End of training
Browse files- README.md +1 -1
- all_results.json +9 -9
- eval_results.json +5 -5
- runs/Nov01_14-36-01_5cb5e668fe5d/events.out.tfevents.1730551526.5cb5e668fe5d.185.1 +3 -0
- train_results.json +4 -4
- trainer_state.json +629 -629
README.md
CHANGED
|
@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
# whisper-large-v3-ft-cv-cy-en
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.2744
|
| 22 |
- Wer: 0.1474
|
|
|
|
| 16 |
|
| 17 |
# whisper-large-v3-ft-cv-cy-en
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/commonvoice_18_0_cy_en train main dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.2744
|
| 22 |
- Wer: 0.1474
|
all_results.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.537318712415989,
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 10763,
|
| 6 |
-
"eval_samples_per_second": 2.
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
"total_flos": 5.435079965953229e+20,
|
| 10 |
-
"train_loss": 0.
|
| 11 |
-
"train_runtime":
|
| 12 |
"train_samples": 45223,
|
| 13 |
-
"train_samples_per_second": 2.
|
| 14 |
-
"train_steps_per_second": 0.
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.537318712415989,
|
| 3 |
+
"eval_loss": 0.27435705065727234,
|
| 4 |
+
"eval_runtime": 4616.5839,
|
| 5 |
"eval_samples": 10763,
|
| 6 |
+
"eval_samples_per_second": 2.331,
|
| 7 |
+
"eval_steps_per_second": 0.146,
|
| 8 |
+
"eval_wer": 0.14744191504341703,
|
| 9 |
"total_flos": 5.435079965953229e+20,
|
| 10 |
+
"train_loss": 0.26693406739234926,
|
| 11 |
+
"train_runtime": 74036.7422,
|
| 12 |
"train_samples": 45223,
|
| 13 |
+
"train_samples_per_second": 2.161,
|
| 14 |
+
"train_steps_per_second": 0.068
|
| 15 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.537318712415989,
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 10763,
|
| 6 |
-
"eval_samples_per_second": 2.
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.537318712415989,
|
| 3 |
+
"eval_loss": 0.27435705065727234,
|
| 4 |
+
"eval_runtime": 4616.5839,
|
| 5 |
"eval_samples": 10763,
|
| 6 |
+
"eval_samples_per_second": 2.331,
|
| 7 |
+
"eval_steps_per_second": 0.146,
|
| 8 |
+
"eval_wer": 0.14744191504341703
|
| 9 |
}
|
runs/Nov01_14-36-01_5cb5e668fe5d/events.out.tfevents.1730551526.5cb5e668fe5d.185.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e9f9015a40f28192d9d7657905e0cd4758baeae33e5bd9b84fc2942db5d4eadb
|
| 3 |
+
size 406
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.537318712415989,
|
| 3 |
"total_flos": 5.435079965953229e+20,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 45223,
|
| 7 |
-
"train_samples_per_second": 2.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.537318712415989,
|
| 3 |
"total_flos": 5.435079965953229e+20,
|
| 4 |
+
"train_loss": 0.26693406739234926,
|
| 5 |
+
"train_runtime": 74036.7422,
|
| 6 |
"train_samples": 45223,
|
| 7 |
+
"train_samples_per_second": 2.161,
|
| 8 |
+
"train_steps_per_second": 0.068
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -10,1457 +10,1457 @@
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.017686593562079942,
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"loss":
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.035373187124159884,
|
| 20 |
-
"grad_norm":
|
| 21 |
-
"learning_rate":
|
| 22 |
-
"loss":
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.05305978068623983,
|
| 27 |
-
"grad_norm":
|
| 28 |
-
"learning_rate": 1.
|
| 29 |
-
"loss": 0.
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.07074637424831977,
|
| 34 |
-
"grad_norm":
|
| 35 |
-
"learning_rate":
|
| 36 |
-
"loss": 0.
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.08843296781039972,
|
| 41 |
-
"grad_norm":
|
| 42 |
-
"learning_rate": 2.
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.10611956137247966,
|
| 48 |
-
"grad_norm":
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.1238061549345596,
|
| 55 |
-
"grad_norm":
|
| 56 |
-
"learning_rate": 3.
|
| 57 |
-
"loss": 0.
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.14149274849663954,
|
| 62 |
-
"grad_norm":
|
| 63 |
-
"learning_rate":
|
| 64 |
-
"loss": 0.
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.1591793420587195,
|
| 69 |
-
"grad_norm":
|
| 70 |
-
"learning_rate": 4.
|
| 71 |
-
"loss": 0.
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.17686593562079944,
|
| 76 |
-
"grad_norm":
|
| 77 |
-
"learning_rate":
|
| 78 |
-
"loss": 0.
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.19455252918287938,
|
| 83 |
-
"grad_norm":
|
| 84 |
-
"learning_rate": 5.
|
| 85 |
-
"loss": 0.
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.21223912274495932,
|
| 90 |
-
"grad_norm":
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss": 0.
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.22992571630703926,
|
| 97 |
-
"grad_norm":
|
| 98 |
-
"learning_rate": 6.
|
| 99 |
-
"loss": 0.
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.2476123098691192,
|
| 104 |
-
"grad_norm":
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.26529890343119916,
|
| 111 |
-
"grad_norm":
|
| 112 |
-
"learning_rate": 7.
|
| 113 |
-
"loss": 0.
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.2829854969932791,
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"loss": 0.
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"epoch": 0.30067209055535904,
|
| 125 |
-
"grad_norm":
|
| 126 |
-
"learning_rate": 8.
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.318358684117439,
|
| 132 |
-
"grad_norm":
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.3360452776795189,
|
| 139 |
-
"grad_norm":
|
| 140 |
-
"learning_rate": 9.
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 0.3537318712415989,
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.3714184648036788,
|
| 153 |
-
"grad_norm":
|
| 154 |
-
"learning_rate": 9.
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.38910505836575876,
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate": 9.
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.40679165192783867,
|
| 167 |
-
"grad_norm":
|
| 168 |
-
"learning_rate": 9.
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.42447824548991864,
|
| 174 |
-
"grad_norm":
|
| 175 |
-
"learning_rate": 9.
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.4421648390519986,
|
| 181 |
-
"grad_norm":
|
| 182 |
-
"learning_rate": 9.
|
| 183 |
-
"loss": 0.
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.4598514326140785,
|
| 188 |
-
"grad_norm":
|
| 189 |
-
"learning_rate": 9.
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 0.4775380261761585,
|
| 195 |
-
"grad_norm":
|
| 196 |
-
"learning_rate": 9.
|
| 197 |
-
"loss": 0.
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
"epoch": 0.4952246197382384,
|
| 202 |
-
"grad_norm":
|
| 203 |
-
"learning_rate": 9.
|
| 204 |
-
"loss": 0.
|
| 205 |
"step": 700
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.5129112133003184,
|
| 209 |
-
"grad_norm":
|
| 210 |
-
"learning_rate": 9.
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 725
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.5305978068623983,
|
| 216 |
-
"grad_norm":
|
| 217 |
-
"learning_rate": 9.
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 750
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.5482844004244782,
|
| 223 |
-
"grad_norm":
|
| 224 |
-
"learning_rate": 9.
|
| 225 |
-
"loss": 0.
|
| 226 |
"step": 775
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.5659709939865581,
|
| 230 |
-
"grad_norm":
|
| 231 |
-
"learning_rate": 9.
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 800
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 0.5836575875486382,
|
| 237 |
-
"grad_norm":
|
| 238 |
-
"learning_rate": 9.
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 825
|
| 241 |
},
|
| 242 |
{
|
| 243 |
"epoch": 0.6013441811107181,
|
| 244 |
-
"grad_norm":
|
| 245 |
-
"learning_rate": 9.
|
| 246 |
-
"loss": 0.
|
| 247 |
"step": 850
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"epoch": 0.619030774672798,
|
| 251 |
-
"grad_norm":
|
| 252 |
-
"learning_rate": 9.
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 875
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.636717368234878,
|
| 258 |
-
"grad_norm":
|
| 259 |
-
"learning_rate": 9.
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 900
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.6544039617969579,
|
| 265 |
-
"grad_norm":
|
| 266 |
-
"learning_rate": 9.
|
| 267 |
-
"loss": 0.
|
| 268 |
"step": 925
|
| 269 |
},
|
| 270 |
{
|
| 271 |
"epoch": 0.6720905553590378,
|
| 272 |
-
"grad_norm":
|
| 273 |
-
"learning_rate":
|
| 274 |
-
"loss": 0.
|
| 275 |
"step": 950
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 0.6897771489211177,
|
| 279 |
-
"grad_norm":
|
| 280 |
-
"learning_rate": 8.
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 975
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 0.7074637424831978,
|
| 286 |
-
"grad_norm":
|
| 287 |
-
"learning_rate": 8.
|
| 288 |
-
"loss": 0.
|
| 289 |
"step": 1000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.7074637424831978,
|
| 293 |
-
"eval_loss": 0.
|
| 294 |
-
"eval_runtime":
|
| 295 |
-
"eval_samples_per_second": 2.
|
| 296 |
-
"eval_steps_per_second": 0.
|
| 297 |
-
"eval_wer": 0.
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.7251503360452777,
|
| 302 |
-
"grad_norm":
|
| 303 |
-
"learning_rate": 8.
|
| 304 |
-
"loss": 0.
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.7428369296073576,
|
| 309 |
-
"grad_norm":
|
| 310 |
-
"learning_rate": 8.
|
| 311 |
-
"loss": 0.
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 0.7605235231694376,
|
| 316 |
-
"grad_norm":
|
| 317 |
-
"learning_rate": 8.
|
| 318 |
-
"loss": 0.
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.7782101167315175,
|
| 323 |
-
"grad_norm":
|
| 324 |
-
"learning_rate": 8.
|
| 325 |
-
"loss": 0.
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.7958967102935974,
|
| 330 |
-
"grad_norm":
|
| 331 |
-
"learning_rate": 8.
|
| 332 |
-
"loss": 0.
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.8135833038556773,
|
| 337 |
-
"grad_norm":
|
| 338 |
-
"learning_rate": 8.
|
| 339 |
-
"loss": 0.
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.8312698974177574,
|
| 344 |
-
"grad_norm":
|
| 345 |
-
"learning_rate": 8.
|
| 346 |
-
"loss": 0.
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 0.8489564909798373,
|
| 351 |
-
"grad_norm":
|
| 352 |
-
"learning_rate": 8.
|
| 353 |
-
"loss": 0.
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.8666430845419172,
|
| 358 |
-
"grad_norm":
|
| 359 |
-
"learning_rate": 8.
|
| 360 |
-
"loss": 0.
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
"epoch": 0.8843296781039972,
|
| 365 |
-
"grad_norm":
|
| 366 |
-
"learning_rate": 8.
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.9020162716660771,
|
| 372 |
-
"grad_norm":
|
| 373 |
-
"learning_rate": 8.
|
| 374 |
-
"loss": 0.
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.919702865228157,
|
| 379 |
-
"grad_norm":
|
| 380 |
-
"learning_rate": 8.
|
| 381 |
-
"loss": 0.
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
"epoch": 0.937389458790237,
|
| 386 |
-
"grad_norm":
|
| 387 |
-
"learning_rate": 8.
|
| 388 |
-
"loss": 0.
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.955076052352317,
|
| 393 |
-
"grad_norm":
|
| 394 |
-
"learning_rate": 8.
|
| 395 |
-
"loss": 0.
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 0.9727626459143969,
|
| 400 |
-
"grad_norm":
|
| 401 |
-
"learning_rate": 8.
|
| 402 |
-
"loss": 0.
