
|
| 179 |
"step": 675
|
| 180 |
},
|
| 181 |
{
|
| 182 |
-
"epoch": 0.
|
| 183 |
-
"
|
| 184 |
-
"
|
|
|
|
| 185 |
"step": 700
|
| 186 |
},
|
| 187 |
{
|
| 188 |
-
"epoch": 0.
|
| 189 |
-
"
|
| 190 |
-
"
|
|
|
|
| 191 |
"step": 725
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch": 0.
|
| 195 |
-
"
|
| 196 |
-
"
|
|
|
|
| 197 |
"step": 750
|
| 198 |
},
|
| 199 |
{
|
| 200 |
-
"epoch": 0.
|
| 201 |
-
"
|
| 202 |
-
"
|
|
|
|
| 203 |
"step": 775
|
| 204 |
},
|
| 205 |
{
|
| 206 |
"epoch": 0.1,
|
| 207 |
-
"
|
| 208 |
-
"
|
|
|
|
| 209 |
"step": 800
|
| 210 |
},
|
| 211 |
{
|
| 212 |
-
"epoch": 0.
|
| 213 |
-
"
|
| 214 |
-
"
|
|
|
|
| 215 |
"step": 825
|
| 216 |
},
|
| 217 |
{
|
| 218 |
-
"epoch": 0.
|
| 219 |
-
"
|
| 220 |
-
"
|
|
|
|
| 221 |
"step": 850
|
| 222 |
},
|
| 223 |
{
|
| 224 |
-
"epoch": 0.
|
| 225 |
-
"
|
| 226 |
-
"
|
|
|
|
| 227 |
"step": 875
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"epoch": 0.
|
| 231 |
-
"
|
| 232 |
-
"
|
|
|
|
| 233 |
"step": 900
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch": 0.
|
| 237 |
-
"
|
| 238 |
-
"
|
|
|
|
| 239 |
"step": 925
|
| 240 |
},
|
| 241 |
{
|
| 242 |
-
"epoch": 0.
|
| 243 |
-
"
|
| 244 |
-
"
|
|
|
|
| 245 |
"step": 950
|
| 246 |
},
|
| 247 |
{
|
| 248 |
-
"epoch": 0.
|
| 249 |
-
"
|
| 250 |
-
"
|
|
|
|
| 251 |
"step": 975
|
| 252 |
},
|
| 253 |
{
|
| 254 |
-
"epoch": 0.
|
| 255 |
-
"
|
| 256 |
-
"
|
|
|
|
| 257 |
"step": 1000
|
| 258 |
},
|
| 259 |
{
|
| 260 |
-
"epoch": 0.
|
| 261 |
-
"eval_loss": 0.
|
| 262 |
-
"eval_runtime":
|
| 263 |
-
"eval_samples_per_second":
|
| 264 |
-
"eval_steps_per_second": 0.
|
| 265 |
-
"eval_wer":
|
| 266 |
"step": 1000
|
| 267 |
},
|
| 268 |
{
|
| 269 |
-
"epoch": 0.
|
| 270 |
-
"
|
| 271 |
-
"
|
|
|
|
| 272 |
"step": 1025
|
| 273 |
},
|
| 274 |
{
|
| 275 |
-
"epoch": 0.
|
| 276 |
-
"
|
| 277 |
-
"
|
|
|
|
| 278 |
"step": 1050
|
| 279 |
},
|
| 280 |
{
|
| 281 |
-
"epoch": 0.
|
| 282 |
-
"
|
| 283 |
-
"
|
|
|
|
| 284 |
"step": 1075
|
| 285 |
},
|
| 286 |
{
|
| 287 |
-
"epoch": 0.
|
| 288 |
-
"
|
| 289 |
-
"
|
|
|
|
| 290 |
"step": 1100
|
| 291 |
},
|
| 292 |
{
|
| 293 |
-
"epoch": 0.
|
| 294 |
-
"
|
| 295 |
-
"
|
|
|
|
| 296 |
"step": 1125
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"epoch": 0.
|
| 300 |
-
"
|
| 301 |
-
"
|
|
|
|
| 302 |
"step": 1150
|
| 303 |
},
|
| 304 |
{
|
| 305 |
-
"epoch": 0.
|
| 306 |
-
"
|
| 307 |
-
"
|
|
|
|
| 308 |
"step": 1175
|
| 309 |
},
|
| 310 |
{
|
| 311 |
"epoch": 0.15,
|
| 312 |
-
"
|
| 313 |
-
"
|
|
|
|
| 314 |
"step": 1200
|
| 315 |
},
|
| 316 |
{
|
| 317 |
-
"epoch": 0.
|
| 318 |
-
"
|
| 319 |
-
"
|
|
|
|
| 320 |
"step": 1225
|
| 321 |
},
|
| 322 |
{
|
| 323 |
-
"epoch": 0.
|
| 324 |
-
"
|
| 325 |
-
"
|
|
|
|
| 326 |
"step": 1250
|
| 327 |
},
|
| 328 |
{
|
| 329 |
-
"epoch": 0.
|
| 330 |
-
"
|
| 331 |
-
"
|
|
|
|
| 332 |
"step": 1275
|
| 333 |
},
|
| 334 |
{
|
| 335 |
-
"epoch": 0.
|
| 336 |
-
"
|
| 337 |
-
"
|
|
|
|
| 338 |
"step": 1300
|
| 339 |
},
|
| 340 |
{
|
| 341 |
-
"epoch": 0.
|
| 342 |
-
"
|
| 343 |
-
"
|
|
|
|
| 344 |
"step": 1325
|
| 345 |
},
|
| 346 |
{
|
| 347 |
-
"epoch": 0.
|
| 348 |
-
"
|
| 349 |
-
"
|
|
|
|
| 350 |
"step": 1350
|
| 351 |
},
|
| 352 |
{
|
| 353 |
-
"epoch": 0.
|
| 354 |
-
"
|
| 355 |
-
"
|
|
|
|
| 356 |
"step": 1375
|
| 357 |
},
|
| 358 |
{
|
| 359 |
-
"epoch": 0.
|
| 360 |
-
"
|
| 361 |
-
"
|
|
|
|
| 362 |
"step": 1400
|
| 363 |
},
|
| 364 |
{
|
| 365 |
-
"epoch": 0.
|
| 366 |
-
"
|
| 367 |
-
"
|
|
|
|
| 368 |
"step": 1425
|
| 369 |
},
|
| 370 |
{
|
| 371 |
-
"epoch": 0.
|
| 372 |
-
"
|
| 373 |
-
"
|
|
|
|
| 374 |
"step": 1450
|
| 375 |
},
|
| 376 |
{
|
| 377 |
-
"epoch": 0.
|
| 378 |
-
"
|
| 379 |
-
"
|
|
|
|
| 380 |
"step": 1475
|
| 381 |
},
|
| 382 |
{
|
| 383 |
-
"epoch": 0.
|
| 384 |
-
"
|
| 385 |
-
"
|
|
|
|
| 386 |
"step": 1500
|
| 387 |
},
|
| 388 |
{
|
| 389 |
-
"epoch": 0.
|
| 390 |
-
"eval_loss": 0.
|
| 391 |
-
"eval_runtime":
|
| 392 |
-
"eval_samples_per_second":
|
| 393 |
-
"eval_steps_per_second": 0.
|
| 394 |
-
"eval_wer":
|
| 395 |
"step": 1500
|
| 396 |
},
|
| 397 |
{
|
| 398 |
-
"epoch": 0.
|
| 399 |
-
"
|
| 400 |
-
"
|
|
|
|
| 401 |
"step": 1525
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch": 0.
|
| 405 |
-
"
|
| 406 |
-
"
|
|
|
|
| 407 |
"step": 1550
|
| 408 |
},
|
| 409 |
{
|
| 410 |
-
"epoch": 0.
|
| 411 |
-
"
|
| 412 |
-
"
|
|
|
|
| 413 |
"step": 1575
|
| 414 |
},
|
| 415 |
{
|
| 416 |
"epoch": 0.2,
|
| 417 |
-
"
|
| 418 |
-
"
|
|
|
|
| 419 |
"step": 1600
|
| 420 |
},
|
| 421 |
{
|
| 422 |
-
"epoch": 0.
|
| 423 |
-
"
|
| 424 |
-
"
|
|
|
|
| 425 |
"step": 1625
|
| 426 |
},
|
| 427 |
{
|
| 428 |
-
"epoch": 0.
|
| 429 |
-
"
|
| 430 |
-
"
|
|
|
|
| 431 |
"step": 1650
|
| 432 |
},
|
| 433 |
{
|
| 434 |
-
"epoch": 0.
|
| 435 |
-
"
|
| 436 |
-
"
|
|
|
|
| 437 |
"step": 1675
|
| 438 |
},
|
| 439 |
{
|
| 440 |
-
"epoch": 0.
|
| 441 |
-
"
|
| 442 |
-
"
|
|
|
|
| 443 |
"step": 1700
|
| 444 |
},
|
| 445 |
{
|
| 446 |
-
"epoch":
|
| 447 |
-
"
|
| 448 |
-
"
|
|
|
|
| 449 |
"step": 1725
|
| 450 |
},
|
| 451 |
{
|
| 452 |
-
"epoch":
|
| 453 |
-
"
|
| 454 |
-
"
|
|
|
|
| 455 |
"step": 1750
|
| 456 |
},
|
| 457 |
{
|
| 458 |
-
"epoch":
|
| 459 |
-
"
|
| 460 |
-
"
|
|
|
|
| 461 |
"step": 1775
|
| 462 |
},
|
| 463 |
{
|
| 464 |
-
"epoch":
|
| 465 |
-
"
|
| 466 |
-
"
|
|
|
|
| 467 |
"step": 1800
|
| 468 |
},
|
| 469 |
{
|
| 470 |
-
"epoch":
|
| 471 |
-
"
|
| 472 |
-
"
|
|
|
|
| 473 |
"step": 1825
|
| 474 |
},
|
| 475 |
{
|
| 476 |
-
"epoch":
|
| 477 |
-
"
|
| 478 |
-
"
|
|
|
|
| 479 |
"step": 1850
|
| 480 |
},
|
| 481 |
{
|
| 482 |
-
"epoch":
|
| 483 |
-
"
|
| 484 |
-
"
|
|
|
|
| 485 |
"step": 1875
|
| 486 |
},
|
| 487 |
{
|
| 488 |
-
"epoch":
|
| 489 |
-
"
|
| 490 |
-
"
|
|
|
|
| 491 |
"step": 1900
|
| 492 |
},
|
| 493 |
{
|
| 494 |
-
"epoch":
|
| 495 |
-
"
|
| 496 |
-
"
|
|
|
|
| 497 |
"step": 1925
|
| 498 |
},
|
| 499 |
{
|
| 500 |
-
"epoch":
|
| 501 |
-
"
|
| 502 |
-
"
|
|
|
|
| 503 |
"step": 1950
|
| 504 |
},
|
| 505 |
{
|
| 506 |
-
"epoch":
|
| 507 |
-
"
|
| 508 |
-
"
|
|
|
|
| 509 |
"step": 1975
|
| 510 |
},
|
| 511 |
{
|
| 512 |
-
"epoch":
|
| 513 |
-
"
|
| 514 |
-
"
|
|
|
|
| 515 |
"step": 2000
|
| 516 |
},
|
| 517 |
{
|
| 518 |
-
"epoch":
|
| 519 |
-
"eval_loss": 0.
|
| 520 |
-
"eval_runtime":
|
| 521 |
-
"eval_samples_per_second":
|
| 522 |
-
"eval_steps_per_second": 0.
|
| 523 |
-
"eval_wer":
|
| 524 |
"step": 2000
|
| 525 |
},
|
| 526 |
{
|
| 527 |
-
"epoch":
|
| 528 |
-
"
|
| 529 |
-
"
|
|
|
|
| 530 |
"step": 2025
|
| 531 |
},
|
| 532 |
{
|
| 533 |
-
"epoch":
|
| 534 |
-
"
|
| 535 |
-
"
|
|
|
|
| 536 |
"step": 2050
|
| 537 |
},
|
| 538 |
{
|
| 539 |
-
"epoch":
|
| 540 |
-
"
|
| 541 |
-
"
|
|
|
|
| 542 |
"step": 2075
|
| 543 |
},
|
| 544 |
{
|
| 545 |
-
"epoch":
|
| 546 |
-
"
|
| 547 |
-
"
|
|
|
|
| 548 |
"step": 2100
|
| 549 |
},
|
| 550 |
{
|
| 551 |
-
"epoch":
|
| 552 |
-
"
|
| 553 |
-
"
|
|
|
|
| 554 |
"step": 2125
|
| 555 |
},
|
| 556 |
{
|
| 557 |
-
"epoch":
|
| 558 |
-
"
|
| 559 |
-
"
|
|
|
|
| 560 |
"step": 2150
|
| 561 |
},
|
| 562 |
{
|
| 563 |
-
"epoch":
|
| 564 |
-
"
|
| 565 |
-
"
|
|
|
|
| 566 |
"step": 2175
|
| 567 |
},
|
| 568 |
{
|
| 569 |
-
"epoch":
|
| 570 |
-
"
|
| 571 |
-
"
|
|
|
|
| 572 |
"step": 2200
|
| 573 |
},
|
| 574 |
{
|
| 575 |
-
"epoch":
|
| 576 |
-
"
|
| 577 |
-
"
|
|
|
|
| 578 |
"step": 2225
|
| 579 |
},
|
| 580 |
{
|
| 581 |
-
"epoch":
|
| 582 |
-
"
|
| 583 |
-
"
|
|
|
|
| 584 |
"step": 2250
|
| 585 |
},
|
| 586 |
{
|
| 587 |
-
"epoch":
|
| 588 |
-
"
|
| 589 |
-
"
|
|
|
|
| 590 |
"step": 2275
|
| 591 |
},
|
| 592 |
{
|
| 593 |
-
"epoch":
|
| 594 |
-
"
|
| 595 |
-
"
|
|
|
|
| 596 |
"step": 2300
|
| 597 |
},
|
| 598 |
{
|
| 599 |
-
"epoch":
|
| 600 |
-
"
|
| 601 |
-
"
|
|
|
|
| 602 |
"step": 2325
|
| 603 |
},
|
| 604 |
{
|
| 605 |
-
"epoch":
|
| 606 |
-
"
|
| 607 |
-
"
|
|
|
|
| 608 |
"step": 2350
|
| 609 |
},
|
| 610 |
{
|
| 611 |
-
"epoch":
|
| 612 |
-
"
|
| 613 |
-
"
|
|
|
|
| 614 |
"step": 2375
|
| 615 |
},
|
| 616 |
{
|
| 617 |
-
"epoch":
|
| 618 |
-
"
|
| 619 |
-
"
|
|
|
|
| 620 |
"step": 2400
|
| 621 |
},
|
| 622 |
{
|
| 623 |
-
"epoch":
|
| 624 |
-
"
|
| 625 |
-
"
|
|
|
|
| 626 |
"step": 2425
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch":
|
| 630 |
-
"
|
| 631 |
-
"
|
|
|
|
| 632 |
"step": 2450
|
| 633 |
},
|
| 634 |
{
|
| 635 |
-
"epoch":
|
| 636 |
-
"
|
| 637 |
-
"
|
|
|
|
| 638 |
"step": 2475
|
| 639 |
},
|
| 640 |
{
|
| 641 |
-
"epoch":
|
| 642 |
-
"
|
| 643 |
-
"
|
|
|
|
| 644 |
"step": 2500
|
| 645 |
},
|
| 646 |
{
|
| 647 |
-
"epoch":
|
| 648 |
-
"eval_loss": 0.
|
| 649 |
-
"eval_runtime":
|
| 650 |
-
"eval_samples_per_second":
|
| 651 |
-
"eval_steps_per_second": 0.
|
| 652 |
-
"eval_wer":
|
| 653 |
"step": 2500
|
| 654 |
},
|
| 655 |
{
|
| 656 |
-
"epoch":
|
| 657 |
-
"
|
| 658 |
-
"
|
|
|
|
| 659 |
"step": 2525
|
| 660 |
},
|
| 661 |
{
|
| 662 |
-
"epoch":
|
| 663 |
-
"
|
| 664 |
-
"
|
|
|
|
| 665 |
"step": 2550
|
| 666 |
},
|
| 667 |
{
|
| 668 |
-
"epoch":
|
| 669 |
-
"
|
| 670 |
-
"
|
|
|
|
| 671 |
"step": 2575
|
| 672 |
},
|
| 673 |
{
|
| 674 |
-
"epoch":
|
| 675 |
-
"
|
| 676 |
-
"
|
|
|
|
| 677 |
"step": 2600
|
| 678 |
},
|
| 679 |
{
|
| 680 |
-
"epoch":
|
| 681 |
-
"
|
| 682 |
-
"
|
|
|
|
| 683 |
"step": 2625
|
| 684 |
},
|
| 685 |
{
|
| 686 |
-
"epoch":
|
| 687 |
-
"
|
| 688 |
-
"
|
|
|
|
| 689 |
"step": 2650
|
| 690 |
},
|
| 691 |
{
|
| 692 |
-
"epoch":
|
| 693 |
-
"
|
| 694 |
-
"
|
|
|
|
| 695 |
"step": 2675
|
| 696 |
},
|
| 697 |
{
|
| 698 |
-
"epoch":
|
| 699 |
-
"
|
| 700 |
-
"
|
|
|
|
| 701 |
"step": 2700
|
| 702 |
},
|
| 703 |
{
|
| 704 |
-
"epoch":
|
| 705 |
-
"
|
| 706 |
-
"
|
|
|
|
| 707 |
"step": 2725
|
| 708 |
},
|
| 709 |
{
|
| 710 |
-
"epoch":
|
| 711 |
-
"
|
| 712 |
-
"
|
|
|
|
| 713 |
"step": 2750
|
| 714 |
},
|
| 715 |
{
|
| 716 |
-
"epoch":
|
| 717 |
-
"
|
| 718 |
-
"
|
|
|
|
| 719 |
"step": 2775
|
| 720 |
},
|
| 721 |
{
|
| 722 |
-
"epoch":
|
| 723 |
-
"
|
| 724 |
-
"
|
|
|
|
| 725 |
"step": 2800
|
| 726 |
},
|
| 727 |
{
|
| 728 |
-
"epoch":
|
| 729 |
-
"
|
| 730 |
-
"
|
|
|
|
| 731 |
"step": 2825
|
| 732 |
},
|
| 733 |
{
|
| 734 |
-
"epoch":
|
| 735 |
-
"
|
| 736 |
-
"
|
|
|
|
| 737 |
"step": 2850
|
| 738 |
},
|
| 739 |
{
|
| 740 |
-
"epoch":
|
| 741 |
-
"
|
| 742 |
-
"
|
|
|
|
| 743 |
"step": 2875
|
| 744 |
},
|
| 745 |
{
|
| 746 |
-
"epoch":
|
| 747 |
-
"
|
| 748 |
-
"
|
|
|
|
| 749 |
"step": 2900
|
| 750 |
},
|
| 751 |
{
|
| 752 |
-
"epoch":
|
| 753 |
-
"
|
| 754 |
-
"
|
|
|
|
| 755 |
"step": 2925
|
| 756 |
},
|
| 757 |
{
|
| 758 |
-
"epoch":
|
| 759 |
-
"
|
| 760 |
-
"
|
|
|
|
| 761 |
"step": 2950
|
| 762 |
},
|
| 763 |
{
|
| 764 |
-
"epoch":
|
| 765 |
-
"
|
| 766 |
-
"
|
|
|
|
| 767 |
"step": 2975
|
| 768 |
},
|
| 769 |
{
|
| 770 |
-
"epoch":
|
| 771 |
-
"
|
| 772 |
-
"
|
|
|
|
| 773 |
"step": 3000
|
| 774 |
},
|
| 775 |
{
|
| 776 |
-
"epoch":
|
| 777 |
-
"eval_loss": 0.
|
| 778 |
-
"eval_runtime":
|
| 779 |
-
"eval_samples_per_second":
|
| 780 |
-
"eval_steps_per_second": 0.
|
| 781 |
-
"eval_wer":
|
| 782 |
"step": 3000
|
| 783 |
},
|
| 784 |
{
|
| 785 |
-
"epoch":
|
| 786 |
-
"
|
| 787 |
-
"
|
|
|
|
| 788 |
"step": 3025
|
| 789 |
},
|
| 790 |
{
|
| 791 |
-
"epoch":
|
| 792 |
-
"
|
| 793 |
-
"
|
|
|
|
| 794 |
"step": 3050
|
| 795 |
},
|
| 796 |
{
|
| 797 |
-
"epoch":
|
| 798 |
-
"
|
| 799 |
-
"
|
|
|
|
| 800 |
"step": 3075
|
| 801 |
},
|
| 802 |
{
|
| 803 |
-
"epoch":
|
| 804 |
-
"
|
| 805 |
-
"
|
|
|
|
| 806 |
"step": 3100
|
| 807 |
},
|
| 808 |
{
|
| 809 |
-
"epoch":
|
| 810 |
-
"
|
| 811 |
-
"
|
|
|
|
| 812 |
"step": 3125
|
| 813 |
},
|
| 814 |
{
|
| 815 |
-
"epoch":
|
| 816 |
-
"
|
| 817 |
-
"
|
|
|
|
| 818 |
"step": 3150
|
| 819 |
},
|
| 820 |
{
|
| 821 |
-
"epoch":
|
| 822 |
-
"
|
| 823 |
-
"
|
|
|
|
| 824 |
"step": 3175
|
| 825 |
},
|
| 826 |
{
|
| 827 |
-
"epoch":
|
| 828 |
-
"
|
| 829 |
-
"
|
|
|
|
| 830 |
"step": 3200
|
| 831 |
},
|
| 832 |
{
|
| 833 |
-
"epoch":
|
| 834 |
-
"
|
| 835 |
-
"
|
|
|
|
| 836 |
"step": 3225
|
| 837 |
},
|
| 838 |
{
|
| 839 |
-
"epoch":
|
| 840 |
-
"
|
| 841 |
-
"
|
|
|
|
| 842 |
"step": 3250
|
| 843 |
},
|
| 844 |
{
|
| 845 |
-
"epoch":
|
| 846 |
-
"
|
| 847 |
-
"
|
|
|
|
| 848 |
"step": 3275
|
| 849 |
},
|
| 850 |
{
|
| 851 |
-
"epoch":
|
| 852 |
-
"
|
| 853 |
-
"
|
|
|
|
| 854 |
"step": 3300
|
| 855 |
},
|
| 856 |
{
|
| 857 |
-
"epoch":
|
| 858 |
-
"
|
| 859 |
-
"
|
|
|
|
| 860 |
"step": 3325
|
| 861 |
},
|
| 862 |
{
|
| 863 |
-
"epoch":
|
| 864 |
-
"
|
| 865 |
-
"
|
|
|
|
| 866 |
"step": 3350
|
| 867 |
},
|
| 868 |
{
|
| 869 |
-
"epoch":
|
| 870 |
-
"
|
| 871 |
-
"
|
|
|
|
| 872 |
"step": 3375
|
| 873 |
},
|
| 874 |
{
|
| 875 |
-
"epoch":
|
| 876 |
-
"
|
| 877 |
-
"
|
|
|
|
| 878 |
"step": 3400
|
| 879 |
},
|
| 880 |
{
|
| 881 |
-
"epoch":
|
| 882 |
-
"
|
| 883 |
-
"
|
|
|
|
| 884 |
"step": 3425
|
| 885 |
},
|
| 886 |
{
|
| 887 |
-
"epoch":
|
| 888 |
-
"
|
| 889 |
-
"
|
|
|
|
| 890 |
"step": 3450
|
| 891 |
},
|
| 892 |
{
|
| 893 |
-
"epoch":
|
| 894 |
-
"
|
| 895 |
-
"
|
|
|
|
| 896 |
"step": 3475
|
| 897 |
},
|
| 898 |
{
|
| 899 |
-
"epoch":
|
| 900 |
-
"
|
| 901 |
-
"
|
|
|
|
| 902 |
"step": 3500
|
| 903 |
},
|
| 904 |
{
|
| 905 |
-
"epoch":
|
| 906 |
-
"eval_loss": 0.
|
| 907 |
-
"eval_runtime":
|
| 908 |
-
"eval_samples_per_second":
|
| 909 |
-
"eval_steps_per_second": 0.
|
| 910 |
-
"eval_wer":
|
| 911 |
"step": 3500
|
| 912 |
},
|
| 913 |
{
|
| 914 |
-
"epoch":
|
| 915 |
-
"
|
| 916 |
-
"
|
|
|
|
| 917 |
"step": 3525
|
| 918 |
},
|
| 919 |
{
|
| 920 |
-
"epoch":
|
| 921 |
-
"
|
| 922 |
-
"
|
|
|
|
| 923 |
"step": 3550
|
| 924 |
},
|
| 925 |
{
|
| 926 |
-
"epoch":
|
| 927 |
-
"
|
| 928 |
-
"
|
|
|
|
| 929 |
"step": 3575
|
| 930 |
},
|
| 931 |
{
|
| 932 |
-
"epoch":
|
| 933 |
-
"
|
| 934 |
-
"
|
|
|
|
| 935 |
"step": 3600
|
| 936 |
},
|
| 937 |
{
|
| 938 |
-
"epoch":
|
| 939 |
-
"
|
| 940 |
-
"
|
|
|
|
| 941 |
"step": 3625
|
| 942 |
},
|
| 943 |
{
|
| 944 |
-
"epoch":
|
| 945 |
-
"
|
| 946 |
-
"
|
|
|
|
| 947 |
"step": 3650
|
| 948 |
},
|
| 949 |
{
|
| 950 |
-
"epoch":
|
| 951 |
-
"
|
| 952 |
-
"
|
|
|
|
| 953 |
"step": 3675
|
| 954 |
},
|
| 955 |
{
|
| 956 |
-
"epoch":
|
| 957 |
-
"
|
| 958 |
-
"
|
|
|
|
| 959 |
"step": 3700
|
| 960 |
},
|
| 961 |
{
|
| 962 |
-
"epoch":
|
| 963 |
-
"
|
| 964 |
-
"
|
|
|
|
| 965 |
"step": 3725
|
| 966 |
},
|
| 967 |
{
|
| 968 |
-
"epoch":
|
| 969 |
-
"
|
| 970 |
-
"
|
|
|
|
| 971 |
"step": 3750
|
| 972 |
},
|
| 973 |
{
|
| 974 |
-
"epoch":
|
| 975 |
-
"
|
| 976 |
-
"
|
|
|
|
| 977 |
"step": 3775
|
| 978 |
},
|
| 979 |
{
|
| 980 |
-
"epoch":
|
| 981 |
-
"
|
| 982 |
-
"
|
|
|
|
| 983 |
"step": 3800
|
| 984 |
},
|
| 985 |
{
|
| 986 |
-
"epoch":
|
| 987 |
-
"
|
| 988 |
-
"
|
|
|
|
| 989 |
"step": 3825
|
| 990 |
},
|
| 991 |
{
|
| 992 |
-
"epoch":
|
| 993 |
-
"
|
| 994 |
-
"
|
|
|
|
| 995 |
"step": 3850
|
| 996 |
},
|
| 997 |
{
|
| 998 |
-
"epoch":
|
| 999 |
-
"
|
| 1000 |
-
"
|
|
|
|
| 1001 |
"step": 3875
|
| 1002 |
},
|
| 1003 |
{
|
| 1004 |
-
"epoch":
|
| 1005 |
-
"
|
| 1006 |
-
"
|
|
|
|
| 1007 |
"step": 3900
|
| 1008 |
},
|
| 1009 |
{
|
| 1010 |
-
"epoch":
|
| 1011 |
-
"
|
| 1012 |
-
"
|
|
|
|
| 1013 |
"step": 3925
|
| 1014 |
},
|
| 1015 |
{
|
| 1016 |
-
"epoch":
|
| 1017 |
-
"
|
| 1018 |
-
"
|
|
|
|
| 1019 |
"step": 3950
|
| 1020 |
},
|
| 1021 |
{
|
| 1022 |
-
"epoch":
|
| 1023 |
-
"
|
| 1024 |
-
"
|
|
|
|
| 1025 |
"step": 3975
|
| 1026 |
},
|
| 1027 |
{
|
| 1028 |
-
"epoch":
|
| 1029 |
-
"
|
| 1030 |
-
"
|
|
|
|
| 1031 |
"step": 4000
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
-
"epoch":
|
| 1035 |
-
"eval_loss": 0.
|
| 1036 |
-
"eval_runtime":
|
| 1037 |
-
"eval_samples_per_second":
|
| 1038 |
-
"eval_steps_per_second": 0.
|
| 1039 |
-
"eval_wer":
|
| 1040 |
"step": 4000
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
-
"epoch":
|
| 1044 |
-
"
|
| 1045 |
-
"
|
|
|
|
| 1046 |
"step": 4025
|
| 1047 |
},
|
| 1048 |
{
|
| 1049 |
-
"epoch":
|
| 1050 |
-
"
|
| 1051 |
-
"
|
|
|
|
| 1052 |
"step": 4050
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
-
"epoch":
|
| 1056 |
-
"
|
| 1057 |
-
"
|
|
|
|
| 1058 |
"step": 4075
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
-
"epoch":
|
| 1062 |
-
"
|
| 1063 |
-
"
|
|
|
|
| 1064 |
"step": 4100
|
| 1065 |
},
|
| 1066 |
{
|
| 1067 |
-
"epoch":
|
| 1068 |
-
"
|
| 1069 |
-
"
|
|
|
|
| 1070 |
"step": 4125
|
| 1071 |
},
|
| 1072 |
{
|
| 1073 |
-
"epoch":
|
| 1074 |
-
"
|
| 1075 |
-
"
|
|
|
|
| 1076 |
"step": 4150
|
| 1077 |
},
|
| 1078 |
{
|
| 1079 |
-
"epoch":
|
| 1080 |
-
"
|
| 1081 |
-
"
|
|
|
|
| 1082 |
"step": 4175
|
| 1083 |
},
|
| 1084 |
{
|
| 1085 |
-
"epoch":
|
| 1086 |
-
"
|
| 1087 |
-
"
|
|
|
|
| 1088 |
"step": 4200
|
| 1089 |
},
|
| 1090 |
{
|
| 1091 |
-
"epoch":
|
| 1092 |
-
"
|
| 1093 |
-
"
|
|
|
|
| 1094 |
"step": 4225
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
-
"epoch":
|
| 1098 |
-
"
|
| 1099 |
-
"
|
|
|
|
| 1100 |
"step": 4250
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
-
"epoch":
|
| 1104 |
-
"
|
| 1105 |
-
"
|
|
|
|
| 1106 |
"step": 4275
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
-
"epoch":
|
| 1110 |
-
"
|
| 1111 |
-
"
|
|
|
|
| 1112 |
"step": 4300
|
| 1113 |
},
|
| 1114 |
{
|
| 1115 |
-
"epoch":
|
| 1116 |
-
"
|
| 1117 |
-
"
|
|
|
|
| 1118 |
"step": 4325
|
| 1119 |
},
|
| 1120 |
{
|
| 1121 |
-
"epoch":
|
| 1122 |
-
"
|
| 1123 |
-
"
|
|
|
|
| 1124 |
"step": 4350
|
| 1125 |
},
|
| 1126 |
{
|
| 1127 |
-
"epoch":
|
| 1128 |
-
"
|
| 1129 |
-
"
|
|
|
|
| 1130 |
"step": 4375
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
-
"epoch":
|
| 1134 |
-
"
|
| 1135 |
-
"
|
|
|
|
| 1136 |
"step": 4400
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
-
"epoch":
|
| 1140 |
-
"
|
| 1141 |
-
"
|
|
|
|
| 1142 |
"step": 4425
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
-
"epoch":
|
| 1146 |
-
"
|
| 1147 |
-
"
|
|
|
|
| 1148 |
"step": 4450
|
| 1149 |
},
|
| 1150 |
{
|
| 1151 |
-
"epoch":
|
| 1152 |
-
"
|
| 1153 |
-
"
|
|
|
|
| 1154 |
"step": 4475
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
-
"epoch":
|
| 1158 |
-
"
|
| 1159 |
-
"
|
|
|
|
| 1160 |
"step": 4500
|
| 1161 |
},
|
| 1162 |
{
|
| 1163 |
-
"epoch":
|
| 1164 |
-
"eval_loss": 0.
|
| 1165 |
-
"eval_runtime":
|
| 1166 |
-
"eval_samples_per_second":
|
| 1167 |
-
"eval_steps_per_second": 0.
|
| 1168 |
-
"eval_wer":
|
| 1169 |
"step": 4500
|
| 1170 |
},
|
| 1171 |
{
|
| 1172 |
-
"epoch":
|
| 1173 |
-
"
|
| 1174 |
-
"
|
|
|
|
| 1175 |
"step": 4525
|
| 1176 |
},
|
| 1177 |
{
|
| 1178 |
-
"epoch":
|
| 1179 |
-
"
|
| 1180 |
-
"
|
|
|
|
| 1181 |
"step": 4550
|
| 1182 |
},
|
| 1183 |
{
|
| 1184 |
-
"epoch":
|
| 1185 |
-
"
|
| 1186 |
-
"
|
|
|
|
| 1187 |
"step": 4575
|
| 1188 |
},
|
| 1189 |
{
|
| 1190 |
-
"epoch":
|
| 1191 |
-
"
|
| 1192 |
-
"
|
|
|
|
| 1193 |
"step": 4600
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
-
"epoch":
|
| 1197 |
-
"
|
| 1198 |
-
"
|
|
|
|
| 1199 |
"step": 4625
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
-
"epoch":
|
| 1203 |
-
"
|
| 1204 |
-
"
|
|
|
|
| 1205 |
"step": 4650
|
| 1206 |
},
|
| 1207 |
{
|
| 1208 |
-
"epoch":
|
| 1209 |
-
"
|
| 1210 |
-
"
|
|
|
|
| 1211 |
"step": 4675
|
| 1212 |
},
|
| 1213 |
{
|
| 1214 |
-
"epoch":
|
| 1215 |
-
"
|
| 1216 |
-
"
|
|
|
|
| 1217 |
"step": 4700
|
| 1218 |
},
|
| 1219 |
{
|
| 1220 |
-
"epoch":
|
| 1221 |
-
"
|
| 1222 |
-
"
|
|
|
|
| 1223 |
"step": 4725
|
| 1224 |
},
|
| 1225 |
{
|
| 1226 |
-
"epoch":
|
| 1227 |
-
"
|
| 1228 |
-
"
|
|
|
|
| 1229 |
"step": 4750
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
-
"epoch":
|
| 1233 |
-
"
|
| 1234 |
-
"
|
|
|
|
| 1235 |
"step": 4775
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
-
"epoch":
|
| 1239 |
-
"
|
| 1240 |
-
"
|
|
|
|
| 1241 |
"step": 4800
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
-
"epoch":
|
| 1245 |
-
"
|
| 1246 |
-
"
|
|
|
|
| 1247 |
"step": 4825
|
| 1248 |
},
|
| 1249 |
{
|
| 1250 |
-
"epoch":
|
| 1251 |
-
"
|
| 1252 |
-
"
|
|
|
|
| 1253 |
"step": 4850
|
| 1254 |
},
|
| 1255 |
{
|
| 1256 |
-
"epoch":
|
| 1257 |
-
"
|
| 1258 |
-
"
|
|
|
|
| 1259 |
"step": 4875
|
| 1260 |
},
|
| 1261 |
{
|
| 1262 |
-
"epoch":
|
| 1263 |
-
"
|
| 1264 |
-
"
|
|
|
|
| 1265 |
"step": 4900
|
| 1266 |
},
|
| 1267 |
{
|
| 1268 |
-
"epoch":
|
| 1269 |
-
"
|
| 1270 |
-
"
|
|
|
|
| 1271 |
"step": 4925
|
| 1272 |
},
|
| 1273 |
{
|
| 1274 |
-
"epoch":
|
| 1275 |
-
"
|
| 1276 |
-
"
|
|
|
|
| 1277 |
"step": 4950
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
-
"epoch":
|
| 1281 |
-
"
|
| 1282 |
-
"
|
|
|
|
| 1283 |
"step": 4975
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
-
"epoch":
|
| 1287 |
-
"
|
| 1288 |
-
"
|
|
|
|
| 1289 |
"step": 5000
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
-
"epoch":
|
| 1293 |
-
"eval_loss": 0.
|
| 1294 |
-
"eval_runtime":
|
| 1295 |
-
"eval_samples_per_second":
|
| 1296 |
-
"eval_steps_per_second": 0.
|
| 1297 |
-
"eval_wer":
|
| 1298 |
"step": 5000
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
-
"epoch":
|
| 1302 |
-
"
|
| 1303 |
-
"
|
|
|
|
| 1304 |
"step": 5025
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
-
"epoch":
|
| 1308 |
-
"
|
| 1309 |
-
"
|
|
|
|
| 1310 |
"step": 5050
|
| 1311 |
},
|
| 1312 |
{
|
| 1313 |
-
"epoch":
|
| 1314 |
-
"
|
| 1315 |
-
"
|
|
|
|
| 1316 |
"step": 5075
|
| 1317 |
},
|
| 1318 |
{
|
| 1319 |
-
"epoch":
|
| 1320 |
-
"
|
| 1321 |
-
"
|
|
|
|
| 1322 |
"step": 5100
|
| 1323 |
},
|
| 1324 |
{
|
| 1325 |
-
"epoch":
|
| 1326 |
-
"
|
| 1327 |
-
"
|
|
|
|
| 1328 |
"step": 5125
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
-
"epoch":
|
| 1332 |
-
"
|
| 1333 |
-
"
|
|
|
|
| 1334 |
"step": 5150
|
| 1335 |
},
|
| 1336 |
{
|
| 1337 |
-
"epoch":
|
| 1338 |
-
"
|
| 1339 |
-
"
|
|
|
|
| 1340 |
"step": 5175
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
-
"epoch":
|
| 1344 |
-
"
|
| 1345 |
-
"
|
|
|
|
| 1346 |
"step": 5200
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
-
"epoch":
|
| 1350 |
-
"
|
| 1351 |
-
"
|
|
|
|
| 1352 |
"step": 5225
|
| 1353 |
},
|
| 1354 |
{
|
| 1355 |
-
"epoch":
|
| 1356 |
-
"
|
| 1357 |
-
"
|
|
|
|
| 1358 |
"step": 5250
|
| 1359 |
},
|
| 1360 |
{
|
| 1361 |
-
"epoch":
|
| 1362 |
-
"
|
| 1363 |
-
"
|
|
|
|
| 1364 |
"step": 5275
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
-
"epoch":
|
| 1368 |
-
"
|
| 1369 |
-
"
|
|
|
|
| 1370 |
"step": 5300
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
-
"epoch":
|
| 1374 |
-
"
|
| 1375 |
-
"
|
|
|
|
| 1376 |
"step": 5325
|
| 1377 |
},
|
| 1378 |
{
|
| 1379 |
-
"epoch":
|
| 1380 |
-
"
|
| 1381 |
-
"
|
|
|
|
| 1382 |
"step": 5350
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
-
"epoch":
|
| 1386 |
-
"
|
| 1387 |
-
"
|
|
|
|
| 1388 |
"step": 5375
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
-
"epoch":
|
| 1392 |
-
"
|
| 1393 |
-
"
|
|
|
|
| 1394 |
"step": 5400
|
| 1395 |
},
|
| 1396 |
{
|
| 1397 |
-
"epoch":
|
| 1398 |
-
"
|
| 1399 |
-
"
|
|
|
|
| 1400 |
"step": 5425
|
| 1401 |
},
|
| 1402 |
{
|
| 1403 |
-
"epoch":
|
| 1404 |
-
"
|
| 1405 |
-
"
|
|
|
|
| 1406 |
"step": 5450
|
| 1407 |
},
|
| 1408 |
{
|
| 1409 |
-
"epoch":
|
| 1410 |
-
"
|
| 1411 |
-
"
|
|
|
|
| 1412 |
"step": 5475
|
| 1413 |
},
|
| 1414 |
{
|
| 1415 |
-
"epoch":
|
| 1416 |
-
"
|
| 1417 |
-
"
|
|
|
|
| 1418 |
"step": 5500
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
-
"epoch":
|
| 1422 |
-
"eval_loss": 0.
|
| 1423 |
-
"eval_runtime":
|
| 1424 |
-
"eval_samples_per_second":
|
| 1425 |
-
"eval_steps_per_second": 0.
|
| 1426 |
-
"eval_wer":
|
| 1427 |
"step": 5500
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
-
"epoch":
|
| 1431 |
-
"
|
| 1432 |
-
"
|
|
|
|
| 1433 |
"step": 5525
|
| 1434 |
},
|
| 1435 |
{
|
| 1436 |
-
"epoch":
|
| 1437 |
-
"
|
| 1438 |
-
"
|
|
|
|
| 1439 |
"step": 5550
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
-
"epoch":
|
| 1443 |
-
"
|
| 1444 |
-
"
|
|
|
|
| 1445 |
"step": 5575
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
-
"epoch":
|
| 1449 |
-
"
|
| 1450 |
-
"
|
|
|
|
| 1451 |
"step": 5600
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
-
"epoch":
|
| 1455 |
-
"
|
| 1456 |
-
"
|
|
|
|
| 1457 |
"step": 5625
|
| 1458 |
},
|
| 1459 |
{
|
| 1460 |
-
"epoch":
|
| 1461 |
-
"
|
| 1462 |
-
"
|
|
|
|
| 1463 |
"step": 5650
|
| 1464 |
},
|
| 1465 |
{
|
| 1466 |
-
"epoch":
|
| 1467 |
-
"
|
| 1468 |
-
"
|
|
|
|
| 1469 |
"step": 5675
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
-
"epoch":
|
| 1473 |
-
"
|
| 1474 |
-
"
|
|
|
|
| 1475 |
"step": 5700
|
| 1476 |
},
|
| 1477 |
{
|
| 1478 |
-
"epoch":
|
| 1479 |
-
"
|
| 1480 |
-
"
|
|
|
|
| 1481 |
"step": 5725
|
| 1482 |
},
|
| 1483 |
{
|
| 1484 |
-
"epoch":
|
| 1485 |
-
"
|
| 1486 |
-
"
|
|
|
|
| 1487 |
"step": 5750
|
| 1488 |
},
|
| 1489 |
{
|
| 1490 |
-
"epoch":
|
| 1491 |
-
"
|
| 1492 |
-
"
|
|
|
|
| 1493 |
"step": 5775
|
| 1494 |
},
|
| 1495 |
{
|
| 1496 |
-
"epoch":
|
| 1497 |
-
"
|
| 1498 |
-
"
|
|
|
|
| 1499 |
"step": 5800
|
| 1500 |
},
|
| 1501 |
{
|
| 1502 |
-
"epoch":
|
| 1503 |
-
"
|
| 1504 |
-
"
|
|
|
|
| 1505 |
"step": 5825
|
| 1506 |
},
|
| 1507 |
{
|
| 1508 |
-
"epoch":
|
| 1509 |
-
"
|
| 1510 |
-
"
|
|
|
|
| 1511 |
"step": 5850
|
| 1512 |
},
|
| 1513 |
{
|
| 1514 |
-
"epoch":
|
| 1515 |
-
"
|
| 1516 |
-
"
|
|
|
|
| 1517 |
"step": 5875
|
| 1518 |
},
|
| 1519 |
{
|
| 1520 |
-
"epoch":
|
| 1521 |
-
"
|
| 1522 |
-
"
|
|
|
|
| 1523 |
"step": 5900
|
| 1524 |
},
|
| 1525 |
{
|
| 1526 |
-
"epoch":
|
| 1527 |
-
"
|
| 1528 |
-
"
|
|
|
|
| 1529 |
"step": 5925
|
| 1530 |
},
|
| 1531 |
{
|
| 1532 |
-
"epoch":
|
| 1533 |
-
"
|
| 1534 |
-
"
|
|
|
|
| 1535 |
"step": 5950
|
| 1536 |
},
|
| 1537 |
{
|
| 1538 |
-
"epoch":
|
| 1539 |
-
"
|
| 1540 |
-
"
|
|
|
|
| 1541 |
"step": 5975
|
| 1542 |
},
|
| 1543 |
{
|
| 1544 |
-
"epoch":
|
| 1545 |
-
"
|
| 1546 |
-
"
|
|
|
|
| 1547 |
"step": 6000
|
| 1548 |
},
|
| 1549 |
{
|
| 1550 |
-
"epoch":
|
| 1551 |
-
"eval_loss": 0.