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
"epoch": 0.9904492394764768,
|
| 407 |
-
"grad_norm":
|
| 408 |
-
"learning_rate": 8.
|
| 409 |
-
"loss": 0.
|
| 410 |
"step": 1400
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 1.0081358330385568,
|
| 414 |
-
"grad_norm":
|
| 415 |
-
"learning_rate": 7.
|
| 416 |
-
"loss": 0.
|
| 417 |
"step": 1425
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 1.0258224266006368,
|
| 421 |
-
"grad_norm":
|
| 422 |
-
"learning_rate": 7.
|
| 423 |
-
"loss": 0.
|
| 424 |
"step": 1450
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 1.0435090201627166,
|
| 428 |
-
"grad_norm":
|
| 429 |
-
"learning_rate": 7.
|
| 430 |
-
"loss": 0.
|
| 431 |
"step": 1475
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 1.0611956137247966,
|
| 435 |
-
"grad_norm":
|
| 436 |
-
"learning_rate": 7.
|
| 437 |
-
"loss": 0.
|
| 438 |
"step": 1500
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 1.0788822072868764,
|
| 442 |
-
"grad_norm":
|
| 443 |
-
"learning_rate": 7.
|
| 444 |
-
"loss": 0.
|
| 445 |
"step": 1525
|
| 446 |
},
|
| 447 |
{
|
| 448 |
"epoch": 1.0965688008489565,
|
| 449 |
-
"grad_norm":
|
| 450 |
-
"learning_rate": 7.
|
| 451 |
-
"loss": 0.
|
| 452 |
"step": 1550
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 1.1142553944110365,
|
| 456 |
-
"grad_norm":
|
| 457 |
-
"learning_rate": 7.
|
| 458 |
-
"loss": 0.
|
| 459 |
"step": 1575
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 1.1319419879731163,
|
| 463 |
-
"grad_norm":
|
| 464 |
-
"learning_rate": 7.
|
| 465 |
-
"loss": 0.
|
| 466 |
"step": 1600
|
| 467 |
},
|
| 468 |
{
|
| 469 |
"epoch": 1.1496285815351963,
|
| 470 |
-
"grad_norm":
|
| 471 |
-
"learning_rate": 7.
|
| 472 |
-
"loss": 0.
|
| 473 |
"step": 1625
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 1.1673151750972763,
|
| 477 |
-
"grad_norm":
|
| 478 |
-
"learning_rate": 7.
|
| 479 |
-
"loss": 0.
|
| 480 |
"step": 1650
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 1.1850017686593561,
|
| 484 |
-
"grad_norm":
|
| 485 |
-
"learning_rate": 7.
|
| 486 |
-
"loss": 0.
|
| 487 |
"step": 1675
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 1.2026883622214362,
|
| 491 |
-
"grad_norm":
|
| 492 |
-
"learning_rate": 7.
|
| 493 |
-
"loss": 0.
|
| 494 |
"step": 1700
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 1.2203749557835162,
|
| 498 |
-
"grad_norm":
|
| 499 |
-
"learning_rate": 7.
|
| 500 |
-
"loss": 0.
|
| 501 |
"step": 1725
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 1.238061549345596,
|
| 505 |
-
"grad_norm":
|
| 506 |
-
"learning_rate": 7.
|
| 507 |
-
"loss": 0.
|
| 508 |
"step": 1750
|
| 509 |
},
|
| 510 |
{
|
| 511 |
"epoch": 1.255748142907676,
|
| 512 |
-
"grad_norm":
|
| 513 |
-
"learning_rate": 7.
|
| 514 |
-
"loss": 0.
|
| 515 |
"step": 1775
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 1.2734347364697558,
|
| 519 |
-
"grad_norm":
|
| 520 |
-
"learning_rate": 7.
|
| 521 |
-
"loss": 0.
|
| 522 |
"step": 1800
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 1.2911213300318358,
|
| 526 |
-
"grad_norm":
|
| 527 |
-
"learning_rate": 7.
|
| 528 |
-
"loss": 0.
|
| 529 |
"step": 1825
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 1.3088079235939158,
|
| 533 |
-
"grad_norm":
|
| 534 |
-
"learning_rate":
|
| 535 |
-
"loss": 0.
|
| 536 |
"step": 1850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 1.3264945171559956,
|
| 540 |
-
"grad_norm":
|
| 541 |
-
"learning_rate": 6.
|
| 542 |
-
"loss": 0.
|
| 543 |
"step": 1875
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 1.3441811107180757,
|
| 547 |
-
"grad_norm":
|
| 548 |
-
"learning_rate": 6.
|
| 549 |
-
"loss": 0.
|
| 550 |
"step": 1900
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 1.3618677042801557,
|
| 554 |
-
"grad_norm":
|
| 555 |
-
"learning_rate": 6.
|
| 556 |
-
"loss": 0.
|
| 557 |
"step": 1925
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 1.3795542978422355,
|
| 561 |
-
"grad_norm":
|
| 562 |
-
"learning_rate": 6.
|
| 563 |
-
"loss": 0.
|
| 564 |
"step": 1950
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 1.3972408914043155,
|
| 568 |
-
"grad_norm":
|
| 569 |
-
"learning_rate": 6.
|
| 570 |
-
"loss": 0.
|
| 571 |
"step": 1975
|
| 572 |
},
|
| 573 |
{
|
| 574 |
"epoch": 1.4149274849663955,
|
| 575 |
-
"grad_norm":
|
| 576 |
-
"learning_rate": 6.
|
| 577 |
-
"loss": 0.
|
| 578 |
"step": 2000
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 1.4149274849663955,
|
| 582 |
-
"eval_loss": 0.
|
| 583 |
-
"eval_runtime":
|
| 584 |
-
"eval_samples_per_second": 2.
|
| 585 |
-
"eval_steps_per_second": 0.
|
| 586 |
-
"eval_wer": 0.
|
| 587 |
"step": 2000
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 1.4326140785284753,
|
| 591 |
-
"grad_norm":
|
| 592 |
-
"learning_rate": 6.
|
| 593 |
-
"loss": 0.
|
| 594 |
"step": 2025
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 1.4503006720905554,
|
| 598 |
-
"grad_norm":
|
| 599 |
-
"learning_rate": 6.
|
| 600 |
-
"loss": 0.
|
| 601 |
"step": 2050
|
| 602 |
},
|
| 603 |
{
|
| 604 |
"epoch": 1.4679872656526354,
|
| 605 |
-
"grad_norm":
|
| 606 |
-
"learning_rate": 6.
|
| 607 |
-
"loss": 0.
|
| 608 |
"step": 2075
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 1.4856738592147152,
|
| 612 |
-
"grad_norm":
|
| 613 |
-
"learning_rate": 6.
|
| 614 |
-
"loss": 0.
|
| 615 |
"step": 2100
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 1.5033604527767952,
|
| 619 |
-
"grad_norm":
|
| 620 |
-
"learning_rate": 6.
|
| 621 |
-
"loss": 0.
|
| 622 |
"step": 2125
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 1.5210470463388752,
|
| 626 |
-
"grad_norm":
|
| 627 |
-
"learning_rate": 6.
|
| 628 |
-
"loss": 0.
|
| 629 |
"step": 2150
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 1.538733639900955,
|
| 633 |
-
"grad_norm":
|
| 634 |
-
"learning_rate": 6.
|
| 635 |
-
"loss": 0.
|
| 636 |
"step": 2175
|
| 637 |
},
|
| 638 |
{
|
| 639 |
"epoch": 1.556420233463035,
|
| 640 |
-
"grad_norm":
|
| 641 |
-
"learning_rate": 6.
|
| 642 |
-
"loss": 0.
|
| 643 |
"step": 2200
|
| 644 |
},
|
| 645 |
{
|
| 646 |
"epoch": 1.574106827025115,
|
| 647 |
-
"grad_norm":
|
| 648 |
-
"learning_rate": 6.
|
| 649 |
-
"loss": 0.
|
| 650 |
"step": 2225
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 1.5917934205871949,
|
| 654 |
-
"grad_norm":
|
| 655 |
-
"learning_rate": 6.
|
| 656 |
-
"loss": 0.
|
| 657 |
"step": 2250
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 1.6094800141492749,
|
| 661 |
-
"grad_norm":
|
| 662 |
-
"learning_rate": 6.
|
| 663 |
-
"loss": 0.
|
| 664 |
"step": 2275
|
| 665 |
},
|
| 666 |
{
|
| 667 |
"epoch": 1.627166607711355,
|
| 668 |
-
"grad_norm":
|
| 669 |
-
"learning_rate":
|
| 670 |
-
"loss": 0.
|
| 671 |
"step": 2300
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 1.6448532012734347,
|
| 675 |
-
"grad_norm":
|
| 676 |
-
"learning_rate": 5.
|
| 677 |
-
"loss": 0.
|
| 678 |
"step": 2325
|
| 679 |
},
|
| 680 |
{
|
| 681 |
"epoch": 1.6625397948355147,
|
| 682 |
-
"grad_norm":
|
| 683 |
-
"learning_rate": 5.
|
| 684 |
-
"loss": 0.
|
| 685 |
"step": 2350
|
| 686 |
},
|
| 687 |
{
|
| 688 |
"epoch": 1.6802263883975948,
|
| 689 |
-
"grad_norm":
|
| 690 |
-
"learning_rate": 5.
|
| 691 |
-
"loss": 0.
|
| 692 |
"step": 2375
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 1.6979129819596745,
|
| 696 |
-
"grad_norm":
|
| 697 |
-
"learning_rate": 5.
|
| 698 |
-
"loss": 0.
|
| 699 |
"step": 2400
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 1.7155995755217543,
|
| 703 |
-
"grad_norm":
|
| 704 |
-
"learning_rate": 5.
|
| 705 |
-
"loss": 0.
|
| 706 |
"step": 2425
|
| 707 |
},
|
| 708 |
{
|
| 709 |
"epoch": 1.7332861690838346,
|
| 710 |
-
"grad_norm":
|
| 711 |
-
"learning_rate": 5.
|
| 712 |
-
"loss": 0.
|
| 713 |
"step": 2450
|
| 714 |
},
|
| 715 |
{
|
| 716 |
"epoch": 1.7509727626459144,
|
| 717 |
-
"grad_norm":
|
| 718 |
-
"learning_rate": 5.
|
| 719 |
-
"loss": 0.
|
| 720 |
"step": 2475
|
| 721 |
},
|
| 722 |
{
|
| 723 |
"epoch": 1.7686593562079942,
|
| 724 |
-
"grad_norm":
|
| 725 |
-
"learning_rate": 5.
|
| 726 |
-
"loss": 0.
|
| 727 |
"step": 2500
|
| 728 |
},
|
| 729 |
{
|
| 730 |
"epoch": 1.7863459497700744,
|
| 731 |
-
"grad_norm":
|
| 732 |
-
"learning_rate": 5.
|
| 733 |
-
"loss": 0.
|
| 734 |
"step": 2525
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 1.8040325433321542,
|
| 738 |
-
"grad_norm":
|
| 739 |
-
"learning_rate": 5.
|
| 740 |
-
"loss": 0.
|
| 741 |
"step": 2550
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 1.821719136894234,
|
| 745 |
-
"grad_norm":
|
| 746 |
-
"learning_rate": 5.
|
| 747 |
-
"loss": 0.
|
| 748 |
"step": 2575
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 1.839405730456314,
|
| 752 |
-
"grad_norm":
|
| 753 |
-
"learning_rate": 5.
|
| 754 |
-
"loss": 0.
|
| 755 |
"step": 2600
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 1.857092324018394,
|
| 759 |
-
"grad_norm":
|
| 760 |
-
"learning_rate": 5.
|
| 761 |
-
"loss": 0.
|
| 762 |
"step": 2625
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 1.8747789175804739,
|
| 766 |
-
"grad_norm":
|
| 767 |
-
"learning_rate": 5.
|
| 768 |
-
"loss": 0.
|
| 769 |
"step": 2650
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 1.892465511142554,
|
| 773 |
-
"grad_norm":
|
| 774 |
-
"learning_rate": 5.
|
| 775 |
-
"loss": 0.