|
| 1552 |
-
"eval_runtime":
|
| 1553 |
-
"eval_samples_per_second":
|
| 1554 |
-
"eval_steps_per_second": 0.
|
| 1555 |
-
"eval_wer":
|
| 1556 |
"step": 6000
|
| 1557 |
},
|
| 1558 |
{
|
| 1559 |
-
"epoch":
|
| 1560 |
-
"
|
| 1561 |
-
"
|
|
|
|
| 1562 |
"step": 6025
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
-
"epoch":
|
| 1566 |
-
"
|
| 1567 |
-
"
|
|
|
|
| 1568 |
"step": 6050
|
| 1569 |
},
|
| 1570 |
{
|
| 1571 |
-
"epoch":
|
| 1572 |
-
"
|
| 1573 |
-
"
|
|
|
|
| 1574 |
"step": 6075
|
| 1575 |
},
|
| 1576 |
{
|
| 1577 |
-
"epoch":
|
| 1578 |
-
"
|
| 1579 |
-
"
|
|
|
|
| 1580 |
"step": 6100
|
| 1581 |
},
|
| 1582 |
{
|
| 1583 |
-
"epoch":
|
| 1584 |
-
"
|
| 1585 |
-
"
|
|
|
|
| 1586 |
"step": 6125
|
| 1587 |
},
|
| 1588 |
{
|
| 1589 |
-
"epoch":
|
| 1590 |
-
"
|
| 1591 |
-
"
|
|
|
|
| 1592 |
"step": 6150
|
| 1593 |
},
|
| 1594 |
{
|
| 1595 |
-
"epoch":
|
| 1596 |
-
"
|
| 1597 |
-
"
|
|
|
|
| 1598 |
"step": 6175
|
| 1599 |
},
|
| 1600 |
{
|
| 1601 |
-
"epoch":
|
| 1602 |
-
"
|
| 1603 |
-
"
|
|
|
|
| 1604 |
"step": 6200
|
| 1605 |
},
|
| 1606 |
{
|
| 1607 |
-
"epoch":
|
| 1608 |
-
"
|
| 1609 |
-
"
|
|
|
|
| 1610 |
"step": 6225
|
| 1611 |
},
|
| 1612 |
{
|
| 1613 |
-
"epoch":
|
| 1614 |
-
"
|
| 1615 |
-
"
|
|
|
|
| 1616 |
"step": 6250
|
| 1617 |
},
|
| 1618 |
{
|
| 1619 |
-
"epoch":
|
| 1620 |
-
"
|
| 1621 |
-
"
|
|
|
|
| 1622 |
"step": 6275
|
| 1623 |
},
|
| 1624 |
{
|
| 1625 |
-
"epoch":
|
| 1626 |
-
"
|
| 1627 |
-
"
|
|
|
|
| 1628 |
"step": 6300
|
| 1629 |
},
|
| 1630 |
{
|
| 1631 |
-
"epoch":
|
| 1632 |
-
"
|
| 1633 |
-
"
|
|
|
|
| 1634 |
"step": 6325
|
| 1635 |
},
|
| 1636 |
{
|
| 1637 |
-
"epoch":
|
| 1638 |
-
"
|
| 1639 |
-
"
|
|
|
|
| 1640 |
"step": 6350
|
| 1641 |
},
|
| 1642 |
{
|
| 1643 |
-
"epoch":
|
| 1644 |
-
"
|
| 1645 |
-
"
|
|
|
|
| 1646 |
"step": 6375
|
| 1647 |
},
|
| 1648 |
{
|
| 1649 |
-
"epoch":
|
| 1650 |
-
"
|
| 1651 |
-
"
|
|
|
|
| 1652 |
"step": 6400
|
| 1653 |
},
|
| 1654 |
{
|
| 1655 |
-
"epoch":
|
| 1656 |
-
"
|
| 1657 |
-
"
|
|
|
|
| 1658 |
"step": 6425
|
| 1659 |
},
|
| 1660 |
{
|
| 1661 |
-
"epoch":
|
| 1662 |
-
"
|
| 1663 |
-
"
|
|
|
|
| 1664 |
"step": 6450
|
| 1665 |
},
|
| 1666 |
{
|
| 1667 |
-
"epoch":
|
| 1668 |
-
"
|
| 1669 |
-
"
|
|
|
|
| 1670 |
"step": 6475
|
| 1671 |
},
|
| 1672 |
{
|
| 1673 |
-
"epoch":
|
| 1674 |
-
"
|
| 1675 |
-
"
|
|
|
|
| 1676 |
"step": 6500
|
| 1677 |
},
|
| 1678 |
{
|
| 1679 |
-
"epoch":
|
| 1680 |
-
"eval_loss": 0.
|
| 1681 |
-
"eval_runtime":
|
| 1682 |
-
"eval_samples_per_second":
|
| 1683 |
-
"eval_steps_per_second": 0.
|
| 1684 |
-
"eval_wer": 9.
|
| 1685 |
"step": 6500
|
| 1686 |
},
|
| 1687 |
{
|
| 1688 |
-
"epoch":
|
| 1689 |
-
"
|
| 1690 |
-
"
|
|
|
|
| 1691 |
"step": 6525
|
| 1692 |
},
|
| 1693 |
{
|
| 1694 |
-
"epoch":
|
| 1695 |
-
"
|
| 1696 |
-
"
|
|
|
|
| 1697 |
"step": 6550
|
| 1698 |
},
|
| 1699 |
{
|
| 1700 |
-
"epoch":
|
| 1701 |
-
"
|
| 1702 |
-
"
|
|
|
|
| 1703 |
"step": 6575
|
| 1704 |
},
|
| 1705 |
{
|
| 1706 |
-
"epoch":
|
| 1707 |
-
"
|
| 1708 |
-
"
|
|
|
|
| 1709 |
"step": 6600
|
| 1710 |
},
|
| 1711 |
{
|
| 1712 |
-
"epoch":
|
| 1713 |
-
"
|
| 1714 |
-
"
|
|
|
|
| 1715 |
"step": 6625
|
| 1716 |
},
|
| 1717 |
{
|
| 1718 |
-
"epoch":
|
| 1719 |
-
"
|
| 1720 |
-
"
|
|
|
|
| 1721 |
"step": 6650
|
| 1722 |
},
|
| 1723 |
{
|
| 1724 |
-
"epoch":
|
| 1725 |
-
"
|
| 1726 |
-
"
|
|
|
|
| 1727 |
"step": 6675
|
| 1728 |
},
|
| 1729 |
{
|
| 1730 |
-
"epoch":
|
| 1731 |
-
"
|
| 1732 |
-
"
|
|
|
|
| 1733 |
"step": 6700
|
| 1734 |
},
|
| 1735 |
{
|
| 1736 |
-
"epoch":
|
| 1737 |
-
"
|
| 1738 |
-
"
|
|
|
|
| 1739 |
"step": 6725
|
| 1740 |
},
|
| 1741 |
{
|
| 1742 |
-
"epoch":
|
| 1743 |
-
"
|
| 1744 |
-
"
|
|
|
|
| 1745 |
"step": 6750
|
| 1746 |
},
|
| 1747 |
{
|
| 1748 |
-
"epoch":
|
| 1749 |
-
"
|
| 1750 |
-
"
|
|
|
|
| 1751 |
"step": 6775
|
| 1752 |
},
|
| 1753 |
{
|
| 1754 |
-
"epoch":
|
| 1755 |
-
"
|
| 1756 |
-
"
|
|
|
|
| 1757 |
"step": 6800
|
| 1758 |
},
|
| 1759 |
{
|
| 1760 |
-
"epoch":
|
| 1761 |
-
"
|
| 1762 |
-
"
|
|
|
|
| 1763 |
"step": 6825
|
| 1764 |
},
|
| 1765 |
{
|
| 1766 |
-
"epoch":
|
| 1767 |
-
"
|
| 1768 |
-
"
|
|
|
|
| 1769 |
"step": 6850
|
| 1770 |
},
|
| 1771 |
{
|
| 1772 |
-
"epoch":
|
| 1773 |
-
"
|
| 1774 |
-
"
|
|
|
|
| 1775 |
"step": 6875
|
| 1776 |
},
|
| 1777 |
{
|
| 1778 |
-
"epoch":
|
| 1779 |
-
"
|
| 1780 |
-
"
|
|
|
|
| 1781 |
"step": 6900
|
| 1782 |
},
|
| 1783 |
{
|
| 1784 |
-
"epoch":
|
| 1785 |
-
"
|
| 1786 |
-
"
|
|
|
|
| 1787 |
"step": 6925
|
| 1788 |
},
|
| 1789 |
{
|
| 1790 |
-
"epoch":
|
| 1791 |
-
"
|
| 1792 |
-
"
|
|
|
|
| 1793 |
"step": 6950
|
| 1794 |
},
|
| 1795 |
{
|
| 1796 |
-
"epoch":
|
| 1797 |
-
"
|
| 1798 |
-
"
|
|
|
|
| 1799 |
"step": 6975
|
| 1800 |
},
|
| 1801 |
{
|
| 1802 |
-
"epoch":
|
| 1803 |
-
"
|
| 1804 |
-
"
|
|
|
|
| 1805 |
"step": 7000
|
| 1806 |
},
|
| 1807 |
{
|
| 1808 |
-
"epoch":
|
| 1809 |
-
"eval_loss": 0.
|
| 1810 |
-
"eval_runtime":
|
| 1811 |
-
"eval_samples_per_second":
|
| 1812 |
-
"eval_steps_per_second": 0.
|
| 1813 |
-
"eval_wer":
|
| 1814 |
"step": 7000
|
| 1815 |
},
|
| 1816 |
{
|
| 1817 |
-
"epoch":
|
| 1818 |
-
"
|
| 1819 |
-
"
|
|
|
|
| 1820 |
"step": 7025
|
| 1821 |
},
|
| 1822 |
{
|
| 1823 |
-
"epoch":
|
| 1824 |
-
"
|
| 1825 |
-
"
|
|
|
|
| 1826 |
"step": 7050
|
| 1827 |
},
|
| 1828 |
{
|
| 1829 |
-
"epoch":
|
| 1830 |
-
"
|
| 1831 |
-
"
|
|
|
|
| 1832 |
"step": 7075
|
| 1833 |
},
|
| 1834 |
{
|
| 1835 |
-
"epoch":
|
| 1836 |
-
"
|
| 1837 |
-
"
|
|
|
|
| 1838 |
"step": 7100
|
| 1839 |
},
|
| 1840 |
{
|
| 1841 |
-
"epoch":
|
| 1842 |
-
"
|
| 1843 |
-
"
|
|
|
|
| 1844 |
"step": 7125
|
| 1845 |
},
|
| 1846 |
{
|
| 1847 |
-
"epoch":
|
| 1848 |
-
"
|
| 1849 |
-
"
|
|
|
|
| 1850 |
"step": 7150
|
| 1851 |
},
|
| 1852 |
{
|
| 1853 |
-
"epoch":
|
| 1854 |
-
"
|
| 1855 |
-
"
|
|
|
|
| 1856 |
"step": 7175
|
| 1857 |
},
|
| 1858 |
{
|
| 1859 |
-
"epoch":
|
| 1860 |
-
"
|
| 1861 |
-
"
|
|
|
|
| 1862 |
"step": 7200
|
| 1863 |
},
|
| 1864 |
{
|
| 1865 |
-
"epoch":
|
| 1866 |
-
"
|
| 1867 |
-
"
|
|
|
|
| 1868 |
"step": 7225
|
| 1869 |
},
|
| 1870 |
{
|
| 1871 |
-
"epoch":
|
| 1872 |
-
"
|
| 1873 |
-
"
|
|
|
|
| 1874 |
"step": 7250
|
| 1875 |
},
|
| 1876 |
{
|
| 1877 |
-
"epoch":
|
| 1878 |
-
"
|
| 1879 |
-
"
|
|
|
|
| 1880 |
"step": 7275
|
| 1881 |
},
|
| 1882 |
{
|
| 1883 |
-
"epoch":
|
| 1884 |
-
"
|
| 1885 |
-
"
|
|
|
|
| 1886 |
"step": 7300
|
| 1887 |
},
|
| 1888 |
{
|
| 1889 |
-
"epoch":
|
| 1890 |
-
"
|
| 1891 |
-
"
|
|
|
|
| 1892 |
"step": 7325
|
| 1893 |
},
|
| 1894 |
{
|
| 1895 |
-
"epoch":
|
| 1896 |
-
"
|
| 1897 |
-
"
|
|
|
|
| 1898 |
"step": 7350
|
| 1899 |
},
|
| 1900 |
{
|
| 1901 |
-
"epoch":
|
| 1902 |
-
"
|
| 1903 |
-
"
|
|
|
|
| 1904 |
"step": 7375
|
| 1905 |
},
|
| 1906 |
{
|
| 1907 |
-
"epoch":
|
| 1908 |
-
"
|
| 1909 |
-
"
|
|
|
|
| 1910 |
"step": 7400
|
| 1911 |
},
|
| 1912 |
{
|
| 1913 |
-
"epoch":
|
| 1914 |
-
"
|
| 1915 |
-
"
|
|
|
|
| 1916 |
"step": 7425
|
| 1917 |
},
|
| 1918 |
{
|
| 1919 |
-
"epoch":
|
| 1920 |
-
"
|
| 1921 |
-
"
|
|
|
|
| 1922 |
"step": 7450
|
| 1923 |
},
|
| 1924 |
{
|
| 1925 |
-
"epoch":
|
| 1926 |
-
"
|
| 1927 |
-
"
|
|
|
|
| 1928 |
"step": 7475
|
| 1929 |
},
|
| 1930 |
{
|
| 1931 |
-
"epoch":
|
| 1932 |
-
"
|
| 1933 |
-
"
|
|
|
|
| 1934 |
"step": 7500
|
| 1935 |
},
|
| 1936 |
{
|
| 1937 |
-
"epoch":
|
| 1938 |
-
"eval_loss": 0.
|
| 1939 |
-
"eval_runtime":
|
| 1940 |
-
"eval_samples_per_second":
|
| 1941 |
-
"eval_steps_per_second": 0.
|
| 1942 |
-
"eval_wer":
|
| 1943 |
"step": 7500
|
| 1944 |
},
|
| 1945 |
{
|
| 1946 |
-
"epoch":
|
| 1947 |
-
"
|
| 1948 |
-
"
|
|
|
|
| 1949 |
"step": 7525
|
| 1950 |
},
|
| 1951 |
{
|
| 1952 |
-
"epoch":
|
| 1953 |
-
"
|
| 1954 |
-
"
|
|
|
|
| 1955 |
"step": 7550
|
| 1956 |
},
|
| 1957 |
{
|
| 1958 |
-
"epoch":
|
| 1959 |
-
"
|
| 1960 |
-
"
|
|
|
|
| 1961 |
"step": 7575
|
| 1962 |
},
|
| 1963 |
{
|
| 1964 |
-
"epoch":
|
| 1965 |
-
"
|
| 1966 |
-
"
|
|
|
|
| 1967 |
"step": 7600
|
| 1968 |
},
|
| 1969 |
{
|
| 1970 |
-
"epoch":
|
| 1971 |
-
"
|
| 1972 |
-
"
|
|
|
|
| 1973 |
"step": 7625
|
| 1974 |
},
|
| 1975 |
{
|
| 1976 |
-
"epoch":
|
| 1977 |
-
"
|
| 1978 |
-
"
|
|
|
|
| 1979 |
"step": 7650
|
| 1980 |
},
|
| 1981 |
{
|
| 1982 |
-
"epoch":
|
| 1983 |
-
"
|
| 1984 |
-
"
|
|
|
|
| 1985 |
"step": 7675
|
| 1986 |
},
|
| 1987 |
{
|
| 1988 |
-
"epoch":
|
| 1989 |
-
"
|
| 1990 |
-
"
|
|
|
|
| 1991 |
"step": 7700
|
| 1992 |
},
|
| 1993 |
{
|
| 1994 |
-
"epoch":
|
| 1995 |
-
"
|
| 1996 |
-
"
|
|
|
|
| 1997 |
"step": 7725
|
| 1998 |
},
|
| 1999 |
{
|
| 2000 |
-
"epoch":
|
| 2001 |
-
"
|
| 2002 |
-
"
|
|
|
|
| 2003 |
"step": 7750
|
| 2004 |
},
|
| 2005 |
{
|
| 2006 |
-
"epoch":
|
| 2007 |
-
"
|
| 2008 |
-
"
|
|
|
|
| 2009 |
"step": 7775
|
| 2010 |
},
|
| 2011 |
{
|
| 2012 |
-
"epoch":
|
| 2013 |
-
"
|
| 2014 |
-
"
|
|
|
|
| 2015 |
"step": 7800
|
| 2016 |
},
|
| 2017 |
{
|
| 2018 |
-
"epoch":
|
| 2019 |
-
"
|
| 2020 |
-
"
|
|
|
|
| 2021 |
"step": 7825
|
| 2022 |
},
|
| 2023 |
{
|
| 2024 |
-
"epoch":
|
| 2025 |
-
"
|
| 2026 |
-
"
|
|
|
|
| 2027 |
"step": 7850
|
| 2028 |
},
|
| 2029 |
{
|
| 2030 |
-
"epoch":
|
| 2031 |
-
"
|
| 2032 |
-
"
|
|
|
|
| 2033 |
"step": 7875
|
| 2034 |
},
|
| 2035 |
{
|
| 2036 |
-
"epoch":
|
| 2037 |
-
"
|
| 2038 |
-
"
|
|
|
|
| 2039 |
"step": 7900
|
| 2040 |
},
|
| 2041 |
{
|
| 2042 |
-
"epoch":
|
| 2043 |
-
"
|
| 2044 |
-
"
|
|
|
|
| 2045 |
"step": 7925
|
| 2046 |
},
|
| 2047 |
{
|
| 2048 |
-
"epoch":
|
| 2049 |
-
"
|
| 2050 |
-
"
|
|
|
|
| 2051 |
"step": 7950
|
| 2052 |
},
|
| 2053 |
{
|
| 2054 |
-
"epoch":
|
| 2055 |
-
"
|
| 2056 |
-
"
|
|
|
|
| 2057 |
"step": 7975
|
| 2058 |
},
|
| 2059 |
{
|
| 2060 |
-
"epoch":
|
| 2061 |
-
"
|
| 2062 |
-
"
|
|
|
|
| 2063 |
"step": 8000
|
| 2064 |
},
|
| 2065 |
{
|
| 2066 |
-
"epoch":
|
| 2067 |
-
"eval_loss": 0.
|
| 2068 |
-
"eval_runtime":
|
| 2069 |
-
"eval_samples_per_second":
|
| 2070 |
-
"eval_steps_per_second": 0.
|
| 2071 |
-
"eval_wer":
|
| 2072 |
"step": 8000
|
| 2073 |
},
|
| 2074 |
{
|
| 2075 |
-
"epoch":
|
| 2076 |
"step": 8000,
|
| 2077 |
-
"total_flos":
|
| 2078 |
-
"train_loss": 0.
|
| 2079 |
-
"train_runtime":
|
| 2080 |
-
"train_samples_per_second":
|
| 2081 |
-
"train_steps_per_second": 0.
|
| 2082 |
}
|
| 2083 |
],
|
|
|
|
| 2084 |
"max_steps": 8000,
|
|
|
|
| 2085 |
"num_train_epochs": 9223372036854775807,
|
| 2086 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2087 |
"trial_name": null,
|
| 2088 |
"trial_params": null
|
| 2089 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 8.8020814247499,
|
| 3 |
"best_model_checkpoint": "./checkpoint-8000",
|
| 4 |
+
"epoch": 4.148,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
"global_step": 8000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.003125,
|
| 13 |
+
"grad_norm": 15.080086708068848,
|
| 14 |
+
"learning_rate": 2.875e-07,
|
| 15 |
+
"loss": 1.3478,
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.00625,
|
| 20 |
+
"grad_norm": 11.962413787841797,
|
| 21 |
+
"learning_rate": 6.000000000000001e-07,
|
| 22 |
+
"loss": 1.14,
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.009375,
|
| 27 |
+
"grad_norm": 10.828227996826172,
|
| 28 |
+
"learning_rate": 9.124999999999999e-07,
|
| 29 |
+
"loss": 0.8401,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.0125,
|
| 34 |
+
"grad_norm": 9.401681900024414,
|
| 35 |
+
"learning_rate": 1.2250000000000001e-06,
|
| 36 |
+
"loss": 0.7112,
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.015625,
|
| 41 |
+
"grad_norm": 9.252142906188965,
|
| 42 |
+
"learning_rate": 1.5375e-06,
|
| 43 |
+
"loss": 0.6528,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.01875,
|
| 48 |
+
"grad_norm": 10.718783378601074,
|
| 49 |
+
"learning_rate": 1.85e-06,
|
| 50 |
+
"loss": 0.6492,
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.021875,
|
| 55 |
+
"grad_norm": 7.8105082511901855,
|
| 56 |
+
"learning_rate": 2.1625e-06,
|
| 57 |
+
"loss": 0.5886,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.025,
|
| 62 |
+
"grad_norm": 8.969450950622559,
|
| 63 |
+
"learning_rate": 2.4750000000000004e-06,
|
| 64 |
+
"loss": 0.5284,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.028125,
|
| 69 |
+
"grad_norm": 8.110017776489258,
|
| 70 |
+
"learning_rate": 2.7875e-06,
|
| 71 |
+
"loss": 0.5097,
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.03125,
|
| 76 |
+
"grad_norm": 8.07854175567627,
|
| 77 |
+
"learning_rate": 3.1e-06,
|
| 78 |
+
"loss": 0.48,
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.034375,
|
| 83 |
+
"grad_norm": 9.469429016113281,
|
| 84 |
+
"learning_rate": 3.4125000000000004e-06,
|
| 85 |
+
"loss": 0.4453,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.0375,
|
| 90 |
+
"grad_norm": 6.849759578704834,
|
| 91 |
+
"learning_rate": 3.725e-06,
|
| 92 |
+
"loss": 0.4249,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.040625,
|
| 97 |
+
"grad_norm": 8.43980598449707,
|
| 98 |
+
"learning_rate": 4.037500000000001e-06,
|
| 99 |
+
"loss": 0.4093,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.04375,
|
| 104 |
+
"grad_norm": 7.9978132247924805,
|
| 105 |
+
"learning_rate": 4.35e-06,
|
| 106 |
+
"loss": 0.4179,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.046875,
|
| 111 |
+
"grad_norm": 7.179676532745361,
|
| 112 |
+
"learning_rate": 4.6625e-06,
|
| 113 |
+
"loss": 0.366,
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.05,
|
| 118 |
+
"grad_norm": 8.746236801147461,
|
| 119 |
+
"learning_rate": 4.975000000000001e-06,
|
| 120 |
+
"loss": 0.3484,
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.053125,
|
| 125 |
+
"grad_norm": 6.08207893371582,
|
| 126 |
+
"learning_rate": 5.2875e-06,
|
| 127 |
+
"loss": 0.3177,
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.05625,
|
| 132 |
+
"grad_norm": 7.5096235275268555,
|
| 133 |
+
"learning_rate": 5.600000000000001e-06,
|
| 134 |
+
"loss": 0.3407,
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.059375,
|
| 139 |
+
"grad_norm": 6.521783828735352,
|
| 140 |
+
"learning_rate": 5.9125e-06,
|
| 141 |
+
"loss": 0.3323,
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 0.0625,
|
| 146 |
+
"grad_norm": 6.018587589263916,
|
| 147 |
+
"learning_rate": 6.2250000000000005e-06,
|
| 148 |
+
"loss": 0.3171,
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.0625,
|
| 153 |
+
"eval_loss": 0.33685919642448425,
|
| 154 |
+
"eval_runtime": 2424.3615,
|
| 155 |
+
"eval_samples_per_second": 5.622,
|
| 156 |
+
"eval_steps_per_second": 0.703,
|
| 157 |
+
"eval_wer": 25.53043350800689,
|
| 158 |
"step": 500
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 0.065625,
|
| 162 |
+
"grad_norm": 6.343411922454834,
|
| 163 |
+
"learning_rate": 6.230833333333334e-06,
|
| 164 |
+
"loss": 0.323,
|
| 165 |
"step": 525
|
| 166 |
},
|
| 167 |
{
|
| 168 |
+
"epoch": 0.06875,
|
| 169 |
+
"grad_norm": 6.164433479309082,
|
| 170 |
+
"learning_rate": 6.210000000000001e-06,
|
| 171 |
+
"loss": 0.2979,
|
| 172 |
"step": 550
|
| 173 |
},
|
| 174 |
{
|
| 175 |
+
"epoch": 0.071875,
|
| 176 |
+
"grad_norm": 7.514057159423828,
|
| 177 |
+
"learning_rate": 6.1891666666666666e-06,
|
| 178 |
+
"loss": 0.2956,
|
| 179 |
"step": 575
|
| 180 |
},
|
| 181 |
{
|
| 182 |
+
"epoch": 0.075,
|
| 183 |
+
"grad_norm": 5.7609992027282715,
|
| 184 |
+
"learning_rate": 6.168333333333333e-06,
|
| 185 |
+
"loss": 0.2966,
|
| 186 |
"step": 600
|
| 187 |
},
|
| 188 |
{
|
| 189 |
+
"epoch": 0.078125,
|
| 190 |
+
"grad_norm": 6.035708427429199,
|
| 191 |
+
"learning_rate": 6.147500000000001e-06,
|
| 192 |
+
"loss": 0.2914,
|
| 193 |
"step": 625
|
| 194 |
},
|
| 195 |
{
|
| 196 |
+
"epoch": 0.08125,
|
| 197 |
+
"grad_norm": 6.485240459442139,
|
| 198 |
+
"learning_rate": 6.126666666666667e-06,
|
| 199 |
+
"loss": 0.2929,
|
| 200 |
"step": 650
|
| 201 |
},
|
| 202 |
{
|
| 203 |
+
"epoch": 0.084375,
|
| 204 |
+
"grad_norm": 5.631970405578613,
|
| 205 |
+
"learning_rate": 6.1058333333333335e-06,
|
| 206 |
+
"loss": 0.2676,
|
| 207 |
"step": 675
|
| 208 |
},
|
| 209 |
{
|
| 210 |
+
"epoch": 0.0875,
|
| 211 |
+
"grad_norm": 6.642117977142334,
|
| 212 |
+
"learning_rate": 6.085e-06,
|
| 213 |
+
"loss": 0.2756,
|
| 214 |
"step": 700
|
| 215 |
},
|
| 216 |
{
|
| 217 |
+
"epoch": 0.090625,
|
| 218 |
+
"grad_norm": 6.473007678985596,
|
| 219 |
+
"learning_rate": 6.064166666666667e-06,
|
| 220 |
+
"loss": 0.2613,
|
| 221 |
"step": 725
|
| 222 |
},
|
| 223 |
{
|
| 224 |
+
"epoch": 0.09375,
|
| 225 |
+
"grad_norm": 6.47869348526001,
|
| 226 |
+
"learning_rate": 6.043333333333334e-06,
|
| 227 |
+
"loss": 0.2667,
|
| 228 |
"step": 750
|
| 229 |
},
|
| 230 |
{
|
| 231 |
+
"epoch": 0.096875,
|
| 232 |
+
"grad_norm": 6.658367156982422,
|
| 233 |
+
"learning_rate": 6.0225e-06,
|
| 234 |
+
"loss": 0.2778,
|
| 235 |
"step": 775
|
| 236 |
},
|
| 237 |
{
|
| 238 |
"epoch": 0.1,
|
| 239 |
+
"grad_norm": 7.887542724609375,
|
| 240 |
+
"learning_rate": 6.001666666666667e-06,
|
| 241 |
+
"loss": 0.3095,
|
| 242 |
"step": 800
|
| 243 |
},
|
| 244 |
{
|
| 245 |
+
"epoch": 0.103125,
|
| 246 |
+
"grad_norm": 6.103913307189941,
|
| 247 |
+
"learning_rate": 5.980833333333334e-06,
|
| 248 |
+
"loss": 0.297,
|
| 249 |
"step": 825
|
| 250 |
},
|
| 251 |
{
|
| 252 |
+
"epoch": 0.10625,
|
| 253 |
+
"grad_norm": 6.087416648864746,
|
| 254 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 255 |
+
"loss": 0.2326,
|
| 256 |
"step": 850
|
| 257 |
},
|
| 258 |
{
|
| 259 |
+
"epoch": 0.109375,
|
| 260 |
+
"grad_norm": 5.486704349517822,
|
| 261 |
+
"learning_rate": 5.939166666666667e-06,
|
| 262 |
+
"loss": 0.2215,
|
| 263 |
"step": 875
|
| 264 |
},
|
| 265 |
{
|
| 266 |
+
"epoch": 0.1125,
|
| 267 |
+
"grad_norm": 5.353816986083984,
|
| 268 |
+
"learning_rate": 5.918333333333333e-06,
|
| 269 |
+
"loss": 0.205,
|
| 270 |
"step": 900
|
| 271 |
},
|
| 272 |
{
|
| 273 |
+
"epoch": 0.115625,
|
| 274 |
+
"grad_norm": 5.429187774658203,
|
| 275 |
+
"learning_rate": 5.8975e-06,
|
| 276 |
+
"loss": 0.2185,
|
| 277 |
"step": 925
|
| 278 |
},
|
| 279 |
{
|
| 280 |
+
"epoch": 0.11875,
|
| 281 |
+
"grad_norm": 5.181328773498535,
|
| 282 |
+
"learning_rate": 5.8766666666666674e-06,
|
| 283 |
+
"loss": 0.2069,
|
| 284 |
"step": 950
|
| 285 |
},
|
| 286 |
{
|
| 287 |
+
"epoch": 0.121875,
|
| 288 |
+
"grad_norm": 4.126291751861572,
|
| 289 |
+
"learning_rate": 5.855833333333333e-06,
|
| 290 |
+
"loss": 0.2062,
|
| 291 |
"step": 975
|
| 292 |
},
|
| 293 |
{
|
| 294 |
+
"epoch": 0.125,
|
| 295 |
+
"grad_norm": 5.896920204162598,
|
| 296 |
+
"learning_rate": 5.835e-06,
|
| 297 |
+
"loss": 0.1852,
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
+
"epoch": 0.125,
|
| 302 |
+
"eval_loss": 0.2408558428287506,
|
| 303 |
+
"eval_runtime": 2415.6545,
|
| 304 |
+
"eval_samples_per_second": 5.642,
|
| 305 |
+
"eval_steps_per_second": 0.705,
|
| 306 |
+
"eval_wer": 17.311004434021037,
|
| 307 |
"step": 1000
|
| 308 |
},
|
| 309 |
{
|
| 310 |
+
"epoch": 0.128125,
|
| 311 |
+
"grad_norm": 5.046872138977051,
|
| 312 |
+
"learning_rate": 5.814166666666667e-06,
|
| 313 |
+
"loss": 0.2142,
|
| 314 |
"step": 1025
|
| 315 |
},
|
| 316 |
{
|
| 317 |
+
"epoch": 0.13125,
|
| 318 |
+
"grad_norm": 6.671748638153076,
|
| 319 |
+
"learning_rate": 5.793333333333334e-06,
|
| 320 |
+
"loss": 0.2208,
|
| 321 |
"step": 1050
|
| 322 |
},
|
| 323 |
{
|
| 324 |
+
"epoch": 0.134375,
|
| 325 |
+
"grad_norm": 6.533191680908203,
|
| 326 |
+
"learning_rate": 5.7725e-06,
|
| 327 |
+
"loss": 0.1944,
|
| 328 |
"step": 1075
|
| 329 |
},
|
| 330 |
{
|
| 331 |
+
"epoch": 0.1375,
|
| 332 |
+
"grad_norm": 4.43493127822876,
|
| 333 |
+
"learning_rate": 5.751666666666667e-06,
|
| 334 |
+
"loss": 0.1816,
|
| 335 |
"step": 1100
|
| 336 |
},
|
| 337 |
{
|
| 338 |
+
"epoch": 0.140625,
|
| 339 |
+
"grad_norm": 5.601134777069092,
|
| 340 |
+
"learning_rate": 5.730833333333334e-06,
|
| 341 |
+
"loss": 0.1884,
|
| 342 |
"step": 1125
|
| 343 |
},
|
| 344 |
{
|
| 345 |
+
"epoch": 0.14375,
|
| 346 |
+
"grad_norm": 4.886395454406738,
|
| 347 |
+
"learning_rate": 5.71e-06,
|
| 348 |
+
"loss": 0.1712,
|
| 349 |
"step": 1150
|
| 350 |
},
|
| 351 |
{
|
| 352 |
+
"epoch": 0.146875,
|
| 353 |
+
"grad_norm": 5.513830184936523,
|
| 354 |
+
"learning_rate": 5.689166666666667e-06,
|
| 355 |
+
"loss": 0.1835,
|
| 356 |
"step": 1175
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 0.15,
|
| 360 |
+
"grad_norm": 6.305610179901123,
|
| 361 |
+
"learning_rate": 5.668333333333334e-06,
|
| 362 |
+
"loss": 0.2221,
|
| 363 |
"step": 1200
|
| 364 |
},
|
| 365 |
{
|
| 366 |
+
"epoch": 0.153125,
|
| 367 |
+
"grad_norm": 6.482845306396484,
|
| 368 |
+
"learning_rate": 5.6475e-06,
|
| 369 |
+
"loss": 0.2138,
|
| 370 |
"step": 1225
|
| 371 |
},
|
| 372 |
{
|
| 373 |
+
"epoch": 0.15625,
|
| 374 |
+
"grad_norm": 6.316329002380371,
|
| 375 |
+
"learning_rate": 5.626666666666667e-06,
|
| 376 |
+
"loss": 0.2338,
|
| 377 |
"step": 1250
|
| 378 |
},
|
| 379 |
{
|
| 380 |
+
"epoch": 0.159375,
|
| 381 |
+
"grad_norm": 6.650278091430664,
|
| 382 |
+
"learning_rate": 5.605833333333334e-06,
|
| 383 |
+
"loss": 0.2231,
|
| 384 |
"step": 1275
|
| 385 |
},
|
| 386 |
{
|
| 387 |
+
"epoch": 0.1625,
|
| 388 |
+
"grad_norm": 5.289668560028076,
|
| 389 |
+
"learning_rate": 5.585e-06,
|
| 390 |
+
"loss": 0.2018,
|
| 391 |
"step": 1300
|
| 392 |
},
|
| 393 |
{
|
| 394 |
+
"epoch": 0.165625,
|
| 395 |
+
"grad_norm": 6.051737308502197,
|
| 396 |
+
"learning_rate": 5.564166666666667e-06,
|
| 397 |
+
"loss": 0.2092,
|
| 398 |
"step": 1325
|
| 399 |
},
|
| 400 |
{
|
| 401 |
+
"epoch": 0.16875,
|
| 402 |
+
"grad_norm": 4.950199604034424,
|
| 403 |
+
"learning_rate": 5.543333333333333e-06,
|
| 404 |
+
"loss": 0.1934,
|
| 405 |
"step": 1350
|
| 406 |
},
|
| 407 |
{
|
| 408 |
+
"epoch": 0.171875,
|
| 409 |
+
"grad_norm": 5.296252727508545,
|
| 410 |
+
"learning_rate": 5.522500000000001e-06,
|
| 411 |
+
"loss": 0.1806,
|
| 412 |
"step": 1375
|
| 413 |
},
|
| 414 |
{
|
| 415 |
+
"epoch": 0.175,
|
| 416 |
+
"grad_norm": 5.999615669250488,
|
| 417 |
+
"learning_rate": 5.501666666666667e-06,
|
| 418 |
+
"loss": 0.1782,
|
| 419 |
"step": 1400
|
| 420 |
},
|
| 421 |
{
|
| 422 |
+
"epoch": 0.178125,
|
| 423 |
+
"grad_norm": 3.7982959747314453,
|
| 424 |
+
"learning_rate": 5.4808333333333335e-06,
|
| 425 |
+
"loss": 0.1712,
|
| 426 |
"step": 1425
|
| 427 |
},
|
| 428 |
{
|
| 429 |
+
"epoch": 0.18125,
|
| 430 |
+
"grad_norm": 6.798803806304932,
|
| 431 |
+
"learning_rate": 5.46e-06,
|
| 432 |
+
"loss": 0.1972,
|
| 433 |
"step": 1450
|
| 434 |
},
|
| 435 |
{
|
| 436 |
+
"epoch": 0.184375,
|
| 437 |
+
"grad_norm": 5.545141220092773,
|
| 438 |
+
"learning_rate": 5.439166666666667e-06,
|
| 439 |
+
"loss": 0.2153,
|
| 440 |
"step": 1475
|
| 441 |
},
|
| 442 |
{
|
| 443 |
+
"epoch": 0.1875,
|
| 444 |
+
"grad_norm": 5.619574546813965,
|
| 445 |
+
"learning_rate": 5.418333333333334e-06,
|
| 446 |
+
"loss": 0.2353,
|
| 447 |
"step": 1500
|
| 448 |
},
|
| 449 |
{
|
| 450 |
+
"epoch": 0.1875,
|
| 451 |
+
"eval_loss": 0.20496371388435364,
|
| 452 |
+
"eval_runtime": 2428.1499,
|
| 453 |
+
"eval_samples_per_second": 5.613,
|
| 454 |
+
"eval_steps_per_second": 0.702,
|
| 455 |
+
"eval_wer": 14.222763750961926,
|
| 456 |
"step": 1500
|
| 457 |
},
|
| 458 |
{
|
| 459 |
+
"epoch": 0.190625,
|
| 460 |
+
"grad_norm": 6.522052764892578,
|
| 461 |
+
"learning_rate": 5.3975e-06,
|
| 462 |
+
"loss": 0.2049,
|
| 463 |
"step": 1525
|
| 464 |
},
|
| 465 |
{
|
| 466 |
+
"epoch": 0.19375,
|
| 467 |
+
"grad_norm": 5.805539608001709,
|
| 468 |
+
"learning_rate": 5.376666666666666e-06,
|
| 469 |
+
"loss": 0.1931,
|
| 470 |
"step": 1550
|
| 471 |
},
|
| 472 |
{
|
| 473 |
+
"epoch": 0.196875,
|
| 474 |
+
"grad_norm": 5.707961082458496,
|
| 475 |
+
"learning_rate": 5.355833333333334e-06,
|
| 476 |
+
"loss": 0.1944,
|
| 477 |
"step": 1575
|
| 478 |
},
|
| 479 |
{
|
| 480 |
"epoch": 0.2,
|
| 481 |
+
"grad_norm": 6.126225471496582,
|
| 482 |
+
"learning_rate": 5.335000000000001e-06,
|
| 483 |
+
"loss": 0.225,
|
| 484 |
"step": 1600
|
| 485 |
},
|
| 486 |
{
|
| 487 |
+
"epoch": 0.203125,
|
| 488 |
+
"grad_norm": 5.682012557983398,
|
| 489 |
+
"learning_rate": 5.3141666666666665e-06,
|
| 490 |
+
"loss": 0.2225,
|
| 491 |
"step": 1625
|
| 492 |
},
|
| 493 |
{
|
| 494 |
+
"epoch": 0.20625,
|
| 495 |
+
"grad_norm": 5.529600620269775,
|
| 496 |
+
"learning_rate": 5.293333333333333e-06,
|
| 497 |
+
"loss": 0.2154,
|
| 498 |
"step": 1650
|
| 499 |
},
|
| 500 |
{
|
| 501 |
+
"epoch": 0.209375,
|
| 502 |
+
"grad_norm": 8.373395919799805,
|
| 503 |
+
"learning_rate": 5.272500000000001e-06,
|
| 504 |
+
"loss": 0.1866,
|
| 505 |
"step": 1675
|
| 506 |
},
|
| 507 |
{
|
| 508 |
+
"epoch": 0.2125,
|
| 509 |
+