|
| 776 |
"step": 2675
|
| 777 |
},
|
| 778 |
{
|
| 779 |
"epoch": 1.910152104704634,
|
| 780 |
-
"grad_norm":
|
| 781 |
-
"learning_rate": 5.
|
| 782 |
-
"loss": 0.
|
| 783 |
"step": 2700
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 1.9278386982667137,
|
| 787 |
-
"grad_norm":
|
| 788 |
-
"learning_rate": 5.
|
| 789 |
-
"loss": 0.
|
| 790 |
"step": 2725
|
| 791 |
},
|
| 792 |
{
|
| 793 |
"epoch": 1.9455252918287937,
|
| 794 |
-
"grad_norm":
|
| 795 |
-
"learning_rate":
|
| 796 |
-
"loss": 0.
|
| 797 |
"step": 2750
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 1.9632118853908738,
|
| 801 |
-
"grad_norm":
|
| 802 |
-
"learning_rate": 4.
|
| 803 |
-
"loss": 0.
|
| 804 |
"step": 2775
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 1.9808984789529536,
|
| 808 |
-
"grad_norm":
|
| 809 |
-
"learning_rate": 4.
|
| 810 |
-
"loss": 0.
|
| 811 |
"step": 2800
|
| 812 |
},
|
| 813 |
{
|
| 814 |
"epoch": 1.9985850725150336,
|
| 815 |
-
"grad_norm":
|
| 816 |
-
"learning_rate": 4.
|
| 817 |
-
"loss": 0.
|
| 818 |
"step": 2825
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 2.0162716660771136,
|
| 822 |
-
"grad_norm":
|
| 823 |
-
"learning_rate": 4.
|
| 824 |
-
"loss": 0.
|
| 825 |
"step": 2850
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 2.0339582596391934,
|
| 829 |
-
"grad_norm":
|
| 830 |
-
"learning_rate": 4.
|
| 831 |
-
"loss": 0.
|
| 832 |
"step": 2875
|
| 833 |
},
|
| 834 |
{
|
| 835 |
"epoch": 2.0516448532012737,
|
| 836 |
-
"grad_norm":
|
| 837 |
-
"learning_rate": 4.
|
| 838 |
-
"loss": 0.
|
| 839 |
"step": 2900
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 2.0693314467633535,
|
| 843 |
-
"grad_norm":
|
| 844 |
-
"learning_rate": 4.
|
| 845 |
-
"loss": 0.
|
| 846 |
"step": 2925
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 2.0870180403254333,
|
| 850 |
-
"grad_norm": 1.
|
| 851 |
-
"learning_rate": 4.
|
| 852 |
-
"loss": 0.
|
| 853 |
"step": 2950
|
| 854 |
},
|
| 855 |
{
|
| 856 |
"epoch": 2.104704633887513,
|
| 857 |
-
"grad_norm":
|
| 858 |
-
"learning_rate": 4.
|
| 859 |
-
"loss": 0.
|
| 860 |
"step": 2975
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 2.1223912274495933,
|
| 864 |
-
"grad_norm": 1.
|
| 865 |
-
"learning_rate": 4.
|
| 866 |
-
"loss": 0.
|
| 867 |
"step": 3000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 2.1223912274495933,
|
| 871 |
-
"eval_loss": 0.
|
| 872 |
-
"eval_runtime":
|
| 873 |
-
"eval_samples_per_second": 2.
|
| 874 |
-
"eval_steps_per_second": 0.
|
| 875 |
-
"eval_wer": 0.
|
| 876 |
"step": 3000
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 2.140077821011673,
|
| 880 |
-
"grad_norm":
|
| 881 |
-
"learning_rate": 4.
|
| 882 |
-
"loss": 0.
|
| 883 |
"step": 3025
|
| 884 |
},
|
| 885 |
{
|
| 886 |
"epoch": 2.157764414573753,
|
| 887 |
-
"grad_norm":
|
| 888 |
-
"learning_rate": 4.
|
| 889 |
-
"loss": 0.
|
| 890 |
"step": 3050
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 2.175451008135833,
|
| 894 |
-
"grad_norm":
|
| 895 |
-
"learning_rate": 4.
|
| 896 |
-
"loss": 0.
|
| 897 |
"step": 3075
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 2.193137601697913,
|
| 901 |
-
"grad_norm":
|
| 902 |
-
"learning_rate": 4.
|
| 903 |
-
"loss": 0.
|
| 904 |
"step": 3100
|
| 905 |
},
|
| 906 |
{
|
| 907 |
"epoch": 2.2108241952599927,
|
| 908 |
-
"grad_norm":
|
| 909 |
-
"learning_rate": 4.
|
| 910 |
-
"loss": 0.
|
| 911 |
"step": 3125
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 2.228510788822073,
|
| 915 |
-
"grad_norm":
|
| 916 |
-
"learning_rate": 4.
|
| 917 |
-
"loss": 0.
|
| 918 |
"step": 3150
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 2.246197382384153,
|
| 922 |
-
"grad_norm":
|
| 923 |
-
"learning_rate": 4.
|
| 924 |
-
"loss": 0.
|
| 925 |
"step": 3175
|
| 926 |
},
|
| 927 |
{
|
| 928 |
"epoch": 2.2638839759462326,
|
| 929 |
-
"grad_norm":
|
| 930 |
-
"learning_rate": 4.
|
| 931 |
-
"loss": 0.
|
| 932 |
"step": 3200
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 2.281570569508313,
|
| 936 |
-
"grad_norm":
|
| 937 |
-
"learning_rate": 3.
|
| 938 |
-
"loss": 0.
|
| 939 |
"step": 3225
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 2.2992571630703926,
|
| 943 |
-
"grad_norm":
|
| 944 |
-
"learning_rate": 3.
|
| 945 |
-
"loss": 0.
|
| 946 |
"step": 3250
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 2.3169437566324724,
|
| 950 |
-
"grad_norm":
|
| 951 |
-
"learning_rate": 3.
|
| 952 |
-
"loss": 0.
|
| 953 |
"step": 3275
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 2.3346303501945527,
|
| 957 |
-
"grad_norm":
|
| 958 |
-
"learning_rate": 3.
|
| 959 |
-
"loss": 0.
|
| 960 |
"step": 3300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 2.3523169437566325,
|
| 964 |
-
"grad_norm":
|
| 965 |
-
"learning_rate": 3.
|
| 966 |
-
"loss": 0.
|
| 967 |
"step": 3325
|
| 968 |
},
|
| 969 |
{
|
| 970 |
"epoch": 2.3700035373187123,
|
| 971 |
-
"grad_norm":
|
| 972 |
-
"learning_rate": 3.
|
| 973 |
-
"loss": 0.
|
| 974 |
"step": 3350
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 2.3876901308807925,
|
| 978 |
-
"grad_norm":
|
| 979 |
-
"learning_rate": 3.
|
| 980 |
-
"loss": 0.
|
| 981 |
"step": 3375
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 2.4053767244428723,
|
| 985 |
-
"grad_norm": 1.
|
| 986 |
-
"learning_rate": 3.
|
| 987 |
-
"loss": 0.
|
| 988 |
"step": 3400
|
| 989 |
},
|
| 990 |
{
|
| 991 |
"epoch": 2.423063318004952,
|
| 992 |
-
"grad_norm":
|
| 993 |
-
"learning_rate": 3.
|
| 994 |
-
"loss": 0.
|
| 995 |
"step": 3425
|
| 996 |
},
|
| 997 |
{
|
| 998 |
"epoch": 2.4407499115670324,
|
| 999 |
-
"grad_norm": 1.
|
| 1000 |
-
"learning_rate": 3.
|
| 1001 |
-
"loss": 0.
|
| 1002 |
"step": 3450
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 2.458436505129112,
|
| 1006 |
-
"grad_norm":
|
| 1007 |
-
"learning_rate": 3.
|
| 1008 |
-
"loss": 0.
|
| 1009 |
"step": 3475
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 2.476123098691192,
|
| 1013 |
-
"grad_norm": 1.
|
| 1014 |
-
"learning_rate": 3.
|
| 1015 |
-
"loss": 0.
|
| 1016 |
"step": 3500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 2.493809692253272,
|
| 1020 |
-
"grad_norm":
|
| 1021 |
-
"learning_rate": 3.
|
| 1022 |
-
"loss": 0.
|
| 1023 |
"step": 3525
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 2.511496285815352,
|
| 1027 |
-
"grad_norm":
|
| 1028 |
-
"learning_rate": 3.
|
| 1029 |
-
"loss": 0.
|
| 1030 |
"step": 3550
|
| 1031 |
},
|
| 1032 |
{
|
| 1033 |
"epoch": 2.529182879377432,
|
| 1034 |
-
"grad_norm":
|
| 1035 |
-
"learning_rate": 3.
|
| 1036 |
-
"loss": 0.
|
| 1037 |
"step": 3575
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 2.5468694729395116,
|
| 1041 |
-
"grad_norm":
|
| 1042 |
-
"learning_rate": 3.
|
| 1043 |
-
"loss": 0.
|
| 1044 |
"step": 3600
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 2.564556066501592,
|
| 1048 |
-
"grad_norm":
|
| 1049 |
-
"learning_rate": 3.
|
| 1050 |
-
"loss": 0.
|
| 1051 |
"step": 3625
|
| 1052 |
},
|
| 1053 |
{
|
| 1054 |
"epoch": 2.5822426600636716,
|
| 1055 |
-
"grad_norm":
|
| 1056 |
-
"learning_rate":
|
| 1057 |
-
"loss": 0.
|
| 1058 |
"step": 3650
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 2.599929253625752,
|
| 1062 |
-
"grad_norm":
|
| 1063 |
-
"learning_rate": 2.
|
| 1064 |
-
"loss": 0.
|
| 1065 |
"step": 3675
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 2.6176158471878317,
|
| 1069 |
-
"grad_norm":
|
| 1070 |
-
"learning_rate": 2.
|
| 1071 |
-
"loss": 0.
|
| 1072 |
"step": 3700
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 2.6353024407499115,
|
| 1076 |
-
"grad_norm":
|
| 1077 |
-
"learning_rate": 2.
|
| 1078 |
-
"loss": 0.
|
| 1079 |
"step": 3725
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 2.6529890343119913,
|
| 1083 |
-
"grad_norm":
|
| 1084 |
-
"learning_rate": 2.
|
| 1085 |
-
"loss": 0.
|
| 1086 |
"step": 3750
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 2.6706756278740715,
|
| 1090 |
-
"grad_norm":
|
| 1091 |
-
"learning_rate": 2.
|
| 1092 |
-
"loss": 0.
|
| 1093 |
"step": 3775
|
| 1094 |
},
|
| 1095 |
{
|
| 1096 |
"epoch": 2.6883622214361513,
|
| 1097 |
-
"grad_norm":
|
| 1098 |
-
"learning_rate": 2.
|
| 1099 |
-
"loss": 0.
|
| 1100 |
"step": 3800
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 2.7060488149982316,
|
| 1104 |
-
"grad_norm":
|
| 1105 |
-
"learning_rate": 2.
|
| 1106 |
-
"loss": 0.
|
| 1107 |
"step": 3825
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 2.7237354085603114,
|
| 1111 |
-
"grad_norm":
|
| 1112 |
-
"learning_rate": 2.
|
| 1113 |
-
"loss": 0.
|
| 1114 |
"step": 3850
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
"epoch": 2.741422002122391,
|
| 1118 |
-
"grad_norm":
|
| 1119 |
-
"learning_rate": 2.
|
| 1120 |
-
"loss": 0.
|
| 1121 |
"step": 3875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 2.759108595684471,
|
| 1125 |
-
"grad_norm":
|
| 1126 |
-
"learning_rate": 2.
|
| 1127 |
-
"loss": 0.
|
| 1128 |
"step": 3900
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 2.776795189246551,
|
| 1132 |
-
"grad_norm": 1.
|
| 1133 |
-
"learning_rate": 2.
|
| 1134 |
-
"loss": 0.