"grad_norm": 4.936291694641113,
|
| 510 |
+
"learning_rate": 5.2516666666666675e-06,
|
| 511 |
+
"loss": 0.1615,
|
| 512 |
"step": 1700
|
| 513 |
},
|
| 514 |
{
|
| 515 |
+
"epoch": 1.002625,
|
| 516 |
+
"grad_norm": 4.893776893615723,
|
| 517 |
+
"learning_rate": 5.230833333333333e-06,
|
| 518 |
+
"loss": 0.172,
|
| 519 |
"step": 1725
|
| 520 |
},
|
| 521 |
{
|
| 522 |
+
"epoch": 1.00575,
|
| 523 |
+
"grad_norm": 4.98813533782959,
|
| 524 |
+
"learning_rate": 5.21e-06,
|
| 525 |
+
"loss": 0.1591,
|
| 526 |
"step": 1750
|
| 527 |
},
|
| 528 |
{
|
| 529 |
+
"epoch": 1.008875,
|
| 530 |
+
"grad_norm": 6.166538715362549,
|
| 531 |
+
"learning_rate": 5.189166666666667e-06,
|
| 532 |
+
"loss": 0.1712,
|
| 533 |
"step": 1775
|
| 534 |
},
|
| 535 |
{
|
| 536 |
+
"epoch": 1.012,
|
| 537 |
+
"grad_norm": 4.1829071044921875,
|
| 538 |
+
"learning_rate": 5.1683333333333335e-06,
|
| 539 |
+
"loss": 0.1908,
|
| 540 |
"step": 1800
|
| 541 |
},
|
| 542 |
{
|
| 543 |
+
"epoch": 1.015125,
|
| 544 |
+
"grad_norm": 5.497808456420898,
|
| 545 |
+
"learning_rate": 5.1475e-06,
|
| 546 |
+
"loss": 0.1791,
|
| 547 |
"step": 1825
|
| 548 |
},
|
| 549 |
{
|
| 550 |
+
"epoch": 1.01825,
|
| 551 |
+
"grad_norm": 4.581858158111572,
|
| 552 |
+
"learning_rate": 5.126666666666667e-06,
|
| 553 |
+
"loss": 0.1685,
|
| 554 |
"step": 1850
|
| 555 |
},
|
| 556 |
{
|
| 557 |
+
"epoch": 1.021375,
|
| 558 |
+
"grad_norm": 4.6282267570495605,
|
| 559 |
+
"learning_rate": 5.105833333333334e-06,
|
| 560 |
+
"loss": 0.1691,
|
| 561 |
"step": 1875
|
| 562 |
},
|
| 563 |
{
|
| 564 |
+
"epoch": 1.0245,
|
| 565 |
+
"grad_norm": 5.748936653137207,
|
| 566 |
+
"learning_rate": 5.0850000000000004e-06,
|
| 567 |
+
"loss": 0.1577,
|
| 568 |
"step": 1900
|
| 569 |
},
|
| 570 |
{
|
| 571 |
+
"epoch": 1.027625,
|
| 572 |
+
"grad_norm": 4.339770793914795,
|
| 573 |
+
"learning_rate": 5.064166666666667e-06,
|
| 574 |
+
"loss": 0.168,
|
| 575 |
"step": 1925
|
| 576 |
},
|
| 577 |
{
|
| 578 |
+
"epoch": 1.03075,
|
| 579 |
+
"grad_norm": 3.72379469871521,
|
| 580 |
+
"learning_rate": 5.043333333333333e-06,
|
| 581 |
+
"loss": 0.1616,
|
| 582 |
"step": 1950
|
| 583 |
},
|
| 584 |
{
|
| 585 |
+
"epoch": 1.033875,
|
| 586 |
+
"grad_norm": 5.169766902923584,
|
| 587 |
+
"learning_rate": 5.0225e-06,
|
| 588 |
+
"loss": 0.1663,
|
| 589 |
"step": 1975
|
| 590 |
},
|
| 591 |
{
|
| 592 |
+
"epoch": 1.037,
|
| 593 |
+
"grad_norm": 6.120935440063477,
|
| 594 |
+
"learning_rate": 5.001666666666667e-06,
|
| 595 |
+
"loss": 0.1569,
|
| 596 |
"step": 2000
|
| 597 |
},
|
| 598 |
{
|
| 599 |
+
"epoch": 1.037,
|
| 600 |
+
"eval_loss": 0.181504026055336,
|
| 601 |
+
"eval_runtime": 2441.5281,
|
| 602 |
+
"eval_samples_per_second": 5.583,
|
| 603 |
+
"eval_steps_per_second": 0.698,
|
| 604 |
+
"eval_wer": 12.286085968705338,
|
| 605 |
"step": 2000
|
| 606 |
},
|
| 607 |
{
|
| 608 |
+
"epoch": 1.040125,
|
| 609 |
+
"grad_norm": 3.2327628135681152,
|
| 610 |
+
"learning_rate": 4.980833333333334e-06,
|
| 611 |
+
"loss": 0.1526,
|
| 612 |
"step": 2025
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 1.04325,
|
| 616 |
+
"grad_norm": 4.2089948654174805,
|
| 617 |
+
"learning_rate": 4.96e-06,
|
| 618 |
+
"loss": 0.1427,
|
| 619 |
"step": 2050
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"epoch": 1.046375,
|
| 623 |
+
"grad_norm": 4.837276458740234,
|
| 624 |
+
"learning_rate": 4.939166666666667e-06,
|
| 625 |
+
"loss": 0.1429,
|
| 626 |
"step": 2075
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 1.0495,
|
| 630 |
+
"grad_norm": 4.590527057647705,
|
| 631 |
+
"learning_rate": 4.918333333333334e-06,
|
| 632 |
+
"loss": 0.1306,
|
| 633 |
"step": 2100
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"epoch": 1.052625,
|
| 637 |
+
"grad_norm": 5.821957111358643,
|
| 638 |
+
"learning_rate": 4.8975e-06,
|
| 639 |
+
"loss": 0.1504,
|
| 640 |
"step": 2125
|
| 641 |
},
|
| 642 |
{
|
| 643 |
+
"epoch": 1.05575,
|
| 644 |
+
"grad_norm": 4.136687278747559,
|
| 645 |
+
"learning_rate": 4.876666666666667e-06,
|
| 646 |
+
"loss": 0.1431,
|
| 647 |
"step": 2150
|
| 648 |
},
|
| 649 |
{
|
| 650 |
+
"epoch": 1.058875,
|
| 651 |
+
"grad_norm": 4.310347080230713,
|
| 652 |
+
"learning_rate": 4.8558333333333336e-06,
|
| 653 |
+
"loss": 0.1364,
|
| 654 |
"step": 2175
|
| 655 |
},
|
| 656 |
{
|
| 657 |
+
"epoch": 1.062,
|
| 658 |
+
"grad_norm": 4.1295485496521,
|
| 659 |
+
"learning_rate": 4.835e-06,
|
| 660 |
+
"loss": 0.1272,
|
| 661 |
"step": 2200
|
| 662 |
},
|
| 663 |
{
|
| 664 |
+
"epoch": 1.065125,
|
| 665 |
+
"grad_norm": 4.432501316070557,
|
| 666 |
+
"learning_rate": 4.814166666666667e-06,
|
| 667 |
+
"loss": 0.1256,
|
| 668 |
"step": 2225
|
| 669 |
},
|
| 670 |
{
|
| 671 |
+
"epoch": 1.06825,
|
| 672 |
+
"grad_norm": 3.792029619216919,
|
| 673 |
+
"learning_rate": 4.793333333333334e-06,
|
| 674 |
+
"loss": 0.1322,
|
| 675 |
"step": 2250
|
| 676 |
},
|
| 677 |
{
|
| 678 |
+
"epoch": 1.071375,
|
| 679 |
+
"grad_norm": 4.919213771820068,
|
| 680 |
+
"learning_rate": 4.7725e-06,
|
| 681 |
+
"loss": 0.1477,
|
| 682 |
"step": 2275
|
| 683 |
},
|
| 684 |
{
|
| 685 |
+
"epoch": 1.0745,
|
| 686 |
+
"grad_norm": 5.92459774017334,
|
| 687 |
+
"learning_rate": 4.751666666666666e-06,
|
| 688 |
+
"loss": 0.1518,
|
| 689 |
"step": 2300
|
| 690 |
},
|
| 691 |
{
|
| 692 |
+
"epoch": 1.077625,
|
| 693 |
+
"grad_norm": 5.6381635665893555,
|
| 694 |
+
"learning_rate": 4.730833333333334e-06,
|
| 695 |
+
"loss": 0.1489,
|
| 696 |
"step": 2325
|
| 697 |
},
|
| 698 |
{
|
| 699 |
+
"epoch": 1.08075,
|
| 700 |
+
"grad_norm": 4.690910816192627,
|
| 701 |
+
"learning_rate": 4.710000000000001e-06,
|
| 702 |
+
"loss": 0.1216,
|
| 703 |
"step": 2350
|
| 704 |
},
|
| 705 |
{
|
| 706 |
+
"epoch": 1.083875,
|
| 707 |
+
"grad_norm": 3.847102165222168,
|
| 708 |
+
"learning_rate": 4.6891666666666665e-06,
|
| 709 |
+
"loss": 0.1243,
|
| 710 |
"step": 2375
|
| 711 |
},
|
| 712 |
{
|
| 713 |
+
"epoch": 1.087,
|
| 714 |
+
"grad_norm": 2.3183677196502686,
|
| 715 |
+
"learning_rate": 4.668333333333333e-06,
|
| 716 |
+
"loss": 0.1301,
|
| 717 |
"step": 2400
|
| 718 |
},
|
| 719 |
{
|
| 720 |
+
"epoch": 1.090125,
|
| 721 |
+
"grad_norm": 3.9808385372161865,
|
| 722 |
+
"learning_rate": 4.647500000000001e-06,
|
| 723 |
+
"loss": 0.1289,
|
| 724 |
"step": 2425
|
| 725 |
},
|
| 726 |
{
|
| 727 |
+
"epoch": 1.09325,
|
| 728 |
+
"grad_norm": 4.1279616355896,
|
| 729 |
+
"learning_rate": 4.626666666666667e-06,
|
| 730 |
+
"loss": 0.119,
|
| 731 |
"step": 2450
|
| 732 |
},
|
| 733 |
{
|
| 734 |
+
"epoch": 1.096375,
|
| 735 |
+
"grad_norm": 4.029735088348389,
|
| 736 |
+
"learning_rate": 4.605833333333333e-06,
|
| 737 |
+
"loss": 0.1235,
|
| 738 |
"step": 2475
|
| 739 |
},
|
| 740 |
{
|
| 741 |
+
"epoch": 1.0995,
|
| 742 |
+
"grad_norm": 5.72581148147583,
|
| 743 |
+
"learning_rate": 4.585e-06,
|
| 744 |
+
"loss": 0.125,
|
| 745 |
"step": 2500
|
| 746 |
},
|
| 747 |
{
|
| 748 |
+
"epoch": 1.0995,
|
| 749 |
+
"eval_loss": 0.16921907663345337,
|
| 750 |
+
"eval_runtime": 2457.6523,
|
| 751 |
+
"eval_samples_per_second": 5.546,
|
| 752 |
+
"eval_steps_per_second": 0.693,
|
| 753 |
+
"eval_wer": 11.114368426838652,
|
| 754 |
"step": 2500
|
| 755 |
},
|
| 756 |
{
|
| 757 |
+
"epoch": 1.102625,
|
| 758 |
+
"grad_norm": 4.8854875564575195,
|
| 759 |
+
"learning_rate": 4.564166666666667e-06,
|
| 760 |
+
"loss": 0.122,
|
| 761 |
"step": 2525
|
| 762 |
},
|
| 763 |
{
|
| 764 |
+
"epoch": 1.10575,
|
| 765 |
+
"grad_norm": 3.682393789291382,
|
| 766 |
+
"learning_rate": 4.543333333333334e-06,
|
| 767 |
+
"loss": 0.1169,
|
| 768 |
"step": 2550
|
| 769 |
},
|
| 770 |
{
|
| 771 |
+
"epoch": 1.108875,
|
| 772 |
+
"grad_norm": 3.8288307189941406,
|
| 773 |
+
"learning_rate": 4.5225e-06,
|
| 774 |
+
"loss": 0.1184,
|
| 775 |
"step": 2575
|
| 776 |
},
|
| 777 |
{
|
| 778 |
+
"epoch": 1.112,
|
| 779 |
+
"grad_norm": 5.098357677459717,
|
| 780 |
+
"learning_rate": 4.501666666666667e-06,
|
| 781 |
+
"loss": 0.1235,
|
| 782 |
"step": 2600
|
| 783 |
},
|
| 784 |
{
|
| 785 |
+
"epoch": 1.115125,
|
| 786 |
+
"grad_norm": 3.420466899871826,
|
| 787 |
+
"learning_rate": 4.480833333333334e-06,
|
| 788 |
+
"loss": 0.1277,
|
| 789 |
"step": 2625
|
| 790 |
},
|
| 791 |
{
|
| 792 |
+
"epoch": 1.11825,
|
| 793 |
+
"grad_norm": 4.542791843414307,
|
| 794 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 795 |
+
"loss": 0.1214,
|
| 796 |
"step": 2650
|
| 797 |
},
|
| 798 |
{
|
| 799 |
+
"epoch": 1.121375,
|
| 800 |
+
"grad_norm": 3.245755195617676,
|
| 801 |
+
"learning_rate": 4.439166666666667e-06,
|
| 802 |
+
"loss": 0.1109,
|
| 803 |
"step": 2675
|
| 804 |
},
|
| 805 |
{
|
| 806 |
+
"epoch": 1.1245,
|
| 807 |
+
"grad_norm": 5.041585445404053,
|
| 808 |
+
"learning_rate": 4.418333333333333e-06,
|
| 809 |
+
"loss": 0.1295,
|
| 810 |
"step": 2700
|
| 811 |
},
|
| 812 |
{
|
| 813 |
+
"epoch": 1.127625,
|
| 814 |
+
"grad_norm": 4.113171577453613,
|
| 815 |
+
"learning_rate": 4.3975e-06,
|
| 816 |
+
"loss": 0.1153,
|
| 817 |
"step": 2725
|
| 818 |
},
|
| 819 |
{
|
| 820 |
+
"epoch": 1.13075,
|
| 821 |
+
"grad_norm": 5.032025337219238,
|
| 822 |
+
"learning_rate": 4.376666666666667e-06,
|
| 823 |
+
"loss": 0.1204,
|
| 824 |
"step": 2750
|
| 825 |
},
|
| 826 |
{
|
| 827 |
+
"epoch": 1.133875,
|
| 828 |
+
"grad_norm": 5.080983638763428,
|
| 829 |
+
"learning_rate": 4.355833333333333e-06,
|
| 830 |
+
"loss": 0.1213,
|
| 831 |
"step": 2775
|
| 832 |
},
|
| 833 |
{
|
| 834 |
+
"epoch": 1.137,
|
| 835 |
+
"grad_norm": 4.304152011871338,
|
| 836 |
+
"learning_rate": 4.335e-06,
|
| 837 |
+
"loss": 0.1069,
|
| 838 |
"step": 2800
|
| 839 |
},
|
| 840 |
{
|
| 841 |
+
"epoch": 1.140125,
|
| 842 |
+
"grad_norm": 3.1360485553741455,
|
| 843 |
+
"learning_rate": 4.314166666666667e-06,
|
| 844 |
+
"loss": 0.1167,
|
| 845 |
"step": 2825
|
| 846 |
},
|
| 847 |
{
|
| 848 |
+
"epoch": 1.14325,
|
| 849 |
+
"grad_norm": 4.444480895996094,
|
| 850 |
+
"learning_rate": 4.2933333333333334e-06,
|
| 851 |
+
"loss": 0.1117,
|
| 852 |
"step": 2850
|
| 853 |
},
|
| 854 |
{
|
| 855 |
+
"epoch": 1.146375,
|
| 856 |
+
"grad_norm": 3.8886942863464355,
|
| 857 |
+
"learning_rate": 4.2725e-06,
|
| 858 |
+
"loss": 0.1134,
|
| 859 |
"step": 2875
|
| 860 |
},
|
| 861 |
{
|
| 862 |
+
"epoch": 1.1495,
|
| 863 |
+
"grad_norm": 4.360485076904297,
|
| 864 |
+
"learning_rate": 4.251666666666667e-06,
|
| 865 |
+
"loss": 0.1334,
|
| 866 |
"step": 2900
|
| 867 |
},
|
| 868 |
{
|
| 869 |
+
"epoch": 1.152625,
|
| 870 |
+
"grad_norm": 5.376945972442627,
|
| 871 |
+
"learning_rate": 4.230833333333334e-06,
|
| 872 |
+
"loss": 0.1163,
|
| 873 |
"step": 2925
|
| 874 |
},
|
| 875 |
{
|
| 876 |
+
"epoch": 1.15575,
|
| 877 |
+
"grad_norm": 4.359200954437256,
|
| 878 |
+
"learning_rate": 4.21e-06,
|
| 879 |
+
"loss": 0.1351,
|
| 880 |
"step": 2950
|
| 881 |
},
|
| 882 |
{
|
| 883 |
+
"epoch": 1.158875,
|
| 884 |
+
"grad_norm": 5.360095977783203,
|
| 885 |
+
"learning_rate": 4.189166666666667e-06,
|
| 886 |
+
"loss": 0.1113,
|
| 887 |
"step": 2975
|
| 888 |
},
|
| 889 |
{
|
| 890 |
+
"epoch": 1.162,
|
| 891 |
+
"grad_norm": 3.6014363765716553,
|
| 892 |
+
"learning_rate": 4.168333333333334e-06,
|
| 893 |
+
"loss": 0.12,
|
| 894 |
"step": 3000
|
| 895 |
},
|
| 896 |
{
|
| 897 |
+
"epoch": 1.162,
|
| 898 |
+
"eval_loss": 0.15999051928520203,
|
| 899 |
+
"eval_runtime": 2448.4453,
|
| 900 |
+
"eval_samples_per_second": 5.567,
|
| 901 |
+
"eval_steps_per_second": 0.696,
|
| 902 |
+
"eval_wer": 10.697533804829785,
|
| 903 |
"step": 3000
|
| 904 |
},
|
| 905 |
{
|
| 906 |
+
"epoch": 1.165125,
|
| 907 |
+
"grad_norm": 4.150473594665527,
|
| 908 |
+
"learning_rate": 4.1475e-06,
|
| 909 |
+
"loss": 0.1196,
|
| 910 |
"step": 3025
|
| 911 |
},
|
| 912 |
{
|
| 913 |
+
"epoch": 1.16825,
|
| 914 |
+
"grad_norm": 4.962148666381836,
|
| 915 |
+
"learning_rate": 4.126666666666667e-06,
|
| 916 |
+
"loss": 0.1261,
|
| 917 |
"step": 3050
|
| 918 |
},
|
| 919 |
{
|
| 920 |
+
"epoch": 1.171375,
|
| 921 |
+
"grad_norm": 2.9675393104553223,
|
| 922 |
+
"learning_rate": 4.105833333333334e-06,
|
| 923 |
+
"loss": 0.1087,
|
| 924 |
"step": 3075
|
| 925 |
},
|
| 926 |
{
|
| 927 |
+
"epoch": 1.1745,
|
| 928 |
+
"grad_norm": 5.842774868011475,
|
| 929 |
+
"learning_rate": 4.085e-06,
|
| 930 |
+
"loss": 0.1067,
|
| 931 |
"step": 3100
|
| 932 |
},
|
| 933 |
{
|
| 934 |
+
"epoch": 1.177625,
|
| 935 |
+
"grad_norm": 5.209335803985596,
|
| 936 |
+
"learning_rate": 4.064166666666667e-06,
|
| 937 |
+
"loss": 0.114,
|
| 938 |
"step": 3125
|
| 939 |
},
|
| 940 |
{
|
| 941 |
+
"epoch": 1.18075,
|
| 942 |
+
"grad_norm": 4.343747138977051,
|
| 943 |
+
"learning_rate": 4.043333333333333e-06,
|
| 944 |
+
"loss": 0.1398,
|
| 945 |
"step": 3150
|
| 946 |
},
|
| 947 |
{
|
| 948 |
+
"epoch": 1.183875,
|
| 949 |
+
"grad_norm": 4.5666890144348145,
|
| 950 |
+
"learning_rate": 4.0225e-06,
|
| 951 |
+
"loss": 0.112,
|
| 952 |
"step": 3175
|
| 953 |
},
|
| 954 |
{
|
| 955 |
+
"epoch": 1.187,
|
| 956 |
+
"grad_norm": 3.5483310222625732,
|
| 957 |
+
"learning_rate": 4.001666666666667e-06,
|
| 958 |
+
"loss": 0.1041,
|
| 959 |
"step": 3200
|
| 960 |
},
|
| 961 |
{
|
| 962 |
+
"epoch": 1.190125,
|
| 963 |
+
"grad_norm": 5.006927013397217,
|
| 964 |
+
"learning_rate": 3.9808333333333335e-06,
|
| 965 |
+
"loss": 0.1184,
|
| 966 |
"step": 3225
|
| 967 |
},
|
| 968 |
{
|
| 969 |
+
"epoch": 1.19325,
|
| 970 |
+
"grad_norm": 5.658267498016357,
|
| 971 |
+
"learning_rate": 3.96e-06,
|
| 972 |
+
"loss": 0.1141,
|
| 973 |
"step": 3250
|
| 974 |
},
|
| 975 |
{
|
| 976 |
+
"epoch": 1.196375,
|
| 977 |
+
"grad_norm": 2.7597854137420654,
|
| 978 |
+
"learning_rate": 3.939166666666667e-06,
|
| 979 |
+
"loss": 0.1107,
|
| 980 |
"step": 3275
|
| 981 |
},
|
| 982 |
{
|
| 983 |
+
"epoch": 1.1995,
|
| 984 |
+
"grad_norm": 4.687257289886475,
|
| 985 |
+
"learning_rate": 3.918333333333334e-06,
|
| 986 |
+
"loss": 0.1223,
|
| 987 |
"step": 3300
|
| 988 |
},
|
| 989 |
{
|
| 990 |
+
"epoch": 1.202625,
|
| 991 |
+
"grad_norm": 5.1347336769104,
|
| 992 |
+
"learning_rate": 3.8975e-06,
|
| 993 |
+
"loss": 0.1277,
|
| 994 |
"step": 3325
|
| 995 |
},
|
| 996 |
{
|
| 997 |
+
"epoch": 1.20575,
|
| 998 |
+
"grad_norm": 7.7792067527771,
|
| 999 |
+
"learning_rate": 3.876666666666666e-06,
|
| 1000 |
+
"loss": 0.1293,
|
| 1001 |
"step": 3350
|
| 1002 |
},
|
| 1003 |
{
|
| 1004 |
+
"epoch": 1.208875,
|
| 1005 |
+
"grad_norm": 3.461538791656494,
|
| 1006 |
+
"learning_rate": 3.855833333333334e-06,