|
| 1135 |
"step": 3925
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
"epoch": 2.794481782808631,
|
| 1139 |
-
"grad_norm":
|
| 1140 |
-
"learning_rate": 2.
|
| 1141 |
-
"loss": 0.
|
| 1142 |
"step": 3950
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 2.8121683763707113,
|
| 1146 |
-
"grad_norm":
|
| 1147 |
-
"learning_rate": 2.
|
| 1148 |
-
"loss": 0.
|
| 1149 |
"step": 3975
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 2.829854969932791,
|
| 1153 |
-
"grad_norm":
|
| 1154 |
-
"learning_rate": 2.
|
| 1155 |
-
"loss": 0.
|
| 1156 |
"step": 4000
|
| 1157 |
},
|
| 1158 |
{
|
| 1159 |
"epoch": 2.829854969932791,
|
| 1160 |
-
"eval_loss": 0.
|
| 1161 |
-
"eval_runtime":
|
| 1162 |
-
"eval_samples_per_second": 2.
|
| 1163 |
-
"eval_steps_per_second": 0.
|
| 1164 |
-
"eval_wer": 0.
|
| 1165 |
"step": 4000
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
"epoch": 2.847541563494871,
|
| 1169 |
-
"grad_norm":
|
| 1170 |
-
"learning_rate": 2.
|
| 1171 |
-
"loss": 0.
|
| 1172 |
"step": 4025
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 2.8652281570569507,
|
| 1176 |
-
"grad_norm":
|
| 1177 |
-
"learning_rate": 2.
|
| 1178 |
-
"loss": 0.
|
| 1179 |
"step": 4050
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 2.882914750619031,
|
| 1183 |
-
"grad_norm":
|
| 1184 |
-
"learning_rate": 2.
|
| 1185 |
-
"loss": 0.
|
| 1186 |
"step": 4075
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 2.9006013441811107,
|
| 1190 |
-
"grad_norm":
|
| 1191 |
-
"learning_rate": 2.
|
| 1192 |
-
"loss": 0.
|
| 1193 |
"step": 4100
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
"epoch": 2.9182879377431905,
|
| 1197 |
-
"grad_norm":
|
| 1198 |
-
"learning_rate": 1.
|
| 1199 |
-
"loss": 0.
|
| 1200 |
"step": 4125
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 2.9359745313052708,
|
| 1204 |
-
"grad_norm":
|
| 1205 |
-
"learning_rate": 1.
|
| 1206 |
-
"loss": 0.
|
| 1207 |
"step": 4150
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
"epoch": 2.9536611248673506,
|
| 1211 |
-
"grad_norm":
|
| 1212 |
-
"learning_rate": 1.
|
| 1213 |
-
"loss": 0.
|
| 1214 |
"step": 4175
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 2.9713477184294304,
|
| 1218 |
-
"grad_norm":
|
| 1219 |
-
"learning_rate": 1.
|
| 1220 |
-
"loss": 0.
|
| 1221 |
"step": 4200
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 2.9890343119915106,
|
| 1225 |
-
"grad_norm":
|
| 1226 |
-
"learning_rate": 1.
|
| 1227 |
-
"loss": 0.
|
| 1228 |
"step": 4225
|
| 1229 |
},
|
| 1230 |
{
|
| 1231 |
"epoch": 3.0067209055535904,
|
| 1232 |
-
"grad_norm":
|
| 1233 |
-
"learning_rate": 1.
|
| 1234 |
-
"loss": 0.
|
| 1235 |
"step": 4250
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 3.02440749911567,
|
| 1239 |
-
"grad_norm":
|
| 1240 |
-
"learning_rate": 1.
|
| 1241 |
-
"loss": 0.
|
| 1242 |
"step": 4275
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 3.0420940926777504,
|
| 1246 |
-
"grad_norm": 1.
|
| 1247 |
-
"learning_rate": 1.
|
| 1248 |
-
"loss": 0.
|
| 1249 |
"step": 4300
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 3.0597806862398302,
|
| 1253 |
-
"grad_norm": 2.
|
| 1254 |
-
"learning_rate": 1.
|
| 1255 |
-
"loss": 0.
|
| 1256 |
"step": 4325
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 3.07746727980191,
|
| 1260 |
-
"grad_norm":
|
| 1261 |
-
"learning_rate": 1.
|
| 1262 |
-
"loss": 0.
|
| 1263 |
"step": 4350
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 3.0951538733639903,
|
| 1267 |
-
"grad_norm": 1.
|
| 1268 |
-
"learning_rate": 1.
|
| 1269 |
-
"loss": 0.
|
| 1270 |
"step": 4375
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 3.11284046692607,
|
| 1274 |
-
"grad_norm": 1.
|
| 1275 |
-
"learning_rate": 1.
|
| 1276 |
-
"loss": 0.
|
| 1277 |
"step": 4400
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 3.13052706048815,
|
| 1281 |
-
"grad_norm": 1.
|
| 1282 |
-
"learning_rate": 1.
|
| 1283 |
-
"loss": 0.
|
| 1284 |
"step": 4425
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 3.14821365405023,
|
| 1288 |
-
"grad_norm": 0.
|
| 1289 |
-
"learning_rate": 1.
|
| 1290 |
-
"loss": 0.
|
| 1291 |
"step": 4450
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 3.16590024761231,
|
| 1295 |
-
"grad_norm":
|
| 1296 |
-
"learning_rate": 1.
|
| 1297 |
-
"loss": 0.
|
| 1298 |
"step": 4475
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 3.1835868411743897,
|
| 1302 |
-
"grad_norm": 1.
|
| 1303 |
-
"learning_rate": 1.
|
| 1304 |
-
"loss": 0.
|
| 1305 |
"step": 4500
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 3.2012734347364695,
|
| 1309 |
-
"grad_norm": 1.
|
| 1310 |
-
"learning_rate": 1.
|
| 1311 |
-
"loss": 0.
|
| 1312 |
"step": 4525
|
| 1313 |
},
|
| 1314 |
{
|
| 1315 |
"epoch": 3.2189600282985498,
|
| 1316 |
-
"grad_norm": 0.
|
| 1317 |
-
"learning_rate": 1.
|
| 1318 |
-
"loss": 0.
|
| 1319 |
"step": 4550
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 3.2366466218606296,
|
| 1323 |
-
"grad_norm":
|
| 1324 |
-
"learning_rate": 9.
|
| 1325 |
-
"loss": 0.
|
| 1326 |
"step": 4575
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 3.25433321542271,
|
| 1330 |
-
"grad_norm":
|
| 1331 |
-
"learning_rate": 8.
|
| 1332 |
-
"loss": 0.
|
| 1333 |
"step": 4600
|
| 1334 |
},
|
| 1335 |
{
|
| 1336 |
"epoch": 3.2720198089847896,
|
| 1337 |
-
"grad_norm":
|
| 1338 |
-
"learning_rate": 8.
|
| 1339 |
-
"loss": 0.
|
| 1340 |
"step": 4625
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 3.2897064025468694,
|
| 1344 |
-
"grad_norm":
|
| 1345 |
-
"learning_rate": 7.
|
| 1346 |
-
"loss": 0.
|
| 1347 |
"step": 4650
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 3.307392996108949,
|
| 1351 |
-
"grad_norm": 1.
|
| 1352 |
-
"learning_rate": 7.
|
| 1353 |
-
"loss": 0.
|
| 1354 |
"step": 4675
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
"epoch": 3.3250795896710295,
|
| 1358 |
-
"grad_norm":
|
| 1359 |
-
"learning_rate": 6.
|
| 1360 |
-
"loss": 0.
|
| 1361 |
"step": 4700
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 3.3427661832331093,
|
| 1365 |
-
"grad_norm": 1.
|
| 1366 |
-
"learning_rate": 6.
|
| 1367 |
-
"loss": 0.
|
| 1368 |
"step": 4725
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 3.360452776795189,
|
| 1372 |
-
"grad_norm": 1.
|
| 1373 |
-
"learning_rate": 5.
|
| 1374 |
-
"loss": 0.
|
| 1375 |
"step": 4750
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 3.3781393703572693,
|
| 1379 |
-
"grad_norm": 0.
|
| 1380 |
-
"learning_rate": 5.
|
| 1381 |
-
"loss": 0.
|
| 1382 |
"step": 4775
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 3.395825963919349,
|
| 1386 |
-
"grad_norm": 0.
|
| 1387 |
-
"learning_rate": 4.
|
| 1388 |
-
"loss": 0.
|
| 1389 |
"step": 4800
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 3.413512557481429,
|
| 1393 |
-
"grad_norm": 1.
|
| 1394 |
-
"learning_rate": 3.
|
| 1395 |
-
"loss": 0.
|
| 1396 |
"step": 4825
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 3.431199151043509,
|
| 1400 |
-
"grad_norm": 1.
|
| 1401 |
-
"learning_rate": 3.
|
| 1402 |
-
"loss": 0.
|
| 1403 |
"step": 4850
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 3.448885744605589,
|
| 1407 |
-
"grad_norm": 0.
|
| 1408 |
-
"learning_rate": 2.
|
| 1409 |
-
"loss": 0.
|
| 1410 |
"step": 4875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 3.4665723381676687,
|
| 1414 |
-
"grad_norm":
|
| 1415 |
-
"learning_rate": 2.
|
| 1416 |
-
"loss": 0.
|
| 1417 |
"step": 4900
|
| 1418 |
},
|
| 1419 |
{
|
| 1420 |
"epoch": 3.484258931729749,
|
| 1421 |
-
"grad_norm":
|
| 1422 |
-
"learning_rate": 1.
|
| 1423 |
-
"loss": 0.
|
| 1424 |
"step": 4925
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 3.501945525291829,
|
| 1428 |
-
"grad_norm":
|
| 1429 |
-
"learning_rate": 1.
|
| 1430 |
-
"loss": 0.
|
| 1431 |
"step": 4950
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 3.5196321188539086,
|
| 1435 |
-
"grad_norm": 1.
|
| 1436 |
-
"learning_rate":
|
| 1437 |
-
"loss": 0.
|
| 1438 |
"step": 4975
|
| 1439 |
},
|
| 1440 |
{
|
| 1441 |
"epoch": 3.537318712415989,
|
| 1442 |
-
"grad_norm": 0.
|
| 1443 |
-
"learning_rate":
|
| 1444 |
-
"loss": 0.
|
| 1445 |
"step": 5000
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 3.537318712415989,
|
| 1449 |
-
"eval_loss": 0.
|
| 1450 |
-
"eval_runtime":
|
| 1451 |
-
"eval_samples_per_second": 2.
|
| 1452 |
-
"eval_steps_per_second": 0.
|
| 1453 |
-
"eval_wer": 0.
|
| 1454 |
"step": 5000
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
"epoch": 3.537318712415989,
|
| 1458 |
"step": 5000,
|
| 1459 |
"total_flos": 5.435079965953229e+20,
|
| 1460 |
-
"train_loss": 0.
|
| 1461 |
-
"train_runtime":
|
| 1462 |
-
"train_samples_per_second": 2.
|
| 1463 |
-
"train_steps_per_second": 0.