|
| 1007 |
+
"loss": 0.1235,
|
| 1008 |
"step": 3375
|
| 1009 |
},
|
| 1010 |
{
|
| 1011 |
+
"epoch": 1.212,
|
| 1012 |
+
"grad_norm": 3.458538055419922,
|
| 1013 |
+
"learning_rate": 3.8350000000000006e-06,
|
| 1014 |
+
"loss": 0.1094,
|
| 1015 |
"step": 3400
|
| 1016 |
},
|
| 1017 |
{
|
| 1018 |
+
"epoch": 2.002125,
|
| 1019 |
+
"grad_norm": 6.5652313232421875,
|
| 1020 |
+
"learning_rate": 3.8141666666666664e-06,
|
| 1021 |
+
"loss": 0.0766,
|
| 1022 |
"step": 3425
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
+
"epoch": 2.00525,
|
| 1026 |
+
"grad_norm": 2.599691152572632,
|
| 1027 |
+
"learning_rate": 3.7933333333333336e-06,
|
| 1028 |
+
"loss": 0.0666,
|
| 1029 |
"step": 3450
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
+
"epoch": 2.008375,
|
| 1033 |
+
"grad_norm": 5.211915493011475,
|
| 1034 |
+
"learning_rate": 3.7725000000000003e-06,
|
| 1035 |
+
"loss": 0.0635,
|
| 1036 |
"step": 3475
|
| 1037 |
},
|
| 1038 |
{
|
| 1039 |
+
"epoch": 2.0115,
|
| 1040 |
+
"grad_norm": 3.1992459297180176,
|
| 1041 |
+
"learning_rate": 3.7516666666666666e-06,
|
| 1042 |
+
"loss": 0.069,
|
| 1043 |
"step": 3500
|
| 1044 |
},
|
| 1045 |
{
|
| 1046 |
+
"epoch": 2.0115,
|
| 1047 |
+
"eval_loss": 0.15404893457889557,
|
| 1048 |
+
"eval_runtime": 2458.0949,
|
| 1049 |
+
"eval_samples_per_second": 5.545,
|
| 1050 |
+
"eval_steps_per_second": 0.693,
|
| 1051 |
+
"eval_wer": 9.764923595587966,
|
| 1052 |
"step": 3500
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
+
"epoch": 2.014625,
|
| 1056 |
+
"grad_norm": 3.740165948867798,
|
| 1057 |
+
"learning_rate": 3.7308333333333333e-06,
|
| 1058 |
+
"loss": 0.0699,
|
| 1059 |
"step": 3525
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
+
"epoch": 2.01775,
|
| 1063 |
+
"grad_norm": 2.624305009841919,
|
| 1064 |
+
"learning_rate": 3.7100000000000005e-06,
|
| 1065 |
+
"loss": 0.0644,
|
| 1066 |
"step": 3550
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
+
"epoch": 2.020875,
|
| 1070 |
+
"grad_norm": 3.66835355758667,
|
| 1071 |
+
"learning_rate": 3.689166666666667e-06,
|
| 1072 |
+
"loss": 0.0552,
|
| 1073 |
"step": 3575
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
+
"epoch": 2.024,
|
| 1077 |
+
"grad_norm": 3.7992517948150635,
|
| 1078 |
+
"learning_rate": 3.6683333333333335e-06,
|
| 1079 |
+
"loss": 0.0715,
|
| 1080 |
"step": 3600
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
+
"epoch": 2.027125,
|
| 1084 |
+
"grad_norm": 3.4328665733337402,
|
| 1085 |
+
"learning_rate": 3.6475000000000002e-06,
|
| 1086 |
+
"loss": 0.0642,
|
| 1087 |
"step": 3625
|
| 1088 |
},
|
| 1089 |
{
|
| 1090 |
+
"epoch": 2.03025,
|
| 1091 |
+
"grad_norm": 3.206305503845215,
|
| 1092 |
+
"learning_rate": 3.626666666666667e-06,
|
| 1093 |
+
"loss": 0.068,
|
| 1094 |
"step": 3650
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
+
"epoch": 2.033375,
|
| 1098 |
+
"grad_norm": 3.6906914710998535,
|
| 1099 |
+
"learning_rate": 3.6058333333333333e-06,
|
| 1100 |
+
"loss": 0.065,
|
| 1101 |
"step": 3675
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
+
"epoch": 2.0365,
|
| 1105 |
+
"grad_norm": 3.6564929485321045,
|
| 1106 |
+
"learning_rate": 3.585e-06,
|
| 1107 |
+
"loss": 0.0634,
|
| 1108 |
"step": 3700
|
| 1109 |
},
|
| 1110 |
{
|
| 1111 |
+
"epoch": 2.039625,
|
| 1112 |
+
"grad_norm": 2.8516416549682617,
|
| 1113 |
+
"learning_rate": 3.564166666666667e-06,
|
| 1114 |
+
"loss": 0.0632,
|
| 1115 |
"step": 3725
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
+
"epoch": 2.04275,
|
| 1119 |
+
"grad_norm": 2.905165672302246,
|
| 1120 |
+
"learning_rate": 3.5433333333333334e-06,
|
| 1121 |
+
"loss": 0.0528,
|
| 1122 |
"step": 3750
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
+
"epoch": 2.045875,
|
| 1126 |
+
"grad_norm": 2.2724170684814453,
|
| 1127 |
+
"learning_rate": 3.5225e-06,
|
| 1128 |
+
"loss": 0.0604,
|
| 1129 |
"step": 3775
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
+
"epoch": 2.049,
|
| 1133 |
+
"grad_norm": 2.6760237216949463,
|
| 1134 |
+
"learning_rate": 3.501666666666667e-06,
|
| 1135 |
+
"loss": 0.0528,
|
| 1136 |
"step": 3800
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
+
"epoch": 2.052125,
|
| 1140 |
+
"grad_norm": 4.052576541900635,
|
| 1141 |
+
"learning_rate": 3.480833333333333e-06,
|
| 1142 |
+
"loss": 0.0535,
|
| 1143 |
"step": 3825
|
| 1144 |
},
|
| 1145 |
{
|
| 1146 |
+
"epoch": 2.05525,
|
| 1147 |
+
"grad_norm": 3.6434435844421387,
|
| 1148 |
+
"learning_rate": 3.46e-06,
|
| 1149 |
+
"loss": 0.0499,
|
| 1150 |
"step": 3850
|
| 1151 |
},
|
| 1152 |
{
|
| 1153 |
+
"epoch": 2.058375,
|
| 1154 |
+
"grad_norm": 2.612650156021118,
|
| 1155 |
+
"learning_rate": 3.439166666666667e-06,
|
| 1156 |
+
"loss": 0.0489,
|
| 1157 |
"step": 3875
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
+
"epoch": 2.0615,
|
| 1161 |
+
"grad_norm": 1.6679494380950928,
|
| 1162 |
+
"learning_rate": 3.4183333333333338e-06,
|
| 1163 |
+
"loss": 0.0475,
|
| 1164 |
"step": 3900
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
+
"epoch": 2.064625,
|
| 1168 |
+
"grad_norm": 2.7951669692993164,
|
| 1169 |
+
"learning_rate": 3.3975e-06,
|
| 1170 |
+
"loss": 0.0517,
|
| 1171 |
"step": 3925
|
| 1172 |
},
|
| 1173 |
{
|
| 1174 |
+
"epoch": 2.06775,
|
| 1175 |
+
"grad_norm": 2.6957435607910156,
|
| 1176 |
+
"learning_rate": 3.376666666666667e-06,
|
| 1177 |
+
"loss": 0.0508,
|
| 1178 |
"step": 3950
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
+
"epoch": 2.070875,
|
| 1182 |
+
"grad_norm": 2.930025577545166,
|
| 1183 |
+
"learning_rate": 3.355833333333334e-06,
|
| 1184 |
+
"loss": 0.0554,
|
| 1185 |
"step": 3975
|
| 1186 |
},
|
| 1187 |
{
|
| 1188 |
+
"epoch": 2.074,
|
| 1189 |
+
"grad_norm": 2.0811214447021484,
|
| 1190 |
+
"learning_rate": 3.335e-06,
|
| 1191 |
+
"loss": 0.0606,
|
| 1192 |
"step": 4000
|
| 1193 |
},
|
| 1194 |
{
|
| 1195 |
+
"epoch": 2.074,
|
| 1196 |
+
"eval_loss": 0.1549645960330963,
|
| 1197 |
+
"eval_runtime": 2475.9667,
|
| 1198 |
+
"eval_samples_per_second": 5.505,
|
| 1199 |
+
"eval_steps_per_second": 0.688,
|
| 1200 |
+
"eval_wer": 9.819890798490235,
|
| 1201 |
"step": 4000
|
| 1202 |
},
|
| 1203 |
{
|
| 1204 |
+
"epoch": 2.077125,
|
| 1205 |
+
"grad_norm": 2.389423370361328,
|
| 1206 |
+
"learning_rate": 3.314166666666667e-06,
|
| 1207 |
+
"loss": 0.0551,
|
| 1208 |
"step": 4025
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
+
"epoch": 2.08025,
|
| 1212 |
+
"grad_norm": 1.8350708484649658,
|
| 1213 |
+
"learning_rate": 3.2933333333333337e-06,
|
| 1214 |
+
"loss": 0.0501,
|
| 1215 |
"step": 4050
|
| 1216 |
},
|
| 1217 |
{
|
| 1218 |
+
"epoch": 2.083375,
|
| 1219 |
+
"grad_norm": 3.053312063217163,
|
| 1220 |
+
"learning_rate": 3.2725e-06,
|
| 1221 |
+
"loss": 0.0526,
|
| 1222 |
"step": 4075
|
| 1223 |
},
|
| 1224 |
{
|
| 1225 |
+
"epoch": 2.0865,
|
| 1226 |
+
"grad_norm": 3.489823341369629,
|
| 1227 |
+
"learning_rate": 3.2516666666666667e-06,
|
| 1228 |
+
"loss": 0.0476,
|
| 1229 |
"step": 4100
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
+
"epoch": 2.089625,
|
| 1233 |
+
"grad_norm": 3.742950916290283,
|
| 1234 |
+
"learning_rate": 3.2308333333333335e-06,
|
| 1235 |
+
"loss": 0.0614,
|
| 1236 |
"step": 4125
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
+
"epoch": 2.09275,
|
| 1240 |
+
"grad_norm": 1.9854848384857178,
|
| 1241 |
+
"learning_rate": 3.2099999999999998e-06,
|
| 1242 |
+
"loss": 0.0487,
|
| 1243 |
"step": 4150
|
| 1244 |
},
|
| 1245 |
{
|
| 1246 |
+
"epoch": 2.095875,
|
| 1247 |
+
"grad_norm": 2.3751885890960693,
|
| 1248 |
+
"learning_rate": 3.1891666666666665e-06,
|
| 1249 |
+
"loss": 0.0496,
|
| 1250 |
"step": 4175
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
+
"epoch": 2.099,
|
| 1254 |
+
"grad_norm": 2.182997703552246,
|
| 1255 |
+
"learning_rate": 3.1683333333333336e-06,
|
| 1256 |
+
"loss": 0.0442,
|
| 1257 |
"step": 4200
|
| 1258 |
},
|
| 1259 |
{
|
| 1260 |
+
"epoch": 2.102125,
|
| 1261 |
+
"grad_norm": 1.963120937347412,
|
| 1262 |
+
"learning_rate": 3.1475000000000004e-06,
|
| 1263 |
+
"loss": 0.0437,
|
| 1264 |
"step": 4225
|
| 1265 |
},
|
| 1266 |
{
|
| 1267 |
+
"epoch": 2.10525,
|
| 1268 |
+
"grad_norm": 1.7127807140350342,
|
| 1269 |
+
"learning_rate": 3.1266666666666667e-06,
|
| 1270 |
+
"loss": 0.0479,
|
| 1271 |
"step": 4250
|
| 1272 |
},
|
| 1273 |
{
|
| 1274 |
+
"epoch": 2.108375,
|
| 1275 |
+
"grad_norm": 3.277400016784668,
|
| 1276 |
+
"learning_rate": 3.1058333333333334e-06,
|
| 1277 |
+
"loss": 0.0535,
|
| 1278 |
"step": 4275
|
| 1279 |
},
|
| 1280 |
{
|
| 1281 |
+
"epoch": 2.1115,
|
| 1282 |
+
"grad_norm": 3.1404621601104736,
|
| 1283 |
+
"learning_rate": 3.085e-06,
|
| 1284 |
+
"loss": 0.0431,
|
| 1285 |
"step": 4300
|
| 1286 |
},
|
| 1287 |
{
|
| 1288 |
+
"epoch": 2.114625,
|
| 1289 |
+
"grad_norm": 3.8738794326782227,
|
| 1290 |
+
"learning_rate": 3.064166666666667e-06,
|
| 1291 |
+
"loss": 0.0436,
|
| 1292 |
"step": 4325
|
| 1293 |
},
|
| 1294 |
{
|
| 1295 |
+
"epoch": 2.11775,
|
| 1296 |
+
"grad_norm": 2.9267311096191406,
|
| 1297 |
+
"learning_rate": 3.0433333333333336e-06,
|
| 1298 |
+
"loss": 0.0496,
|
| 1299 |
"step": 4350
|
| 1300 |
},
|
| 1301 |
{
|
| 1302 |
+
"epoch": 2.120875,
|
| 1303 |
+
"grad_norm": 4.0710320472717285,
|
| 1304 |
+
"learning_rate": 3.0225e-06,
|
| 1305 |
+
"loss": 0.047,
|
| 1306 |
"step": 4375
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
+
"epoch": 2.124,
|
| 1310 |
+
"grad_norm": 2.9265050888061523,
|
| 1311 |
+
"learning_rate": 3.001666666666667e-06,
|
| 1312 |
+
"loss": 0.0434,
|
| 1313 |
"step": 4400
|
| 1314 |
},
|
| 1315 |
{
|
| 1316 |
+
"epoch": 2.127125,
|
| 1317 |
+
"grad_norm": 2.666417121887207,
|
| 1318 |
+
"learning_rate": 2.9808333333333333e-06,
|
| 1319 |
+
"loss": 0.0357,
|
| 1320 |
"step": 4425
|
| 1321 |
},
|
| 1322 |
{
|
| 1323 |
+
"epoch": 2.13025,
|
| 1324 |
+
"grad_norm": 3.082674741744995,
|
| 1325 |
+
"learning_rate": 2.9600000000000005e-06,
|
| 1326 |
+
"loss": 0.0486,
|
| 1327 |
"step": 4450
|
| 1328 |
},
|
| 1329 |
{
|
| 1330 |
+
"epoch": 2.133375,
|
| 1331 |
+
"grad_norm": 2.291168689727783,
|
| 1332 |
+
"learning_rate": 2.9391666666666668e-06,
|
| 1333 |
+
"loss": 0.046,
|
| 1334 |
"step": 4475
|
| 1335 |
},
|
| 1336 |
{
|
| 1337 |
+
"epoch": 2.1365,
|
| 1338 |
+
"grad_norm": 2.878455877304077,
|
| 1339 |
+
"learning_rate": 2.9183333333333335e-06,
|
| 1340 |
+
"loss": 0.0434,
|
| 1341 |
"step": 4500
|
| 1342 |
},
|
| 1343 |
{
|
| 1344 |
+
"epoch": 2.1365,
|
| 1345 |
+
"eval_loss": 0.15800228714942932,
|
| 1346 |
+
"eval_runtime": 2430.3639,
|
| 1347 |
+
"eval_samples_per_second": 5.608,
|
| 1348 |
+
"eval_steps_per_second": 0.701,
|
| 1349 |
+
"eval_wer": 9.457107259335263,
|
| 1350 |
"step": 4500
|
| 1351 |
},
|
| 1352 |
{
|
| 1353 |
+
"epoch": 2.139625,
|
| 1354 |
+
"grad_norm": 2.2872824668884277,
|
| 1355 |
+
"learning_rate": 2.8975e-06,
|
| 1356 |
+
"loss": 0.0507,
|
| 1357 |
"step": 4525
|
| 1358 |
},
|
| 1359 |
{
|
| 1360 |
+
"epoch": 2.14275,
|
| 1361 |
+
"grad_norm": 1.6797173023223877,
|
| 1362 |
+
"learning_rate": 2.876666666666667e-06,
|
| 1363 |
+
"loss": 0.044,
|
| 1364 |
"step": 4550
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
+
"epoch": 2.145875,
|
| 1368 |
+
"grad_norm": 2.690349817276001,
|
| 1369 |
+
"learning_rate": 2.8558333333333337e-06,
|
| 1370 |
+
"loss": 0.041,
|
| 1371 |
"step": 4575
|
| 1372 |
},
|
| 1373 |
{
|
| 1374 |
+
"epoch": 2.149,
|
| 1375 |
+
"grad_norm": 2.515873670578003,
|
| 1376 |
+
"learning_rate": 2.835e-06,
|
| 1377 |
+
"loss": 0.0503,
|
| 1378 |
"step": 4600
|
| 1379 |
},
|
| 1380 |
{
|
| 1381 |
+
"epoch": 2.152125,
|
| 1382 |
+
"grad_norm": 3.210235595703125,
|
| 1383 |
+
"learning_rate": 2.8141666666666667e-06,
|
| 1384 |
+
"loss": 0.0539,
|
| 1385 |
"step": 4625
|
| 1386 |
},
|
| 1387 |
{
|
| 1388 |
+
"epoch": 2.15525,
|
| 1389 |
+
"grad_norm": 2.19775128364563,
|
| 1390 |
+
"learning_rate": 2.7933333333333334e-06,
|
| 1391 |
+
"loss": 0.0486,
|
| 1392 |
"step": 4650
|
| 1393 |
},
|
| 1394 |
{
|
| 1395 |
+
"epoch": 2.158375,
|
| 1396 |
+
"grad_norm": 2.8275535106658936,
|
| 1397 |
+
"learning_rate": 2.7725e-06,
|
| 1398 |
+
"loss": 0.0455,
|
| 1399 |
"step": 4675
|
| 1400 |
},
|
| 1401 |
{
|
| 1402 |
+
"epoch": 2.1615,
|
| 1403 |
+
"grad_norm": 2.911288022994995,
|
| 1404 |
+
"learning_rate": 2.7516666666666664e-06,
|
| 1405 |
+
"loss": 0.0495,
|
| 1406 |
"step": 4700
|
| 1407 |
},
|
| 1408 |
{
|
| 1409 |
+
"epoch": 2.164625,
|
| 1410 |
+
"grad_norm": 3.446293592453003,
|
| 1411 |
+
"learning_rate": 2.7308333333333336e-06,
|
| 1412 |
+
"loss": 0.0527,
|
| 1413 |
"step": 4725
|
| 1414 |
},
|
| 1415 |
{
|
| 1416 |
+
"epoch": 2.16775,
|
| 1417 |
+
"grad_norm": 2.1223082542419434,
|
| 1418 |
+
"learning_rate": 2.71e-06,
|
| 1419 |
+
"loss": 0.0437,
|
| 1420 |
"step": 4750
|
| 1421 |
},
|
| 1422 |
{
|
| 1423 |
+
"epoch": 2.170875,
|
| 1424 |
+
"grad_norm": 2.4212393760681152,
|
| 1425 |
+
"learning_rate": 2.689166666666667e-06,
|
| 1426 |
+
"loss": 0.0504,
|
| 1427 |
"step": 4775
|
| 1428 |
},
|
| 1429 |
{
|
| 1430 |
+
"epoch": 2.174,
|
| 1431 |
+
"grad_norm": 2.440826177597046,
|
| 1432 |
+
"learning_rate": 2.6683333333333333e-06,
|
| 1433 |
+
"loss": 0.0432,
|
| 1434 |
"step": 4800
|
| 1435 |
},
|
| 1436 |
{
|
| 1437 |
+
"epoch": 2.177125,
|
| 1438 |
+
"grad_norm": 2.3568456172943115,
|
| 1439 |
+
"learning_rate": 2.6475e-06,
|
| 1440 |
+
"loss": 0.0404,
|
| 1441 |
"step": 4825
|
| 1442 |
},
|
| 1443 |
{
|
| 1444 |
+
"epoch": 2.18025,
|
| 1445 |
+
"grad_norm": 2.638301372528076,
|
| 1446 |
+
"learning_rate": 2.6266666666666668e-06,
|
| 1447 |
+
"loss": 0.0528,
|
| 1448 |
"step": 4850
|
| 1449 |
},
|
| 1450 |
{
|
| 1451 |
+
"epoch": 2.183375,
|
| 1452 |
+
"grad_norm": 2.1262662410736084,
|
| 1453 |
+
"learning_rate": 2.6058333333333335e-06,
|
| 1454 |
+
"loss": 0.0468,
|
| 1455 |
"step": 4875
|
| 1456 |
},
|
| 1457 |
{
|
| 1458 |
+
"epoch": 2.1865,
|
| 1459 |
+
"grad_norm": 1.8121278285980225,
|
| 1460 |
+
"learning_rate": 2.5850000000000002e-06,
|
| 1461 |
+
"loss": 0.0414,
|
| 1462 |
"step": 4900
|
| 1463 |
},
|
| 1464 |
{
|
| 1465 |
+
"epoch": 2.189625,
|
| 1466 |
+
"grad_norm": 3.5271337032318115,
|
| 1467 |
+
"learning_rate": 2.564166666666667e-06,
|
| 1468 |
+
"loss": 0.0477,
|
| 1469 |
"step": 4925
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
+
"epoch": 2.19275,
|
| 1473 |
+
"grad_norm": 2.0672607421875,
|
| 1474 |
+
"learning_rate": 2.5433333333333333e-06,
|
| 1475 |
+
"loss": 0.0402,
|
| 1476 |
"step": 4950
|
| 1477 |
},
|
| 1478 |
{
|
| 1479 |
+
"epoch": 2.195875,
|
| 1480 |
+
"grad_norm": 2.3378803730010986,
|
| 1481 |
+
"learning_rate": 2.5225000000000004e-06,
|
| 1482 |
+
"loss": 0.0458,
|
| 1483 |
"step": 4975
|
| 1484 |
},
|
| 1485 |
{
|
| 1486 |
+
"epoch": 2.199,
|
| 1487 |
+
"grad_norm": 2.3618690967559814,
|
| 1488 |
+
"learning_rate": 2.5016666666666667e-06,
|
| 1489 |
+
"loss": 0.0455,
|
| 1490 |
"step": 5000
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
+
"epoch": 2.199,
|
| 1494 |
+
"eval_loss": 0.15326862037181854,
|
| 1495 |
+
"eval_runtime": 2431.3758,
|
| 1496 |
+
"eval_samples_per_second": 5.606,
|
| 1497 |
+
"eval_steps_per_second": 0.701,
|
| 1498 |
+
"eval_wer": 9.14104584264722,
|
| 1499 |
"step": 5000
|
| 1500 |
},
|
| 1501 |
{
|
| 1502 |
+
"epoch": 2.202125,
|
| 1503 |
+
"grad_norm": 3.066169261932373,
|
| 1504 |
+
"learning_rate": 2.4808333333333334e-06,
|
| 1505 |
+
"loss": 0.0514,
|
| 1506 |
"step": 5025
|
| 1507 |
},
|
| 1508 |
{
|
| 1509 |
+
"epoch": 2.20525,
|
| 1510 |
+
"grad_norm": 2.8991472721099854,
|
| 1511 |
+
"learning_rate": 2.46e-06,
|
| 1512 |
+
"loss": 0.0454,
|
| 1513 |
"step": 5050
|
| 1514 |
},
|
| 1515 |
{
|
| 1516 |
+
"epoch": 2.208375,
|
| 1517 |
+
"grad_norm": 2.641075611114502,
|
| 1518 |
+
"learning_rate": 2.439166666666667e-06,
|
| 1519 |
+
"loss": 0.0448,
|
| 1520 |
"step": 5075
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
+
"epoch": 2.2115,
|
| 1524 |
+
"grad_norm": 3.234225034713745,
|
| 1525 |
+
"learning_rate": 2.4183333333333336e-06,
|
| 1526 |
+
"loss": 0.0504,
|
| 1527 |
"step": 5100
|
| 1528 |
},
|
| 1529 |
{
|
| 1530 |
+
"epoch": 3.001625,
|
| 1531 |
+
"grad_norm": 1.9421311616897583,
|
| 1532 |
+
"learning_rate": 2.3975e-06,
|
| 1533 |
+
"loss": 0.0377,
|
| 1534 |
"step": 5125
|
| 1535 |
},
|
| 1536 |
{
|
| 1537 |
+
"epoch": 3.00475,
|
| 1538 |
+
"grad_norm": 1.9280989170074463,
|
| 1539 |
+
"learning_rate": 2.3766666666666666e-06,
|
| 1540 |
+
"loss": 0.0291,
|
| 1541 |
"step": 5150
|
| 1542 |
},
|
| 1543 |
{
|
| 1544 |
+
"epoch": 3.007875,
|
| 1545 |
+
"grad_norm": 1.701106309890747,
|
| 1546 |
+
"learning_rate": 2.3558333333333334e-06,
|
| 1547 |
+
"loss": 0.0254,
|
| 1548 |
"step": 5175
|
| 1549 |
},
|
| 1550 |
{
|
| 1551 |
+
"epoch": 3.011,
|
| 1552 |
+
"grad_norm": 1.4093270301818848,
|
| 1553 |
+
"learning_rate": 2.335e-06,
|
| 1554 |
+
"loss": 0.0267,
|
| 1555 |
"step": 5200
|
| 1556 |
},
|
| 1557 |
{
|
| 1558 |
+
"epoch": 3.014125,
|
| 1559 |
+
"grad_norm": 2.7806079387664795,
|
| 1560 |
+
"learning_rate": 2.314166666666667e-06,
|
| 1561 |
+
"loss": 0.0271,
|
| 1562 |
"step": 5225
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
+
"epoch": 3.01725,
|
| 1566 |
+
"grad_norm": 2.454941511154175,
|
| 1567 |
+
"learning_rate": 2.2933333333333335e-06,
|
| 1568 |
+
"loss": 0.0296,
|
| 1569 |
"step": 5250
|
| 1570 |
},
|
| 1571 |
{
|
| 1572 |
+
"epoch": 3.020375,
|
| 1573 |
+
"grad_norm": 1.8080620765686035,
|
| 1574 |
+
"learning_rate": 2.2725e-06,
|
| 1575 |
+
"loss": 0.0254,
|
| 1576 |
"step": 5275
|
| 1577 |
},
|
| 1578 |
{
|
| 1579 |
+
"epoch": 3.0235,
|
| 1580 |
+
"grad_norm": 2.4920754432678223,
|
| 1581 |
+
"learning_rate": 2.251666666666667e-06,
|
| 1582 |
+
"loss": 0.0242,
|
| 1583 |
"step": 5300
|
| 1584 |
},
|
| 1585 |
{
|
| 1586 |
+
"epoch": 3.026625,
|
| 1587 |
+
"grad_norm": 1.4802467823028564,
|
| 1588 |
+
"learning_rate": 2.2308333333333333e-06,
|
| 1589 |
+
"loss": 0.0242,
|
| 1590 |
"step": 5325
|
| 1591 |
},
|
| 1592 |
{
|
| 1593 |
+
"epoch": 3.02975,
|
| 1594 |
+
"grad_norm": 4.049746036529541,
|
| 1595 |
+
"learning_rate": 2.2100000000000004e-06,
|
| 1596 |
+
"loss": 0.026,
|
| 1597 |
"step": 5350
|
| 1598 |
},
|
| 1599 |
{
|
| 1600 |
+
"epoch": 3.032875,
|
| 1601 |
+
"grad_norm": 2.304689884185791,
|
| 1602 |
+
"learning_rate": 2.1891666666666667e-06,
|
| 1603 |
+
"loss": 0.0261,
|
| 1604 |
"step": 5375
|
| 1605 |
},
|
| 1606 |
{
|
| 1607 |
+
"epoch": 3.036,
|
| 1608 |
+
"grad_norm": 1.9863944053649902,
|
| 1609 |
+
"learning_rate": 2.1683333333333335e-06,
|
| 1610 |
+
"loss": 0.0251,
|
| 1611 |
"step": 5400
|
| 1612 |
},
|
| 1613 |
{
|
| 1614 |
+
"epoch": 3.039125,
|
| 1615 |
+
"grad_norm": 2.0497422218322754,
|
| 1616 |
+
"learning_rate": 2.1475e-06,
|
| 1617 |
+
"loss": 0.0254,
|
| 1618 |
"step": 5425
|
| 1619 |
},
|
| 1620 |
{
|
| 1621 |
+
"epoch": 3.04225,
|
| 1622 |
+
"grad_norm": 2.5717766284942627,
|
| 1623 |
+
"learning_rate": 2.126666666666667e-06,
|
| 1624 |
+
"loss": 0.0238,
|
| 1625 |
"step": 5450
|
| 1626 |
},
|
| 1627 |
{
|
| 1628 |
+
"epoch": 3.045375,
|
| 1629 |
+
"grad_norm": 1.1389164924621582,
|
| 1630 |
+
"learning_rate": 2.1058333333333332e-06,
|
| 1631 |
+
"loss": 0.0208,
|
| 1632 |
"step": 5475
|
| 1633 |
},
|
| 1634 |
{
|
| 1635 |
+
"epoch": 3.0485,
|
| 1636 |
+
"grad_norm": 1.5230340957641602,
|
| 1637 |
+
"learning_rate": 2.085e-06,
|
| 1638 |
+
"loss": 0.0216,
|
| 1639 |
"step": 5500
|
| 1640 |
},
|
| 1641 |
{
|
| 1642 |
+
"epoch": 3.0485,
|
| 1643 |
+
"eval_loss": 0.1619918942451477,
|
| 1644 |
+
"eval_runtime": 2437.6868,
|
| 1645 |
+
"eval_samples_per_second": 5.591,
|
| 1646 |
+
"eval_steps_per_second": 0.699,
|
| 1647 |
+
"eval_wer": 9.08424639964821,
|
| 1648 |
"step": 5500
|
| 1649 |
},
|
| 1650 |
{
|
| 1651 |
+
"epoch": 3.051625,
|
| 1652 |
+
"grad_norm": 1.907436728477478,
|
| 1653 |
+
"learning_rate": 2.0641666666666667e-06,
|
| 1654 |
+
"loss": 0.0224,
|
| 1655 |
"step": 5525
|
| 1656 |
},
|
| 1657 |
{
|
| 1658 |
+
"epoch": 3.05475,
|
| 1659 |
+
"grad_norm": 1.5222604274749756,
|
| 1660 |
+
"learning_rate": 2.0433333333333334e-06,
|
| 1661 |
+
"loss": 0.0197,
|
| 1662 |
"step": 5550
|
| 1663 |
},
|
| 1664 |
{
|
| 1665 |
+
"epoch": 3.057875,
|
| 1666 |
+
"grad_norm": 1.5284624099731445,
|
| 1667 |
+
"learning_rate": 2.0225e-06,
|
| 1668 |
+
"loss": 0.0152,
|
| 1669 |
"step": 5575
|
| 1670 |
},
|
| 1671 |
{
|
| 1672 |
+
"epoch": 3.061,
|
| 1673 |
+
"grad_norm": 0.7098265886306763,
|
| 1674 |
+
"learning_rate": 2.0016666666666664e-06,
|
| 1675 |
+
"loss": 0.019,
|
| 1676 |
"step": 5600
|
| 1677 |
},
|
| 1678 |
{
|
| 1679 |
+
"epoch": 3.064125,
|
| 1680 |
+
"grad_norm": 2.1706318855285645,
|
| 1681 |
+
"learning_rate": 1.9808333333333336e-06,
|
| 1682 |
+
"loss": 0.0189,
|
| 1683 |
"step": 5625
|
| 1684 |
},
|
| 1685 |
{
|
| 1686 |
+
"epoch": 3.06725,
|
| 1687 |
+
"grad_norm": 2.888962745666504,
|
| 1688 |
+
"learning_rate": 1.96e-06,
|
| 1689 |
+
"loss": 0.0201,
|
| 1690 |
"step": 5650
|
| 1691 |
},
|
| 1692 |
{
|
| 1693 |
+
"epoch": 3.070375,
|
| 1694 |
+
"grad_norm": 3.0322179794311523,
|
| 1695 |
+
"learning_rate": 1.939166666666667e-06,
|
| 1696 |
+
"loss": 0.0211,
|
| 1697 |
"step": 5675
|
| 1698 |
},
|
| 1699 |
{
|
| 1700 |
+
"epoch": 3.0735,
|
| 1701 |
+
"grad_norm": 3.066394329071045,
|
| 1702 |
+
"learning_rate": 1.9183333333333333e-06,
|
| 1703 |
+
"loss": 0.0253,
|
| 1704 |
"step": 5700
|
| 1705 |
},
|
| 1706 |
{
|
| 1707 |
+
"epoch": 3.076625,
|
| 1708 |
+
"grad_norm": 2.5032601356506348,
|
| 1709 |
+
"learning_rate": 1.8975e-06,
|
| 1710 |
+
"loss": 0.0236,
|
| 1711 |
"step": 5725
|
| 1712 |
},
|
| 1713 |
{
|
| 1714 |
+
"epoch": 3.07975,
|
| 1715 |
+
"grad_norm": 1.5625773668289185,
|
| 1716 |
+
"learning_rate": 1.8766666666666668e-06,
|
| 1717 |
+
"loss": 0.0193,
|
| 1718 |
"step": 5750
|
| 1719 |
},
|
| 1720 |
{
|
| 1721 |
+
"epoch": 3.082875,
|
| 1722 |
+
"grad_norm": 1.7981406450271606,
|
| 1723 |
+
"learning_rate": 1.8558333333333333e-06,
|
| 1724 |
+
"loss": 0.0157,
|
| 1725 |
"step": 5775
|
| 1726 |
},
|
| 1727 |
{
|
| 1728 |
+
"epoch": 3.086,
|
| 1729 |
+
"grad_norm": 1.1411957740783691,
|
| 1730 |
+
"learning_rate": 1.8350000000000002e-06,
|
| 1731 |
+
"loss": 0.0217,
|
| 1732 |
"step": 5800
|
| 1733 |
},
|
| 1734 |
{
|
| 1735 |
+
"epoch": 3.089125,
|
| 1736 |
+
"grad_norm": 1.1639596223831177,
|
| 1737 |
+
"learning_rate": 1.8141666666666667e-06,
|
| 1738 |
+
"loss": 0.0239,
|
| 1739 |
"step": 5825
|
| 1740 |
},
|
| 1741 |
{
|
| 1742 |
+
"epoch": 3.09225,
|
| 1743 |
+
"grad_norm": 0.7168381810188293,
|
| 1744 |
+
"learning_rate": 1.7933333333333332e-06,
|
| 1745 |
+
"loss": 0.0218,
|
| 1746 |
"step": 5850
|
| 1747 |
},
|
| 1748 |
{
|
| 1749 |
+
"epoch": 3.095375,
|
| 1750 |
+
"grad_norm": 1.1308009624481201,
|
| 1751 |
+
"learning_rate": 1.7725000000000002e-06,
|
| 1752 |
+
"loss": 0.0214,
|
| 1753 |
"step": 5875
|
| 1754 |
},
|
| 1755 |
{
|
| 1756 |
+
"epoch": 3.0985,
|
| 1757 |
+
"grad_norm": 2.1142430305480957,
|
| 1758 |
+
"learning_rate": 1.7516666666666667e-06,
|
| 1759 |
+
"loss": 0.0155,
|
| 1760 |
"step": 5900
|
| 1761 |
},
|
| 1762 |
{
|
| 1763 |
+
"epoch": 3.101625,
|
| 1764 |
+
"grad_norm": 1.479252815246582,
|
| 1765 |
+
"learning_rate": 1.7308333333333332e-06,
|
| 1766 |
+
"loss": 0.0176,
|
| 1767 |
"step": 5925
|
| 1768 |
},
|
| 1769 |
{
|
| 1770 |
+
"epoch": 3.10475,
|
| 1771 |
+
"grad_norm": 1.3254590034484863,
|
| 1772 |
+
"learning_rate": 1.7100000000000001e-06,
|
| 1773 |
+
"loss": 0.0144,
|
| 1774 |
"step": 5950
|
| 1775 |
},
|
| 1776 |
{
|
| 1777 |
+
"epoch": 3.107875,
|
| 1778 |
+
"grad_norm": 2.1925458908081055,
|
| 1779 |
+
"learning_rate": 1.6891666666666667e-06,
|
| 1780 |
+
"loss": 0.0162,
|
| 1781 |
"step": 5975
|
| 1782 |
},
|
| 1783 |
{
|
| 1784 |
+
"epoch": 3.111,
|
| 1785 |
+
"grad_norm": 1.774483561515808,
|
| 1786 |
+
"learning_rate": 1.6683333333333336e-06,
|
| 1787 |
+
"loss": 0.017,
|
| 1788 |
"step": 6000
|
| 1789 |
},
|
| 1790 |
{
|
| 1791 |
+
"epoch": 3.111,
|
| 1792 |
+
"eval_loss": 0.17044013738632202,
|
| 1793 |
+
"eval_runtime": 2431.1861,
|
| 1794 |
+
"eval_samples_per_second": 5.606,
|
| 1795 |
+
"eval_steps_per_second": 0.701,
|
| 1796 |
+
"eval_wer": 9.097988200373777,
|
| 1797 |
"step": 6000
|
| 1798 |
},
|
| 1799 |
{
|
| 1800 |
+
"epoch": 3.114125,
|
| 1801 |
+
"grad_norm": 3.372814893722534,
|
| 1802 |
+
"learning_rate": 1.6483333333333334e-06,
|
| 1803 |
+
"loss": 0.0216,
|
| 1804 |
"step": 6025
|
| 1805 |
},
|
| 1806 |
{
|
| 1807 |
+
"epoch": 3.11725,
|
| 1808 |
+
"grad_norm": 0.39716529846191406,
|
| 1809 |
+
"learning_rate": 1.6275000000000003e-06,
|
| 1810 |
+
"loss": 0.0161,
|
| 1811 |
"step": 6050
|
| 1812 |
},
|
| 1813 |
{
|
| 1814 |
+
"epoch": 3.120375,
|
| 1815 |
+
"grad_norm": 1.4220778942108154,
|
| 1816 |
+
"learning_rate": 1.6066666666666668e-06,
|
| 1817 |
+
"loss": 0.0211,
|
| 1818 |
"step": 6075
|
| 1819 |
},
|
| 1820 |
{
|
| 1821 |
+
"epoch": 3.1235,
|
| 1822 |
+
"grad_norm": 1.2959835529327393,
|
| 1823 |
+
"learning_rate": 1.5858333333333333e-06,
|
| 1824 |
+
"loss": 0.0187,
|
| 1825 |
"step": 6100
|
| 1826 |
},
|
| 1827 |
{
|
| 1828 |
+
"epoch": 3.1266249999999998,
|
| 1829 |
+
"grad_norm": 1.9399360418319702,
|
| 1830 |
+
"learning_rate": 1.565e-06,
|
| 1831 |
+
"loss": 0.0123,
|
| 1832 |
"step": 6125
|
| 1833 |
},
|
| 1834 |
{
|
| 1835 |
+
"epoch": 3.12975,
|
| 1836 |
+
"grad_norm": 2.898085594177246,
|
| 1837 |
+
"learning_rate": 1.5441666666666668e-06,
|
| 1838 |
+
"loss": 0.0197,
|
| 1839 |
"step": 6150
|
| 1840 |
},
|
| 1841 |
{
|
| 1842 |
+
"epoch": 3.132875,
|
| 1843 |
+
"grad_norm": 3.642364263534546,
|
| 1844 |
+
"learning_rate": 1.5233333333333333e-06,
|
| 1845 |
+
"loss": 0.0162,
|
| 1846 |
"step": 6175
|
| 1847 |
},
|
| 1848 |
{
|
| 1849 |
+
"epoch": 3.136,
|
| 1850 |
+
"grad_norm": 1.7888002395629883,
|
| 1851 |
+
"learning_rate": 1.5025e-06,
|
| 1852 |
+
"loss": 0.0164,
|
| 1853 |
"step": 6200
|
| 1854 |
},
|
| 1855 |
{
|
| 1856 |
+
"epoch": 3.139125,
|
| 1857 |
+
"grad_norm": 2.3871753215789795,
|
| 1858 |
+
"learning_rate": 1.4816666666666667e-06,
|
| 1859 |
+
"loss": 0.0175,
|
| 1860 |
"step": 6225
|
| 1861 |
},
|
| 1862 |
{
|
| 1863 |
+
"epoch": 3.1422499999999998,
|
| 1864 |
+
"grad_norm": 1.4263437986373901,
|
| 1865 |
+
"learning_rate": 1.4608333333333333e-06,
|
| 1866 |
+
"loss": 0.0152,
|
| 1867 |
"step": 6250
|
| 1868 |
},
|
| 1869 |
{
|
| 1870 |
+
"epoch": 3.145375,
|
| 1871 |
+
"grad_norm": 2.174880027770996,
|
| 1872 |
+
"learning_rate": 1.44e-06,
|
| 1873 |
+
"loss": 0.0172,
|
| 1874 |
"step": 6275
|
| 1875 |
},
|
| 1876 |
{
|
| 1877 |
+
"epoch": 3.1485,
|
| 1878 |
+
"grad_norm": 0.9558664560317993,
|
| 1879 |
+
"learning_rate": 1.4191666666666667e-06,
|
| 1880 |
+
"loss": 0.0172,
|
| 1881 |
"step": 6300
|
| 1882 |
},
|
| 1883 |
{
|
| 1884 |
+
"epoch": 3.151625,
|
| 1885 |
+
"grad_norm": 2.228374719619751,
|
| 1886 |
+
"learning_rate": 1.3983333333333334e-06,
|
| 1887 |
+
"loss": 0.0173,
|
| 1888 |
"step": 6325
|
| 1889 |
},
|
| 1890 |
{
|
| 1891 |
+
"epoch": 3.15475,
|
| 1892 |
+
"grad_norm": 1.3473858833312988,
|
| 1893 |
+
"learning_rate": 1.3775000000000002e-06,
|
| 1894 |
+
"loss": 0.0211,
|
| 1895 |
"step": 6350
|
| 1896 |
},
|
| 1897 |
{
|
| 1898 |
+
"epoch": 3.1578749999999998,
|
| 1899 |
+
"grad_norm": 1.575217843055725,
|
| 1900 |
+
"learning_rate": 1.3566666666666667e-06,
|
| 1901 |
+
"loss": 0.0195,
|
| 1902 |
"step": 6375
|
| 1903 |
},
|
| 1904 |
{
|
| 1905 |
+
"epoch": 3.161,
|
| 1906 |
+
"grad_norm": 1.4997564554214478,
|
| 1907 |
+
"learning_rate": 1.3358333333333334e-06,
|
| 1908 |
+
"loss": 0.0174,
|
| 1909 |
"step": 6400
|
| 1910 |
},
|
| 1911 |
{
|
| 1912 |
+
"epoch": 3.164125,
|
| 1913 |
+
"grad_norm": 1.6651334762573242,
|
| 1914 |
+
"learning_rate": 1.3150000000000001e-06,
|
| 1915 |
+
"loss": 0.0178,
|
| 1916 |
"step": 6425
|
| 1917 |
},
|
| 1918 |
{
|
| 1919 |
+
"epoch": 3.16725,
|
| 1920 |
+
"grad_norm": 1.9218391180038452,
|
| 1921 |
+
"learning_rate": 1.2941666666666668e-06,
|
| 1922 |
+
"loss": 0.0168,
|
| 1923 |
"step": 6450
|
| 1924 |
},
|
| 1925 |
{
|
| 1926 |
+
"epoch": 3.170375,
|
| 1927 |
+
"grad_norm": 1.322709560394287,
|
| 1928 |
+
"learning_rate": 1.2733333333333334e-06,
|
| 1929 |
+
"loss": 0.0155,
|
| 1930 |
"step": 6475
|
| 1931 |
},
|
| 1932 |
{
|
| 1933 |
+
"epoch": 3.1734999999999998,
|
| 1934 |
+
"grad_norm": 0.7265995144844055,
|
| 1935 |
+
"learning_rate": 1.2525e-06,
|
| 1936 |
+
"loss": 0.0174,
|
| 1937 |
"step": 6500
|
| 1938 |
},
|
| 1939 |
{
|
| 1940 |
+
"epoch": 3.1734999999999998,
|
| 1941 |
+
"eval_loss": 0.16809040307998657,
|
| 1942 |
+
"eval_runtime": 2425.0716,
|
| 1943 |
+
"eval_samples_per_second": 5.62,
|
| 1944 |
+
"eval_steps_per_second": 0.703,
|
| 1945 |
+
"eval_wer": 9.072336839019385,
|
| 1946 |
"step": 6500
|
| 1947 |
},
|
| 1948 |
{
|
| 1949 |
+
"epoch": 3.176625,