|
| 1464 |
}
|
| 1465 |
],
|
| 1466 |
"logging_steps": 25,
|
|
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.017686593562079942,
|
| 13 |
+
"grad_norm": 11.730201721191406,
|
| 14 |
+
"learning_rate": 4.800000000000001e-07,
|
| 15 |
+
"loss": 1.906,
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.035373187124159884,
|
| 20 |
+
"grad_norm": 9.43655014038086,
|
| 21 |
+
"learning_rate": 9.800000000000001e-07,
|
| 22 |
+
"loss": 1.5891,
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.05305978068623983,
|
| 27 |
+
"grad_norm": 8.554606437683105,
|
| 28 |
+
"learning_rate": 1.48e-06,
|
| 29 |
+
"loss": 0.9582,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.07074637424831977,
|
| 34 |
+
"grad_norm": 7.075723171234131,
|
| 35 |
+
"learning_rate": 1.98e-06,
|
| 36 |
+
"loss": 0.8164,
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.08843296781039972,
|
| 41 |
+
"grad_norm": 7.806982517242432,
|
| 42 |
+
"learning_rate": 2.46e-06,
|
| 43 |
+
"loss": 0.7283,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.10611956137247966,
|
| 48 |
+
"grad_norm": 6.871402740478516,
|
| 49 |
+
"learning_rate": 2.96e-06,
|
| 50 |
+
"loss": 0.721,
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.1238061549345596,
|
| 55 |
+
"grad_norm": 7.813017845153809,
|
| 56 |
+
"learning_rate": 3.46e-06,
|
| 57 |
+
"loss": 0.68,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.14149274849663954,
|
| 62 |
+
"grad_norm": 7.54969596862793,
|
| 63 |
+
"learning_rate": 3.96e-06,
|
| 64 |
+
"loss": 0.7248,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.1591793420587195,
|
| 69 |
+
"grad_norm": 7.196774482727051,
|
| 70 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 71 |
+
"loss": 0.6417,
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.17686593562079944,
|
| 76 |
+
"grad_norm": 8.30762004852295,
|
| 77 |
+
"learning_rate": 4.960000000000001e-06,
|
| 78 |
+
"loss": 0.6594,
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.19455252918287938,
|
| 83 |
+
"grad_norm": 6.641123294830322,
|
| 84 |
+
"learning_rate": 5.460000000000001e-06,
|
| 85 |
+
"loss": 0.6359,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.21223912274495932,
|
| 90 |
+
"grad_norm": 8.47127914428711,
|
| 91 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 92 |
+
"loss": 0.6455,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.22992571630703926,
|
| 97 |
+
"grad_norm": 8.078540802001953,
|
| 98 |
+
"learning_rate": 6.460000000000001e-06,
|
| 99 |
+
"loss": 0.6563,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.2476123098691192,
|
| 104 |
+
"grad_norm": 7.481531620025635,
|
| 105 |
+
"learning_rate": 6.96e-06,
|
| 106 |
+
"loss": 0.6099,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.26529890343119916,
|
| 111 |
+
"grad_norm": 8.171069145202637,
|
| 112 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 113 |
+
"loss": 0.5922,
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.2829854969932791,
|
| 118 |
+
"grad_norm": 7.401958465576172,
|
| 119 |
+
"learning_rate": 7.960000000000002e-06,
|
| 120 |
+
"loss": 0.6232,
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"epoch": 0.30067209055535904,
|
| 125 |
+
"grad_norm": 6.448988437652588,
|
| 126 |
+
"learning_rate": 8.46e-06,
|
| 127 |
+
"loss": 0.6328,
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.318358684117439,
|
| 132 |
+
"grad_norm": 6.956504821777344,
|
| 133 |
+
"learning_rate": 8.96e-06,
|
| 134 |
+
"loss": 0.6026,
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.3360452776795189,
|
| 139 |
+
"grad_norm": 8.774309158325195,
|
| 140 |
+
"learning_rate": 9.460000000000001e-06,
|
| 141 |
+
"loss": 0.6318,
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 0.3537318712415989,
|
| 146 |
+
"grad_norm": 6.220521450042725,
|
| 147 |
+
"learning_rate": 9.960000000000001e-06,
|
| 148 |
+
"loss": 0.6186,
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.3714184648036788,
|
| 153 |
+
"grad_norm": 6.165685176849365,
|
| 154 |
+
"learning_rate": 9.94888888888889e-06,
|
| 155 |
+
"loss": 0.5509,
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.38910505836575876,
|
| 160 |
+
"grad_norm": 6.3863348960876465,
|
| 161 |
+
"learning_rate": 9.893333333333334e-06,
|
| 162 |
+
"loss": 0.63,
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.40679165192783867,
|
| 167 |
+
"grad_norm": 6.65989351272583,
|
| 168 |
+
"learning_rate": 9.837777777777778e-06,
|
| 169 |
+
"loss": 0.5981,
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.42447824548991864,
|
| 174 |
+
"grad_norm": 5.605571746826172,
|
| 175 |
+
"learning_rate": 9.782222222222222e-06,
|
| 176 |
+
"loss": 0.5611,
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.4421648390519986,
|
| 181 |
+
"grad_norm": 7.080026149749756,
|
| 182 |
+
"learning_rate": 9.726666666666668e-06,
|
| 183 |
+
"loss": 0.5498,
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.4598514326140785,
|
| 188 |
+
"grad_norm": 5.857707977294922,
|
| 189 |
+
"learning_rate": 9.671111111111112e-06,
|
| 190 |
+
"loss": 0.5942,
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 0.4775380261761585,
|
| 195 |
+
"grad_norm": 6.002816200256348,
|
| 196 |
+
"learning_rate": 9.615555555555558e-06,
|
| 197 |
+
"loss": 0.5488,
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
"epoch": 0.4952246197382384,
|
| 202 |
+
"grad_norm": 6.393026828765869,
|
| 203 |
+
"learning_rate": 9.56e-06,
|
| 204 |
+
"loss": 0.5464,
|
| 205 |
"step": 700
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.5129112133003184,
|
| 209 |
+
"grad_norm": 6.212553024291992,
|
| 210 |
+
"learning_rate": 9.504444444444446e-06,
|
| 211 |
+
"loss": 0.5416,
|
| 212 |
"step": 725
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.5305978068623983,
|
| 216 |
+
"grad_norm": 6.353979110717773,
|
| 217 |
+
"learning_rate": 9.44888888888889e-06,
|
| 218 |
+
"loss": 0.5493,
|
| 219 |
"step": 750
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.5482844004244782,
|
| 223 |
+
"grad_norm": 5.198184013366699,
|
| 224 |
+
"learning_rate": 9.393333333333334e-06,
|
| 225 |
+
"loss": 0.562,
|
| 226 |
"step": 775
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.5659709939865581,
|
| 230 |
+
"grad_norm": 5.739233493804932,
|
| 231 |
+
"learning_rate": 9.33777777777778e-06,
|
| 232 |
+
"loss": 0.5245,
|
| 233 |
"step": 800
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 0.5836575875486382,
|
| 237 |
+
"grad_norm": 5.637094974517822,
|
| 238 |
+
"learning_rate": 9.282222222222222e-06,
|
| 239 |
+
"loss": 0.5056,
|
| 240 |
"step": 825
|
| 241 |
},
|
| 242 |
{
|
| 243 |
"epoch": 0.6013441811107181,
|
| 244 |
+
"grad_norm": 5.2869181632995605,
|
| 245 |
+
"learning_rate": 9.226666666666668e-06,
|
| 246 |
+
"loss": 0.495,
|
| 247 |
"step": 850
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"epoch": 0.619030774672798,
|
| 251 |
+
"grad_norm": 6.895484447479248,
|
| 252 |
+
"learning_rate": 9.171111111111112e-06,
|
| 253 |
+
"loss": 0.5311,
|
| 254 |
"step": 875
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.636717368234878,
|
| 258 |
+
"grad_norm": 5.642014026641846,
|
| 259 |
+
"learning_rate": 9.115555555555556e-06,
|
| 260 |
+
"loss": 0.4611,
|
| 261 |
"step": 900
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.6544039617969579,
|
| 265 |
+
"grad_norm": 6.39215087890625,
|
| 266 |
+
"learning_rate": 9.060000000000001e-06,
|
| 267 |
+
"loss": 0.5128,
|
| 268 |
"step": 925
|
| 269 |
},
|
| 270 |
{
|
| 271 |
"epoch": 0.6720905553590378,
|
| 272 |
+
"grad_norm": 4.879695892333984,
|
| 273 |
+
"learning_rate": 9.004444444444445e-06,
|
| 274 |
+
"loss": 0.502,
|
| 275 |
"step": 950
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 0.6897771489211177,
|
| 279 |
+
"grad_norm": 6.497096538543701,
|
| 280 |
+
"learning_rate": 8.94888888888889e-06,
|
| 281 |
+
"loss": 0.4927,
|
| 282 |
"step": 975
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 0.7074637424831978,
|
| 286 |
+
"grad_norm": 6.05295991897583,
|
| 287 |
+
"learning_rate": 8.893333333333333e-06,
|
| 288 |
+
"loss": 0.4825,
|
| 289 |
"step": 1000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.7074637424831978,
|
| 293 |
+
"eval_loss": 0.2707957625389099,
|
| 294 |
+
"eval_runtime": 4792.033,
|
| 295 |
+
"eval_samples_per_second": 2.246,
|
| 296 |
+
"eval_steps_per_second": 0.14,
|
| 297 |
+
"eval_wer": 0.18103144801689744,
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.7251503360452777,
|
| 302 |
+
"grad_norm": 6.068356990814209,
|
| 303 |
+
"learning_rate": 8.83777777777778e-06,
|
| 304 |
+
"loss": 0.4787,
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.7428369296073576,
|
| 309 |
+
"grad_norm": 6.832317352294922,
|
| 310 |
+
"learning_rate": 8.782222222222223e-06,
|
| 311 |
+
"loss": 0.4583,
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 0.7605235231694376,
|
| 316 |
+
"grad_norm": 5.528912544250488,
|
| 317 |
+
"learning_rate": 8.726666666666667e-06,
|
| 318 |
+
"loss": 0.4733,
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.7782101167315175,
|
| 323 |
+
"grad_norm": 6.220973968505859,
|
| 324 |
+
"learning_rate": 8.671111111111113e-06,
|
| 325 |
+
"loss": 0.4994,
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.7958967102935974,
|
| 330 |
+
"grad_norm": 6.543335914611816,
|
| 331 |
+
"learning_rate": 8.615555555555555e-06,
|
| 332 |
+
"loss": 0.4694,
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.8135833038556773,
|
| 337 |
+
"grad_norm": 4.369144439697266,
|
| 338 |
+
"learning_rate": 8.560000000000001e-06,
|
| 339 |
+
"loss": 0.4525,
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.8312698974177574,
|
| 344 |
+
"grad_norm": 4.928637981414795,
|
| 345 |
+
"learning_rate": 8.504444444444445e-06,
|
| 346 |
+
"loss": 0.4802,
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 0.8489564909798373,
|
| 351 |
+
"grad_norm": 5.016990661621094,
|
| 352 |
+
"learning_rate": 8.448888888888889e-06,
|
| 353 |
+
"loss": 0.462,
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.8666430845419172,
|
| 358 |
+
"grad_norm": 5.185001373291016,
|
| 359 |
+
"learning_rate": 8.393333333333335e-06,
|
| 360 |
+
"loss": 0.4189,
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
"epoch": 0.8843296781039972,
|
| 365 |
+
"grad_norm": 5.9500627517700195,
|
| 366 |
+
"learning_rate": 8.337777777777777e-06,
|
| 367 |
+
"loss": 0.4862,
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.9020162716660771,
|
| 372 |
+
"grad_norm": 4.7840166091918945,
|
| 373 |
+
"learning_rate": 8.282222222222223e-06,
|
| 374 |
+
"loss": 0.4201,
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.919702865228157,
|
| 379 |
+
"grad_norm": 5.590398788452148,
|
| 380 |
+
"learning_rate": 8.226666666666667e-06,
|
| 381 |
+
"loss": 0.438,
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
"epoch": 0.937389458790237,
|
| 386 |
+
"grad_norm": 5.356804370880127,
|
| 387 |
+
"learning_rate": 8.171111111111113e-06,
|
| 388 |
+
"loss": 0.4343,
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.955076052352317,
|
| 393 |
+
"grad_norm": 5.994735240936279,
|
| 394 |
+
"learning_rate": 8.115555555555557e-06,
|
| 395 |
+
"loss": 0.4281,
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 0.9727626459143969,
|
| 400 |
+
"grad_norm": 4.92245626449585,
|
| 401 |
+
"learning_rate": 8.06e-06,
|
| 402 |
+
"loss": 0.4711,
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
"epoch": 0.9904492394764768,
|
| 407 |
+
"grad_norm": 6.673391342163086,
|
| 408 |
+
"learning_rate": 8.004444444444445e-06,
|
| 409 |
+
"loss": 0.4396,
|
| 410 |
"step": 1400
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 1.0081358330385568,
|
| 414 |
+
"grad_norm": 4.157800197601318,
|
| 415 |
+
"learning_rate": 7.948888888888889e-06,
|
| 416 |
+
"loss": 0.3655,
|
| 417 |
"step": 1425
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 1.0258224266006368,
|
| 421 |
+
"grad_norm": 4.4156060218811035,
|
| 422 |
+
"learning_rate": 7.893333333333335e-06,
|
| 423 |
+
"loss": 0.3032,
|
| 424 |
"step": 1450
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 1.0435090201627166,
|
| 428 |
+
"grad_norm": 3.762485980987549,
|
| 429 |
+
"learning_rate": 7.837777777777779e-06,
|
| 430 |
+
"loss": 0.2597,
|
| 431 |
"step": 1475
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 1.0611956137247966,
|
| 435 |
+
"grad_norm": 3.0248515605926514,
|
| 436 |
+
"learning_rate": 7.782222222222223e-06,
|
| 437 |
+
"loss": 0.2829,
|
| 438 |
"step": 1500
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 1.0788822072868764,
|
| 442 |
+
"grad_norm": 5.492848873138428,
|
| 443 |
+
"learning_rate": 7.726666666666667e-06,
|
| 444 |
+
"loss": 0.229,
|
| 445 |
"step": 1525
|
| 446 |
},
|
| 447 |
{
|
| 448 |
"epoch": 1.0965688008489565,
|
| 449 |
+
"grad_norm": 5.225216388702393,
|
| 450 |
+
"learning_rate": 7.67111111111111e-06,
|
| 451 |
+
"loss": 0.2677,
|
| 452 |
"step": 1550
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 1.1142553944110365,
|
| 456 |
+
"grad_norm": 3.8239035606384277,
|
| 457 |
+
"learning_rate": 7.6155555555555564e-06,
|
| 458 |
+
"loss": 0.2517,
|
| 459 |
"step": 1575
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 1.1319419879731163,
|
| 463 |
+
"grad_norm": 5.077475070953369,
|
| 464 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 465 |
+
"loss": 0.2489,
|
| 466 |
"step": 1600
|
| 467 |
},
|
| 468 |
{
|
| 469 |
"epoch": 1.1496285815351963,
|
| 470 |
+
"grad_norm": 4.364721298217773,
|
| 471 |
+
"learning_rate": 7.504444444444445e-06,
|
| 472 |
+
"loss": 0.2598,
|
| 473 |
"step": 1625
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 1.1673151750972763,
|
| 477 |
+
"grad_norm": 3.3700525760650635,
|
| 478 |
+
"learning_rate": 7.44888888888889e-06,
|
| 479 |
+
"loss": 0.2472,
|
| 480 |
"step": 1650
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 1.1850017686593561,
|
| 484 |
+
"grad_norm": 4.548206329345703,
|
| 485 |
+
"learning_rate": 7.393333333333333e-06,
|
| 486 |
+
"loss": 0.2635,
|
| 487 |
"step": 1675
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 1.2026883622214362,
|
| 491 |
+
"grad_norm": 4.475757122039795,
|
| 492 |
+
"learning_rate": 7.337777777777778e-06,
|
| 493 |
+
"loss": 0.2345,
|
| 494 |
"step": 1700
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 1.2203749557835162,
|
| 498 |
+
"grad_norm": 4.205091953277588,
|
| 499 |
+
"learning_rate": 7.282222222222222e-06,
|
| 500 |
+
"loss": 0.2464,
|
| 501 |
"step": 1725
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 1.238061549345596,
|
| 505 |
+
"grad_norm": 4.778344631195068,
|
| 506 |
+
"learning_rate": 7.226666666666667e-06,
|
| 507 |
+
"loss": 0.238,
|
| 508 |
"step": 1750
|
| 509 |
},
|
| 510 |
{
|
| 511 |
"epoch": 1.255748142907676,
|
| 512 |
+
"grad_norm": 3.172917604446411,
|
| 513 |
+
"learning_rate": 7.171111111111112e-06,
|
| 514 |
+
"loss": 0.2364,
|
| 515 |
"step": 1775
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 1.2734347364697558,
|
| 519 |
+
"grad_norm": 4.275852203369141,
|
| 520 |
+
"learning_rate": 7.115555555555557e-06,
|
| 521 |
+
"loss": 0.2573,
|
| 522 |
"step": 1800
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 1.2911213300318358,
|
| 526 |
+
"grad_norm": 4.308451175689697,
|
| 527 |
+
"learning_rate": 7.06e-06,
|
| 528 |
+
"loss": 0.2569,
|
| 529 |
"step": 1825
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 1.3088079235939158,
|
| 533 |
+
"grad_norm": 4.079460144042969,
|
| 534 |
+
"learning_rate": 7.004444444444445e-06,
|
| 535 |
+
"loss": 0.2261,
|
| 536 |
"step": 1850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 1.3264945171559956,
|
| 540 |
+
"grad_norm": 6.641946792602539,
|
| 541 |
+
"learning_rate": 6.948888888888889e-06,
|
| 542 |
+
"loss": 0.2334,
|
| 543 |
"step": 1875
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 1.3441811107180757,
|
| 547 |
+
"grad_norm": 4.23126745223999,
|
| 548 |
+
"learning_rate": 6.893333333333334e-06,
|
| 549 |
+
"loss": 0.2307,
|
| 550 |
"step": 1900
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 1.3618677042801557,
|
| 554 |
+
"grad_norm": 3.1530086994171143,
|
| 555 |
+
"learning_rate": 6.837777777777779e-06,
|
| 556 |
+
"loss": 0.2154,
|
| 557 |
"step": 1925
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 1.3795542978422355,
|
| 561 |
+
"grad_norm": 3.8190791606903076,
|
| 562 |
+
"learning_rate": 6.782222222222222e-06,
|
| 563 |
+
"loss": 0.2274,
|
| 564 |
"step": 1950
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 1.3972408914043155,
|
| 568 |
+
"grad_norm": 3.953749179840088,
|
| 569 |
+
"learning_rate": 6.726666666666667e-06,
|
| 570 |
+
"loss": 0.2426,
|
| 571 |
"step": 1975
|
| 572 |
},
|
| 573 |
{
|
| 574 |
"epoch": 1.4149274849663955,
|
| 575 |
+
"grad_norm": 3.527085304260254,
|
| 576 |
+
"learning_rate": 6.671111111111112e-06,
|
| 577 |
+
"loss": 0.2262,
|
| 578 |
"step": 2000
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 1.4149274849663955,
|
| 582 |
+
"eval_loss": 0.2485629916191101,
|
| 583 |
+
"eval_runtime": 4723.0929,
|
| 584 |
+
"eval_samples_per_second": 2.279,
|
| 585 |
+
"eval_steps_per_second": 0.142,
|
| 586 |
+
"eval_wer": 0.15938160056324807,
|
| 587 |
"step": 2000
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 1.4326140785284753,
|
| 591 |
+
"grad_norm": 4.03377628326416,
|
| 592 |
+
"learning_rate": 6.615555555555556e-06,
|
| 593 |
+
"loss": 0.2367,
|
| 594 |
"step": 2025
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 1.4503006720905554,
|
| 598 |
+
"grad_norm": 3.128779888153076,
|
| 599 |
+
"learning_rate": 6.560000000000001e-06,
|
| 600 |
+
"loss": 0.2372,
|
| 601 |
"step": 2050
|
| 602 |
},
|
| 603 |
{
|
| 604 |
"epoch": 1.4679872656526354,
|
| 605 |
+
"grad_norm": 3.7657673358917236,
|
| 606 |
+
"learning_rate": 6.504444444444446e-06,
|
| 607 |
+
"loss": 0.2394,
|
| 608 |
"step": 2075
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 1.4856738592147152,
|
| 612 |
+
"grad_norm": 3.009976387023926,
|
| 613 |
+
"learning_rate": 6.448888888888889e-06,
|
| 614 |
+
"loss": 0.2285,
|
| 615 |
"step": 2100
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 1.5033604527767952,
|
| 619 |
+
"grad_norm": 3.916525363922119,
|
| 620 |
+
"learning_rate": 6.393333333333334e-06,
|
| 621 |
+
"loss": 0.2359,
|
| 622 |
"step": 2125
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 1.5210470463388752,
|
| 626 |
+
"grad_norm": 4.418327808380127,
|
| 627 |
+
"learning_rate": 6.3377777777777786e-06,
|
| 628 |
+
"loss": 0.2499,
|
| 629 |
"step": 2150
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 1.538733639900955,
|
| 633 |
+
"grad_norm": 4.2388105392456055,
|
| 634 |
+
"learning_rate": 6.282222222222223e-06,
|
| 635 |
+
"loss": 0.226,
|
| 636 |
"step": 2175
|
| 637 |
},
|
| 638 |
{
|
| 639 |
"epoch": 1.556420233463035,
|
| 640 |
+
"grad_norm": 3.664215564727783,
|
| 641 |
+
"learning_rate": 6.2266666666666675e-06,
|
| 642 |
+
"loss": 0.2316,
|
| 643 |
"step": 2200
|
| 644 |
},
|
| 645 |
{
|
| 646 |
"epoch": 1.574106827025115,
|
| 647 |
+
"grad_norm": 4.053125858306885,
|
| 648 |
+
"learning_rate": 6.171111111111112e-06,
|
| 649 |
+
"loss": 0.228,
|
| 650 |
"step": 2225
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 1.5917934205871949,
|
| 654 |
+
"grad_norm": 4.452697277069092,
|
| 655 |
+
"learning_rate": 6.1155555555555555e-06,
|
| 656 |
+
"loss": 0.2196,
|
| 657 |
"step": 2250
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 1.6094800141492749,
|
| 661 |
+
"grad_norm": 3.919962167739868,
|
| 662 |
+
"learning_rate": 6.0600000000000004e-06,
|
| 663 |
+
"loss": 0.2495,
|
| 664 |
"step": 2275
|
| 665 |
},
|
| 666 |
{
|
| 667 |
"epoch": 1.627166607711355,
|
| 668 |
+
"grad_norm": 4.33357572555542,
|
| 669 |
+
"learning_rate": 6.004444444444445e-06,