|
| 1950 |
+
"grad_norm": 3.6835756301879883,
|
| 1951 |
+
"learning_rate": 1.2316666666666668e-06,
|
| 1952 |
+
"loss": 0.0177,
|
| 1953 |
"step": 6525
|
| 1954 |
},
|
| 1955 |
{
|
| 1956 |
+
"epoch": 3.17975,
|
| 1957 |
+
"grad_norm": 1.7619342803955078,
|
| 1958 |
+
"learning_rate": 1.2108333333333335e-06,
|
| 1959 |
+
"loss": 0.0212,
|
| 1960 |
"step": 6550
|
| 1961 |
},
|
| 1962 |
{
|
| 1963 |
+
"epoch": 3.182875,
|
| 1964 |
+
"grad_norm": 1.547814130783081,
|
| 1965 |
+
"learning_rate": 1.19e-06,
|
| 1966 |
+
"loss": 0.0213,
|
| 1967 |
"step": 6575
|
| 1968 |
},
|
| 1969 |
{
|
| 1970 |
+
"epoch": 3.186,
|
| 1971 |
+
"grad_norm": 2.0077638626098633,
|
| 1972 |
+
"learning_rate": 1.1691666666666668e-06,
|
| 1973 |
+
"loss": 0.016,
|
| 1974 |
"step": 6600
|
| 1975 |
},
|
| 1976 |
{
|
| 1977 |
+
"epoch": 3.1891249999999998,
|
| 1978 |
+
"grad_norm": 1.806665062904358,
|
| 1979 |
+
"learning_rate": 1.1483333333333333e-06,
|
| 1980 |
+
"loss": 0.0165,
|
| 1981 |
"step": 6625
|
| 1982 |
},
|
| 1983 |
{
|
| 1984 |
+
"epoch": 3.19225,
|
| 1985 |
+
"grad_norm": 1.9576735496520996,
|
| 1986 |
+
"learning_rate": 1.1283333333333333e-06,
|
| 1987 |
+
"loss": 0.0173,
|
| 1988 |
"step": 6650
|
| 1989 |
},
|
| 1990 |
{
|
| 1991 |
+
"epoch": 3.195375,
|
| 1992 |
+
"grad_norm": 1.992625117301941,
|
| 1993 |
+
"learning_rate": 1.1075e-06,
|
| 1994 |
+
"loss": 0.0179,
|
| 1995 |
"step": 6675
|
| 1996 |
},
|
| 1997 |
{
|
| 1998 |
+
"epoch": 3.1985,
|
| 1999 |
+
"grad_norm": 1.9625897407531738,
|
| 2000 |
+
"learning_rate": 1.0866666666666667e-06,
|
| 2001 |
+
"loss": 0.0174,
|
| 2002 |
"step": 6700
|
| 2003 |
},
|
| 2004 |
{
|
| 2005 |
+
"epoch": 3.201625,
|
| 2006 |
+
"grad_norm": 1.079950213432312,
|
| 2007 |
+
"learning_rate": 1.0658333333333334e-06,
|
| 2008 |
+
"loss": 0.0189,
|
| 2009 |
"step": 6725
|
| 2010 |
},
|
| 2011 |
{
|
| 2012 |
+
"epoch": 3.2047499999999998,
|
| 2013 |
+
"grad_norm": 1.7856156826019287,
|
| 2014 |
+
"learning_rate": 1.045e-06,
|
| 2015 |
+
"loss": 0.0179,
|
| 2016 |
"step": 6750
|
| 2017 |
},
|
| 2018 |
{
|
| 2019 |
+
"epoch": 3.207875,
|
| 2020 |
+
"grad_norm": 0.44674980640411377,
|
| 2021 |
+
"learning_rate": 1.0241666666666667e-06,
|
| 2022 |
+
"loss": 0.0185,
|
| 2023 |
"step": 6775
|
| 2024 |
},
|
| 2025 |
{
|
| 2026 |
+
"epoch": 3.211,
|
| 2027 |
+
"grad_norm": 2.1849780082702637,
|
| 2028 |
+
"learning_rate": 1.0033333333333334e-06,
|
| 2029 |
+
"loss": 0.0208,
|
| 2030 |
"step": 6800
|
| 2031 |
},
|
| 2032 |
{
|
| 2033 |
+
"epoch": 4.001125,
|
| 2034 |
+
"grad_norm": 1.4259858131408691,
|
| 2035 |
+
"learning_rate": 9.825000000000001e-07,
|
| 2036 |
+
"loss": 0.0162,
|
| 2037 |
"step": 6825
|
| 2038 |
},
|
| 2039 |
{
|
| 2040 |
+
"epoch": 4.00425,
|
| 2041 |
+
"grad_norm": 0.5507416129112244,
|
| 2042 |
+
"learning_rate": 9.616666666666668e-07,
|
| 2043 |
+
"loss": 0.0099,
|
| 2044 |
"step": 6850
|
| 2045 |
},
|
| 2046 |
{
|
| 2047 |
+
"epoch": 4.007375,
|
| 2048 |
+
"grad_norm": 1.9662590026855469,
|
| 2049 |
+
"learning_rate": 9.408333333333334e-07,
|
| 2050 |
+
"loss": 0.0117,
|
| 2051 |
"step": 6875
|
| 2052 |
},
|
| 2053 |
{
|
| 2054 |
+
"epoch": 4.0105,
|
| 2055 |
+
"grad_norm": 1.9439060688018799,
|
| 2056 |
+
"learning_rate": 9.2e-07,
|
| 2057 |
+
"loss": 0.0114,
|
| 2058 |
"step": 6900
|
| 2059 |
},
|
| 2060 |
{
|
| 2061 |
+
"epoch": 4.013625,
|
| 2062 |
+
"grad_norm": 1.266473412513733,
|
| 2063 |
+
"learning_rate": 8.991666666666667e-07,
|
| 2064 |
+
"loss": 0.0085,
|
| 2065 |
"step": 6925
|
| 2066 |
},
|
| 2067 |
{
|
| 2068 |
+
"epoch": 4.01675,
|
| 2069 |
+
"grad_norm": 1.4318881034851074,
|
| 2070 |
+
"learning_rate": 8.783333333333334e-07,
|
| 2071 |
+
"loss": 0.0095,
|
| 2072 |
"step": 6950
|
| 2073 |
},
|
| 2074 |
{
|
| 2075 |
+
"epoch": 4.019875,
|
| 2076 |
+
"grad_norm": 1.7301366329193115,
|
| 2077 |
+
"learning_rate": 8.574999999999999e-07,
|
| 2078 |
+
"loss": 0.01,
|
| 2079 |
"step": 6975
|
| 2080 |
},
|
| 2081 |
{
|
| 2082 |
+
"epoch": 4.023,
|
| 2083 |
+
"grad_norm": 0.7250347137451172,
|
| 2084 |
+
"learning_rate": 8.366666666666667e-07,
|
| 2085 |
+
"loss": 0.0098,
|
| 2086 |
"step": 7000
|
| 2087 |
},
|
| 2088 |
{
|
| 2089 |
+
"epoch": 4.023,
|
| 2090 |
+
"eval_loss": 0.17248913645744324,
|
| 2091 |
+
"eval_runtime": 2424.7286,
|
| 2092 |
+
"eval_samples_per_second": 5.621,
|
| 2093 |
+
"eval_steps_per_second": 0.703,
|
| 2094 |
+
"eval_wer": 8.862545347942394,
|
| 2095 |
"step": 7000
|
| 2096 |
},
|
| 2097 |
{
|
| 2098 |
+
"epoch": 4.026125,
|
| 2099 |
+
"grad_norm": 0.9742090702056885,
|
| 2100 |
+
"learning_rate": 8.158333333333334e-07,
|
| 2101 |
+
"loss": 0.0083,
|
| 2102 |
"step": 7025
|
| 2103 |
},
|
| 2104 |
{
|
| 2105 |
+
"epoch": 4.02925,
|
| 2106 |
+
"grad_norm": 1.966017723083496,
|
| 2107 |
+
"learning_rate": 7.950000000000001e-07,
|
| 2108 |
+
"loss": 0.0098,
|
| 2109 |
"step": 7050
|
| 2110 |
},
|
| 2111 |
{
|
| 2112 |
+
"epoch": 4.032375,
|
| 2113 |
+
"grad_norm": 0.3238462209701538,
|
| 2114 |
+
"learning_rate": 7.741666666666667e-07,
|
| 2115 |
+
"loss": 0.0106,
|
| 2116 |
"step": 7075
|
| 2117 |
},
|
| 2118 |
{
|
| 2119 |
+
"epoch": 4.0355,
|
| 2120 |
+
"grad_norm": 1.1980866193771362,
|
| 2121 |
+
"learning_rate": 7.533333333333334e-07,
|
| 2122 |
+
"loss": 0.0095,
|
| 2123 |
"step": 7100
|
| 2124 |
},
|
| 2125 |
{
|
| 2126 |
+
"epoch": 4.038625,
|
| 2127 |
+
"grad_norm": 1.5639567375183105,
|
| 2128 |
+
"learning_rate": 7.325e-07,
|
| 2129 |
+
"loss": 0.0079,
|
| 2130 |
"step": 7125
|
| 2131 |
},
|
| 2132 |
{
|
| 2133 |
+
"epoch": 4.04175,
|
| 2134 |
+
"grad_norm": 2.4864511489868164,
|
| 2135 |
+
"learning_rate": 7.116666666666667e-07,
|
| 2136 |
+
"loss": 0.0081,
|
| 2137 |
"step": 7150
|
| 2138 |
},
|
| 2139 |
{
|
| 2140 |
+
"epoch": 4.044875,
|
| 2141 |
+
"grad_norm": 0.3260515332221985,
|
| 2142 |
+
"learning_rate": 6.908333333333333e-07,
|
| 2143 |
+
"loss": 0.0078,
|
| 2144 |
"step": 7175
|
| 2145 |
},
|
| 2146 |
{
|
| 2147 |
+
"epoch": 4.048,
|
| 2148 |
+
"grad_norm": 1.6305910348892212,
|
| 2149 |
+
"learning_rate": 6.7e-07,
|
| 2150 |
+
"loss": 0.0093,
|
| 2151 |
"step": 7200
|
| 2152 |
},
|
| 2153 |
{
|
| 2154 |
+
"epoch": 4.051125,
|
| 2155 |
+
"grad_norm": 0.6141969561576843,
|
| 2156 |
+
"learning_rate": 6.491666666666667e-07,
|
| 2157 |
+
"loss": 0.0082,
|
| 2158 |
"step": 7225
|
| 2159 |
},
|
| 2160 |
{
|
| 2161 |
+
"epoch": 4.05425,
|
| 2162 |
+
"grad_norm": 1.9699019193649292,
|
| 2163 |
+
"learning_rate": 6.283333333333334e-07,
|
| 2164 |
+
"loss": 0.0089,
|
| 2165 |
"step": 7250
|
| 2166 |
},
|
| 2167 |
{
|
| 2168 |
+
"epoch": 4.057375,
|
| 2169 |
+
"grad_norm": 0.2788224518299103,
|
| 2170 |
+
"learning_rate": 6.075e-07,
|
| 2171 |
+
"loss": 0.0053,
|
| 2172 |
"step": 7275
|
| 2173 |
},
|
| 2174 |
{
|
| 2175 |
+
"epoch": 4.0605,
|
| 2176 |
+
"grad_norm": 1.0774401426315308,
|
| 2177 |
+
"learning_rate": 5.866666666666667e-07,
|
| 2178 |
+
"loss": 0.0074,
|
| 2179 |
"step": 7300
|
| 2180 |
},
|
| 2181 |
{
|
| 2182 |
+
"epoch": 4.063625,
|
| 2183 |
+
"grad_norm": 0.2802826166152954,
|
| 2184 |
+
"learning_rate": 5.658333333333334e-07,
|
| 2185 |
+
"loss": 0.007,
|
| 2186 |
"step": 7325
|
| 2187 |
},
|
| 2188 |
{
|
| 2189 |
+
"epoch": 4.06675,
|
| 2190 |
+
"grad_norm": 0.459989458322525,
|
| 2191 |
+
"learning_rate": 5.450000000000001e-07,
|
| 2192 |
+
"loss": 0.0061,
|
| 2193 |
"step": 7350
|
| 2194 |
},
|
| 2195 |
{
|
| 2196 |
+
"epoch": 4.069875,
|
| 2197 |
+
"grad_norm": 1.6869885921478271,
|
| 2198 |
+
"learning_rate": 5.241666666666667e-07,
|
| 2199 |
+
"loss": 0.007,
|
| 2200 |
"step": 7375
|
| 2201 |
},
|
| 2202 |
{
|
| 2203 |
+
"epoch": 4.073,
|
| 2204 |
+
"grad_norm": 1.150252103805542,
|
| 2205 |
+
"learning_rate": 5.033333333333333e-07,
|
| 2206 |
+
"loss": 0.009,
|
| 2207 |
"step": 7400
|
| 2208 |
},
|
| 2209 |
{
|
| 2210 |
+
"epoch": 4.076125,
|
| 2211 |
+
"grad_norm": 0.7096041440963745,
|
| 2212 |
+
"learning_rate": 4.825e-07,
|
| 2213 |
+
"loss": 0.0074,
|
| 2214 |
"step": 7425
|
| 2215 |
},
|
| 2216 |
{
|
| 2217 |
+
"epoch": 4.07925,
|
| 2218 |
+
"grad_norm": 1.9114930629730225,
|
| 2219 |
+
"learning_rate": 4.6166666666666666e-07,
|
| 2220 |
+
"loss": 0.0081,
|
| 2221 |
"step": 7450
|
| 2222 |
},
|
| 2223 |
{
|
| 2224 |
+
"epoch": 4.082375,
|
| 2225 |
+
"grad_norm": 0.41340845823287964,
|
| 2226 |
+
"learning_rate": 4.408333333333334e-07,
|
| 2227 |
+
"loss": 0.0078,
|
| 2228 |
"step": 7475
|
| 2229 |
},
|
| 2230 |
{
|
| 2231 |
+
"epoch": 4.0855,
|
| 2232 |
+
"grad_norm": 1.241665005683899,
|
| 2233 |
+
"learning_rate": 4.2e-07,
|
| 2234 |
+
"loss": 0.0076,
|
| 2235 |
"step": 7500
|
| 2236 |
},
|
| 2237 |
{
|
| 2238 |
+
"epoch": 4.0855,
|
| 2239 |
+
"eval_loss": 0.17651478946208954,
|
| 2240 |
+
"eval_runtime": 2432.1699,
|
| 2241 |
+
"eval_samples_per_second": 5.604,
|
| 2242 |
+
"eval_steps_per_second": 0.701,
|
| 2243 |
+
"eval_wer": 8.83506174649126,
|
| 2244 |
"step": 7500
|
| 2245 |
},
|
| 2246 |
{
|
| 2247 |
+
"epoch": 4.088625,
|
| 2248 |
+
"grad_norm": 0.6228423714637756,
|
| 2249 |
+
"learning_rate": 3.991666666666667e-07,
|
| 2250 |
+
"loss": 0.0075,
|
| 2251 |
"step": 7525
|
| 2252 |
},
|
| 2253 |
{
|
| 2254 |
+
"epoch": 4.09175,
|
| 2255 |
+
"grad_norm": 0.5548863410949707,
|
| 2256 |
+
"learning_rate": 3.7833333333333335e-07,
|
| 2257 |
+
"loss": 0.007,
|
| 2258 |
"step": 7550
|
| 2259 |
},
|
| 2260 |
{
|
| 2261 |
+
"epoch": 4.094875,
|
| 2262 |
+
"grad_norm": 0.33007875084877014,
|
| 2263 |
+
"learning_rate": 3.575e-07,
|
| 2264 |
+
"loss": 0.0058,
|
| 2265 |
"step": 7575
|
| 2266 |
},
|
| 2267 |
{
|
| 2268 |
+
"epoch": 4.098,
|
| 2269 |
+
"grad_norm": 1.3314837217330933,
|
| 2270 |
+
"learning_rate": 3.366666666666667e-07,
|
| 2271 |
+
"loss": 0.0068,
|
| 2272 |
"step": 7600
|
| 2273 |
},
|
| 2274 |
{
|
| 2275 |
+
"epoch": 4.101125,
|
| 2276 |
+
"grad_norm": 2.230480909347534,
|
| 2277 |
+
"learning_rate": 3.1583333333333336e-07,
|
| 2278 |
+
"loss": 0.007,
|
| 2279 |
"step": 7625
|
| 2280 |
},
|
| 2281 |
{
|
| 2282 |
+
"epoch": 4.10425,
|
| 2283 |
+
"grad_norm": 0.3246554136276245,
|
| 2284 |
+
"learning_rate": 2.95e-07,
|
| 2285 |
+
"loss": 0.0058,
|
| 2286 |
"step": 7650
|
| 2287 |
},
|
| 2288 |
{
|
| 2289 |
+
"epoch": 4.107375,
|
| 2290 |
+
"grad_norm": 3.8842251300811768,
|
| 2291 |
+
"learning_rate": 2.7416666666666665e-07,
|
| 2292 |
+
"loss": 0.0071,
|
| 2293 |
"step": 7675
|
| 2294 |
},
|
| 2295 |
{
|
| 2296 |
+
"epoch": 4.1105,
|
| 2297 |
+
"grad_norm": 0.27049753069877625,
|
| 2298 |
+
"learning_rate": 2.533333333333333e-07,
|
| 2299 |
+
"loss": 0.0068,
|
| 2300 |
"step": 7700
|
| 2301 |
},
|
| 2302 |
{
|
| 2303 |
+
"epoch": 4.113625,
|
| 2304 |
+
"grad_norm": 1.5919678211212158,
|
| 2305 |
+
"learning_rate": 2.325e-07,
|
| 2306 |
+
"loss": 0.009,
|
| 2307 |
"step": 7725
|
| 2308 |
},
|
| 2309 |
{
|
| 2310 |
+
"epoch": 4.11675,
|
| 2311 |
+
"grad_norm": 1.0917800664901733,
|
| 2312 |
+
"learning_rate": 2.116666666666667e-07,
|
| 2313 |
+
"loss": 0.0083,
|
| 2314 |
"step": 7750
|
| 2315 |
},
|
| 2316 |
{
|
| 2317 |
+
"epoch": 4.119875,
|
| 2318 |
+
"grad_norm": 1.6236002445220947,
|
| 2319 |
+
"learning_rate": 1.9083333333333334e-07,
|
| 2320 |
+
"loss": 0.0077,
|
| 2321 |
"step": 7775
|
| 2322 |
},
|
| 2323 |
{
|
| 2324 |
+
"epoch": 4.123,
|
| 2325 |
+
"grad_norm": 0.34154245257377625,
|
| 2326 |
+
"learning_rate": 1.7e-07,
|
| 2327 |
+
"loss": 0.0064,
|
| 2328 |
"step": 7800
|
| 2329 |
},
|
| 2330 |
{
|
| 2331 |
+
"epoch": 4.126125,
|
| 2332 |
+
"grad_norm": 0.7562180757522583,
|
| 2333 |
+
"learning_rate": 1.4916666666666669e-07,
|
| 2334 |
+
"loss": 0.0073,
|
| 2335 |
"step": 7825
|
| 2336 |
},
|
| 2337 |
{
|
| 2338 |
+
"epoch": 4.12925,
|
| 2339 |
+
"grad_norm": 1.4198355674743652,
|
| 2340 |
+
"learning_rate": 1.2833333333333333e-07,
|
| 2341 |
+
"loss": 0.0051,
|
| 2342 |
"step": 7850
|
| 2343 |
},
|
| 2344 |
{
|
| 2345 |
+
"epoch": 4.132375,
|
| 2346 |
+
"grad_norm": 0.38768094778060913,
|
| 2347 |
+
"learning_rate": 1.075e-07,
|
| 2348 |
+
"loss": 0.0073,
|
| 2349 |
"step": 7875
|
| 2350 |
},
|
| 2351 |
{
|
| 2352 |
+
"epoch": 4.1355,
|
| 2353 |
+
"grad_norm": 2.1387102603912354,
|
| 2354 |
+
"learning_rate": 8.666666666666666e-08,
|
| 2355 |
+
"loss": 0.0078,
|
| 2356 |
"step": 7900
|
| 2357 |
},
|
| 2358 |
{
|
| 2359 |
+
"epoch": 4.138625,
|
| 2360 |
+
"grad_norm": 0.7683693170547485,
|
| 2361 |
+
"learning_rate": 6.583333333333333e-08,
|
| 2362 |
+
"loss": 0.006,
|
| 2363 |
"step": 7925
|
| 2364 |
},
|
| 2365 |
{
|
| 2366 |
+
"epoch": 4.14175,
|
| 2367 |
+
"grad_norm": 0.40976452827453613,
|
| 2368 |
+
"learning_rate": 4.5e-08,
|
| 2369 |
+
"loss": 0.0059,
|
| 2370 |
"step": 7950
|
| 2371 |
},
|
| 2372 |
{
|
| 2373 |
+
"epoch": 4.144875,
|
| 2374 |
+
"grad_norm": 0.3620973229408264,
|
| 2375 |
+
"learning_rate": 2.4166666666666668e-08,
|
| 2376 |
+
"loss": 0.0054,
|
| 2377 |
"step": 7975
|
| 2378 |
},
|
| 2379 |
{
|
| 2380 |
+
"epoch": 4.148,
|
| 2381 |
+
"grad_norm": 0.28266406059265137,
|
| 2382 |
+
"learning_rate": 3.333333333333334e-09,
|
| 2383 |
+
"loss": 0.007,
|
| 2384 |
"step": 8000
|
| 2385 |
},
|
| 2386 |
{
|
| 2387 |
+
"epoch": 4.148,
|
| 2388 |
+
"eval_loss": 0.17873740196228027,
|
| 2389 |
+
"eval_runtime": 2435.5612,
|
| 2390 |
+
"eval_samples_per_second": 5.596,
|
| 2391 |
+
"eval_steps_per_second": 0.7,
|
| 2392 |
+
"eval_wer": 8.8020814247499,
|
| 2393 |
"step": 8000
|
| 2394 |
},
|
| 2395 |
{
|
| 2396 |
+
"epoch": 4.148,
|
| 2397 |
"step": 8000,
|
| 2398 |
+
"total_flos": 1.3062109848403968e+20,
|
| 2399 |
+
"train_loss": 0.11351343880966305,
|
| 2400 |
+
"train_runtime": 52381.6915,
|
| 2401 |
+
"train_samples_per_second": 2.444,
|
| 2402 |
+
"train_steps_per_second": 0.153
|
| 2403 |
}
|
| 2404 |
],
|
| 2405 |
+
"logging_steps": 25,
|
| 2406 |
"max_steps": 8000,
|
| 2407 |
+
"num_input_tokens_seen": 0,
|
| 2408 |
"num_train_epochs": 9223372036854775807,
|
| 2409 |
+
"save_steps": 1000,
|
| 2410 |
+
"stateful_callbacks": {
|
| 2411 |
+
"TrainerControl": {
|
| 2412 |
+
"args": {
|
| 2413 |
+
"should_epoch_stop": false,
|
| 2414 |
+
"should_evaluate": false,
|
| 2415 |
+
"should_log": false,
|
| 2416 |
+
"should_save": true,
|
| 2417 |
+
"should_training_stop": true
|
| 2418 |
+
},
|
| 2419 |
+
"attributes": {}
|
| 2420 |
+
}
|
| 2421 |
+
},
|
| 2422 |
+
"total_flos": 1.3062109848403968e+20,
|
| 2423 |
+
"train_batch_size": 16,
|
| 2424 |
"trial_name": null,
|
| 2425 |
"trial_params": null
|
| 2426 |
}
|