|
| 670 |
+
"loss": 0.2245,
|
| 671 |
"step": 2300
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 1.6448532012734347,
|
| 675 |
+
"grad_norm": 4.634097099304199,
|
| 676 |
+
"learning_rate": 5.948888888888889e-06,
|
| 677 |
+
"loss": 0.2438,
|
| 678 |
"step": 2325
|
| 679 |
},
|
| 680 |
{
|
| 681 |
"epoch": 1.6625397948355147,
|
| 682 |
+
"grad_norm": 4.564420223236084,
|
| 683 |
+
"learning_rate": 5.893333333333334e-06,
|
| 684 |
+
"loss": 0.2047,
|
| 685 |
"step": 2350
|
| 686 |
},
|
| 687 |
{
|
| 688 |
"epoch": 1.6802263883975948,
|
| 689 |
+
"grad_norm": 4.051385879516602,
|
| 690 |
+
"learning_rate": 5.837777777777777e-06,
|
| 691 |
+
"loss": 0.2392,
|
| 692 |
"step": 2375
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 1.6979129819596745,
|
| 696 |
+
"grad_norm": 3.313216209411621,
|
| 697 |
+
"learning_rate": 5.782222222222222e-06,
|
| 698 |
+
"loss": 0.2286,
|
| 699 |
"step": 2400
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 1.7155995755217543,
|
| 703 |
+
"grad_norm": 3.334672451019287,
|
| 704 |
+
"learning_rate": 5.726666666666667e-06,
|
| 705 |
+
"loss": 0.2296,
|
| 706 |
"step": 2425
|
| 707 |
},
|
| 708 |
{
|
| 709 |
"epoch": 1.7332861690838346,
|
| 710 |
+
"grad_norm": 3.519235134124756,
|
| 711 |
+
"learning_rate": 5.671111111111112e-06,
|
| 712 |
+
"loss": 0.2107,
|
| 713 |
"step": 2450
|
| 714 |
},
|
| 715 |
{
|
| 716 |
"epoch": 1.7509727626459144,
|
| 717 |
+
"grad_norm": 4.1753339767456055,
|
| 718 |
+
"learning_rate": 5.615555555555556e-06,
|
| 719 |
+
"loss": 0.2464,
|
| 720 |
"step": 2475
|
| 721 |
},
|
| 722 |
{
|
| 723 |
"epoch": 1.7686593562079942,
|
| 724 |
+
"grad_norm": 3.395427703857422,
|
| 725 |
+
"learning_rate": 5.560000000000001e-06,
|
| 726 |
+
"loss": 0.2323,
|
| 727 |
"step": 2500
|
| 728 |
},
|
| 729 |
{
|
| 730 |
"epoch": 1.7863459497700744,
|
| 731 |
+
"grad_norm": 3.480686902999878,
|
| 732 |
+
"learning_rate": 5.504444444444444e-06,
|
| 733 |
+
"loss": 0.2205,
|
| 734 |
"step": 2525
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 1.8040325433321542,
|
| 738 |
+
"grad_norm": 4.275322914123535,
|
| 739 |
+
"learning_rate": 5.448888888888889e-06,
|
| 740 |
+
"loss": 0.2147,
|
| 741 |
"step": 2550
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 1.821719136894234,
|
| 745 |
+
"grad_norm": 3.6306569576263428,
|
| 746 |
+
"learning_rate": 5.393333333333334e-06,
|
| 747 |
+
"loss": 0.2404,
|
| 748 |
"step": 2575
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 1.839405730456314,
|
| 752 |
+
"grad_norm": 3.426995038986206,
|
| 753 |
+
"learning_rate": 5.337777777777779e-06,
|
| 754 |
+
"loss": 0.2259,
|
| 755 |
"step": 2600
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 1.857092324018394,
|
| 759 |
+
"grad_norm": 4.106579303741455,
|
| 760 |
+
"learning_rate": 5.282222222222223e-06,
|
| 761 |
+
"loss": 0.2043,
|
| 762 |
"step": 2625
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 1.8747789175804739,
|
| 766 |
+
"grad_norm": 3.365471601486206,
|
| 767 |
+
"learning_rate": 5.226666666666667e-06,
|
| 768 |
+
"loss": 0.2179,
|
| 769 |
"step": 2650
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 1.892465511142554,
|
| 773 |
+
"grad_norm": 3.4525575637817383,
|
| 774 |
+
"learning_rate": 5.171111111111111e-06,
|
| 775 |
+
"loss": 0.2185,
|
| 776 |
"step": 2675
|
| 777 |
},
|
| 778 |
{
|
| 779 |
"epoch": 1.910152104704634,
|
| 780 |
+
"grad_norm": 3.5960958003997803,
|
| 781 |
+
"learning_rate": 5.115555555555556e-06,
|
| 782 |
+
"loss": 0.2007,
|
| 783 |
"step": 2700
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 1.9278386982667137,
|
| 787 |
+
"grad_norm": 3.91766619682312,
|
| 788 |
+
"learning_rate": 5.060000000000001e-06,
|
| 789 |
+
"loss": 0.2284,
|
| 790 |
"step": 2725
|
| 791 |
},
|
| 792 |
{
|
| 793 |
"epoch": 1.9455252918287937,
|
| 794 |
+
"grad_norm": 4.662227630615234,
|
| 795 |
+
"learning_rate": 5.004444444444445e-06,
|
| 796 |
+
"loss": 0.2238,
|
| 797 |
"step": 2750
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 1.9632118853908738,
|
| 801 |
+
"grad_norm": 3.227084159851074,
|
| 802 |
+
"learning_rate": 4.94888888888889e-06,
|
| 803 |
+
"loss": 0.2068,
|
| 804 |
"step": 2775
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 1.9808984789529536,
|
| 808 |
+
"grad_norm": 3.7078394889831543,
|
| 809 |
+
"learning_rate": 4.893333333333334e-06,
|
| 810 |
+
"loss": 0.225,
|
| 811 |
"step": 2800
|
| 812 |
},
|
| 813 |
{
|
| 814 |
"epoch": 1.9985850725150336,
|
| 815 |
+
"grad_norm": 3.540781259536743,
|
| 816 |
+
"learning_rate": 4.837777777777778e-06,
|
| 817 |
+
"loss": 0.2115,
|
| 818 |
"step": 2825
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 2.0162716660771136,
|
| 822 |
+
"grad_norm": 2.6315805912017822,
|
| 823 |
+
"learning_rate": 4.7822222222222226e-06,
|
| 824 |
+
"loss": 0.1164,
|
| 825 |
"step": 2850
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 2.0339582596391934,
|
| 829 |
+
"grad_norm": 3.7077784538269043,
|
| 830 |
+
"learning_rate": 4.7266666666666674e-06,
|
| 831 |
+
"loss": 0.1052,
|
| 832 |
"step": 2875
|
| 833 |
},
|
| 834 |
{
|
| 835 |
"epoch": 2.0516448532012737,
|
| 836 |
+
"grad_norm": 2.5548624992370605,
|
| 837 |
+
"learning_rate": 4.6711111111111115e-06,
|
| 838 |
+
"loss": 0.1033,
|
| 839 |
"step": 2900
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 2.0693314467633535,
|
| 843 |
+
"grad_norm": 3.648667097091675,
|
| 844 |
+
"learning_rate": 4.6155555555555555e-06,
|
| 845 |
+
"loss": 0.0954,
|
| 846 |
"step": 2925
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 2.0870180403254333,
|
| 850 |
+
"grad_norm": 1.9462636709213257,
|
| 851 |
+
"learning_rate": 4.56e-06,
|
| 852 |
+
"loss": 0.0907,
|
| 853 |
"step": 2950
|
| 854 |
},
|
| 855 |
{
|
| 856 |
"epoch": 2.104704633887513,
|
| 857 |
+
"grad_norm": 2.353156089782715,
|
| 858 |
+
"learning_rate": 4.504444444444444e-06,
|
| 859 |
+
"loss": 0.0962,
|
| 860 |
"step": 2975
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 2.1223912274495933,
|
| 864 |
+
"grad_norm": 1.8443711996078491,
|
| 865 |
+
"learning_rate": 4.448888888888889e-06,
|
| 866 |
+
"loss": 0.0867,
|
| 867 |
"step": 3000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 2.1223912274495933,
|
| 871 |
+
"eval_loss": 0.25060394406318665,
|
| 872 |
+
"eval_runtime": 4656.2877,
|
| 873 |
+
"eval_samples_per_second": 2.311,
|
| 874 |
+
"eval_steps_per_second": 0.145,
|
| 875 |
+
"eval_wer": 0.15110889462567473,
|
| 876 |
"step": 3000
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 2.140077821011673,
|
| 880 |
+
"grad_norm": 2.377577781677246,
|
| 881 |
+
"learning_rate": 4.393333333333334e-06,
|
| 882 |
+
"loss": 0.0825,
|
| 883 |
"step": 3025
|
| 884 |
},
|
| 885 |
{
|
| 886 |
"epoch": 2.157764414573753,
|
| 887 |
+
"grad_norm": 2.523731231689453,
|
| 888 |
+
"learning_rate": 4.337777777777778e-06,
|
| 889 |
+
"loss": 0.1005,
|
| 890 |
"step": 3050
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 2.175451008135833,
|
| 894 |
+
"grad_norm": 3.5590755939483643,
|
| 895 |
+
"learning_rate": 4.282222222222222e-06,
|
| 896 |
+
"loss": 0.1085,
|
| 897 |
"step": 3075
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 2.193137601697913,
|
| 901 |
+
"grad_norm": 2.7134900093078613,
|
| 902 |
+
"learning_rate": 4.226666666666667e-06,
|
| 903 |
+
"loss": 0.0957,
|
| 904 |
"step": 3100
|
| 905 |
},
|
| 906 |
{
|
| 907 |
"epoch": 2.2108241952599927,
|
| 908 |
+
"grad_norm": 3.198664903640747,
|
| 909 |
+
"learning_rate": 4.171111111111111e-06,
|
| 910 |
+
"loss": 0.0898,
|
| 911 |
"step": 3125
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 2.228510788822073,
|
| 915 |
+
"grad_norm": 2.207460403442383,
|
| 916 |
+
"learning_rate": 4.115555555555556e-06,
|
| 917 |
+
"loss": 0.0922,
|
| 918 |
"step": 3150
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 2.246197382384153,
|
| 922 |
+
"grad_norm": 3.6293723583221436,
|
| 923 |
+
"learning_rate": 4.060000000000001e-06,
|
| 924 |
+
"loss": 0.0847,
|
| 925 |
"step": 3175
|
| 926 |
},
|
| 927 |
{
|
| 928 |
"epoch": 2.2638839759462326,
|
| 929 |
+
"grad_norm": 2.8405864238739014,
|
| 930 |
+
"learning_rate": 4.004444444444445e-06,
|
| 931 |
+
"loss": 0.097,
|
| 932 |
"step": 3200
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 2.281570569508313,
|
| 936 |
+
"grad_norm": 3.193572521209717,
|
| 937 |
+
"learning_rate": 3.948888888888889e-06,
|
| 938 |
+
"loss": 0.0997,
|
| 939 |
"step": 3225
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 2.2992571630703926,
|
| 943 |
+
"grad_norm": 2.4953181743621826,
|
| 944 |
+
"learning_rate": 3.893333333333333e-06,
|
| 945 |
+
"loss": 0.0947,
|
| 946 |
"step": 3250
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 2.3169437566324724,
|
| 950 |
+
"grad_norm": 2.287153959274292,
|
| 951 |
+
"learning_rate": 3.837777777777778e-06,
|
| 952 |
+
"loss": 0.0829,
|
| 953 |
"step": 3275
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 2.3346303501945527,
|
| 957 |
+
"grad_norm": 3.0835964679718018,
|
| 958 |
+
"learning_rate": 3.782222222222223e-06,
|
| 959 |
+
"loss": 0.0986,
|
| 960 |
"step": 3300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 2.3523169437566325,
|
| 964 |
+
"grad_norm": 3.700576066970825,
|
| 965 |
+
"learning_rate": 3.726666666666667e-06,
|
| 966 |
+
"loss": 0.0938,
|
| 967 |
"step": 3325
|
| 968 |
},
|
| 969 |
{
|
| 970 |
"epoch": 2.3700035373187123,
|
| 971 |
+
"grad_norm": 2.685814380645752,
|
| 972 |
+
"learning_rate": 3.6711111111111113e-06,
|
| 973 |
+
"loss": 0.0912,
|
| 974 |
"step": 3350
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 2.3876901308807925,
|
| 978 |
+
"grad_norm": 2.80168080329895,
|
| 979 |
+
"learning_rate": 3.615555555555556e-06,
|
| 980 |
+
"loss": 0.1046,
|
| 981 |
"step": 3375
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 2.4053767244428723,
|
| 985 |
+
"grad_norm": 1.993931531906128,
|
| 986 |
+
"learning_rate": 3.5600000000000002e-06,
|
| 987 |
+
"loss": 0.0956,
|
| 988 |
"step": 3400
|
| 989 |
},
|
| 990 |
{
|
| 991 |
"epoch": 2.423063318004952,
|
| 992 |
+
"grad_norm": 2.2069132328033447,
|
| 993 |
+
"learning_rate": 3.5044444444444447e-06,
|
| 994 |
+
"loss": 0.0897,
|
| 995 |
"step": 3425
|
| 996 |
},
|
| 997 |
{
|
| 998 |
"epoch": 2.4407499115670324,
|
| 999 |
+
"grad_norm": 1.9554574489593506,
|
| 1000 |
+
"learning_rate": 3.4488888888888896e-06,
|
| 1001 |
+
"loss": 0.0916,
|
| 1002 |
"step": 3450
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 2.458436505129112,
|
| 1006 |
+
"grad_norm": 2.17236590385437,
|
| 1007 |
+
"learning_rate": 3.3933333333333336e-06,
|
| 1008 |
+
"loss": 0.0914,
|
| 1009 |
"step": 3475
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 2.476123098691192,
|
| 1013 |
+
"grad_norm": 1.3875476121902466,
|
| 1014 |
+
"learning_rate": 3.337777777777778e-06,
|
| 1015 |
+
"loss": 0.0841,
|
| 1016 |
"step": 3500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 2.493809692253272,
|
| 1020 |
+
"grad_norm": 2.773275852203369,
|
| 1021 |
+
"learning_rate": 3.282222222222223e-06,
|
| 1022 |
+
"loss": 0.0776,
|
| 1023 |
"step": 3525
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 2.511496285815352,
|
| 1027 |
+
"grad_norm": 2.6522185802459717,
|
| 1028 |
+
"learning_rate": 3.226666666666667e-06,
|
| 1029 |
+
"loss": 0.1017,
|
| 1030 |
"step": 3550
|
| 1031 |
},
|
| 1032 |
{
|
| 1033 |
"epoch": 2.529182879377432,
|
| 1034 |
+
"grad_norm": 2.5875155925750732,
|
| 1035 |
+
"learning_rate": 3.1711111111111114e-06,
|
| 1036 |
+
"loss": 0.0873,
|
| 1037 |
"step": 3575
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 2.5468694729395116,
|
| 1041 |
+
"grad_norm": 2.877279281616211,
|
| 1042 |
+
"learning_rate": 3.1155555555555555e-06,
|
| 1043 |
+
"loss": 0.1071,
|
| 1044 |
"step": 3600
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 2.564556066501592,
|
| 1048 |
+
"grad_norm": 2.6924633979797363,
|
| 1049 |
+
"learning_rate": 3.0600000000000003e-06,
|
| 1050 |
+
"loss": 0.0885,
|
| 1051 |
"step": 3625
|
| 1052 |
},
|
| 1053 |
{
|
| 1054 |
"epoch": 2.5822426600636716,
|
| 1055 |
+
"grad_norm": 3.1496784687042236,
|
| 1056 |
+
"learning_rate": 3.004444444444445e-06,
|
| 1057 |
+
"loss": 0.0875,
|
| 1058 |
"step": 3650
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 2.599929253625752,
|
| 1062 |
+
"grad_norm": 2.8723342418670654,
|
| 1063 |
+
"learning_rate": 2.948888888888889e-06,
|
| 1064 |
+
"loss": 0.0849,
|
| 1065 |
"step": 3675
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 2.6176158471878317,
|
| 1069 |
+
"grad_norm": 2.9840595722198486,
|
| 1070 |
+
"learning_rate": 2.8933333333333337e-06,
|
| 1071 |
+
"loss": 0.085,
|
| 1072 |
"step": 3700
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 2.6353024407499115,
|
| 1076 |
+
"grad_norm": 2.006192684173584,
|
| 1077 |
+
"learning_rate": 2.837777777777778e-06,
|
| 1078 |
+
"loss": 0.082,
|
| 1079 |
"step": 3725
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 2.6529890343119913,
|
| 1083 |
+
"grad_norm": 2.020374298095703,
|
| 1084 |
+
"learning_rate": 2.7822222222222222e-06,
|
| 1085 |
+
"loss": 0.083,
|
| 1086 |
"step": 3750
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 2.6706756278740715,
|
| 1090 |
+
"grad_norm": 2.8968160152435303,
|
| 1091 |
+
"learning_rate": 2.726666666666667e-06,
|
| 1092 |
+
"loss": 0.0896,
|
| 1093 |
"step": 3775
|
| 1094 |
},
|
| 1095 |
{
|
| 1096 |
"epoch": 2.6883622214361513,
|
| 1097 |
+
"grad_norm": 2.3445401191711426,
|
| 1098 |
+
"learning_rate": 2.6711111111111116e-06,
|
| 1099 |
+
"loss": 0.0956,
|
| 1100 |
"step": 3800
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 2.7060488149982316,
|
| 1104 |
+
"grad_norm": 2.18906307220459,
|
| 1105 |
+
"learning_rate": 2.6155555555555556e-06,
|
| 1106 |
+
"loss": 0.0968,
|
| 1107 |
"step": 3825
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 2.7237354085603114,
|
| 1111 |
+
"grad_norm": 3.4939792156219482,
|
| 1112 |
+
"learning_rate": 2.56e-06,
|
| 1113 |
+
"loss": 0.0807,
|
| 1114 |
"step": 3850
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
"epoch": 2.741422002122391,
|
| 1118 |
+
"grad_norm": 2.769930839538574,
|
| 1119 |
+
"learning_rate": 2.504444444444445e-06,
|
| 1120 |
+
"loss": 0.0807,
|
| 1121 |
"step": 3875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 2.759108595684471,
|
| 1125 |
+
"grad_norm": 2.675159215927124,
|
| 1126 |
+
"learning_rate": 2.448888888888889e-06,
|
| 1127 |
+
"loss": 0.0852,
|
| 1128 |
"step": 3900
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 2.776795189246551,
|
| 1132 |
+
"grad_norm": 1.9144662618637085,
|
| 1133 |
+
"learning_rate": 2.3933333333333334e-06,
|
| 1134 |
+
"loss": 0.0827,
|
| 1135 |
"step": 3925
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
"epoch": 2.794481782808631,
|
| 1139 |
+
"grad_norm": 2.1294009685516357,
|
| 1140 |
+
"learning_rate": 2.337777777777778e-06,
|
| 1141 |
+
"loss": 0.0723,
|
| 1142 |
"step": 3950
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 2.8121683763707113,
|
| 1146 |
+
"grad_norm": 2.5433287620544434,
|
| 1147 |
+
"learning_rate": 2.2822222222222223e-06,
|
| 1148 |
+
"loss": 0.0895,
|
| 1149 |
"step": 3975
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 2.829854969932791,
|
| 1153 |
+
"grad_norm": 2.70267391204834,
|
| 1154 |
+
"learning_rate": 2.226666666666667e-06,
|
| 1155 |
+
"loss": 0.0973,
|
| 1156 |
"step": 4000
|
| 1157 |
},
|
| 1158 |
{
|
| 1159 |
"epoch": 2.829854969932791,
|
| 1160 |
+
"eval_loss": 0.24439197778701782,
|
| 1161 |
+
"eval_runtime": 4587.2812,
|
| 1162 |
+
"eval_samples_per_second": 2.346,
|
| 1163 |
+
"eval_steps_per_second": 0.147,
|
| 1164 |
+
"eval_wer": 0.14902605022295237,
|
| 1165 |
"step": 4000
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
"epoch": 2.847541563494871,
|
| 1169 |
+
"grad_norm": 2.82698392868042,
|
| 1170 |
+
"learning_rate": 2.1711111111111113e-06,
|
| 1171 |
+
"loss": 0.087,
|
| 1172 |
"step": 4025
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 2.8652281570569507,
|
| 1176 |
+
"grad_norm": 2.4205260276794434,
|
| 1177 |
+
"learning_rate": 2.1155555555555557e-06,
|
| 1178 |
+
"loss": 0.073,
|
| 1179 |
"step": 4050
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 2.882914750619031,
|
| 1183 |
+
"grad_norm": 2.8398914337158203,
|
| 1184 |
+
"learning_rate": 2.06e-06,
|
| 1185 |
+
"loss": 0.0829,
|
| 1186 |
"step": 4075
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 2.9006013441811107,
|
| 1190 |
+
"grad_norm": 2.544459581375122,
|
| 1191 |
+
"learning_rate": 2.0044444444444446e-06,
|
| 1192 |
+
"loss": 0.0944,
|
| 1193 |
"step": 4100
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
"epoch": 2.9182879377431905,
|
| 1197 |
+
"grad_norm": 3.273761034011841,
|
| 1198 |
+
"learning_rate": 1.948888888888889e-06,
|
| 1199 |
+
"loss": 0.0894,
|
| 1200 |
"step": 4125
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 2.9359745313052708,
|
| 1204 |
+
"grad_norm": 2.9628796577453613,
|
| 1205 |
+
"learning_rate": 1.8933333333333333e-06,
|
| 1206 |
+
"loss": 0.0779,
|
| 1207 |
"step": 4150
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
"epoch": 2.9536611248673506,
|
| 1211 |
+
"grad_norm": 2.375798225402832,
|
| 1212 |
+
"learning_rate": 1.837777777777778e-06,
|
| 1213 |
+
"loss": 0.0887,
|
| 1214 |
"step": 4175
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 2.9713477184294304,
|
| 1218 |
+
"grad_norm": 3.536142587661743,
|
| 1219 |
+
"learning_rate": 1.7822222222222225e-06,
|
| 1220 |
+
"loss": 0.0802,
|
| 1221 |
"step": 4200
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 2.9890343119915106,
|
| 1225 |
+
"grad_norm": 2.959638833999634,
|
| 1226 |
+
"learning_rate": 1.7266666666666667e-06,
|
| 1227 |
+
"loss": 0.0736,
|
| 1228 |
"step": 4225
|
| 1229 |
},
|
| 1230 |
{
|
| 1231 |
"epoch": 3.0067209055535904,
|
| 1232 |
+
"grad_norm": 1.6278612613677979,
|
| 1233 |
+
"learning_rate": 1.6711111111111112e-06,
|
| 1234 |
+
"loss": 0.0726,
|
| 1235 |
"step": 4250
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 3.02440749911567,
|
| 1239 |
+
"grad_norm": 2.070136308670044,
|
| 1240 |
+
"learning_rate": 1.6155555555555559e-06,
|
| 1241 |
+
"loss": 0.0342,
|
| 1242 |
"step": 4275
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 3.0420940926777504,
|
| 1246 |
+
"grad_norm": 1.3914729356765747,
|
| 1247 |
+
"learning_rate": 1.56e-06,
|
| 1248 |
+
"loss": 0.0358,
|
| 1249 |
"step": 4300
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 3.0597806862398302,
|
| 1253 |
+
"grad_norm": 2.264988422393799,
|
| 1254 |
+
"learning_rate": 1.5044444444444446e-06,
|
| 1255 |
+
"loss": 0.0333,
|
| 1256 |
"step": 4325
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 3.07746727980191,
|
| 1260 |
+
"grad_norm": 2.115565538406372,
|
| 1261 |
+
"learning_rate": 1.4488888888888892e-06,
|
| 1262 |
+
"loss": 0.0441,
|
| 1263 |
"step": 4350
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 3.0951538733639903,
|
| 1267 |
+
"grad_norm": 1.555413007736206,
|
| 1268 |
+
"learning_rate": 1.3933333333333335e-06,
|
| 1269 |
+
"loss": 0.0333,
|
| 1270 |
"step": 4375
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 3.11284046692607,
|
| 1274 |
+
"grad_norm": 1.6975215673446655,
|
| 1275 |
+
"learning_rate": 1.337777777777778e-06,
|
| 1276 |
+
"loss": 0.0347,
|
| 1277 |
"step": 4400
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 3.13052706048815,
|
| 1281 |
+
"grad_norm": 1.0206265449523926,
|
| 1282 |
+
"learning_rate": 1.2822222222222222e-06,
|
| 1283 |
+
"loss": 0.0288,
|
| 1284 |
"step": 4425
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 3.14821365405023,
|
| 1288 |
+
"grad_norm": 0.8475884199142456,
|
| 1289 |
+
"learning_rate": 1.2266666666666666e-06,
|
| 1290 |
+
"loss": 0.0282,
|
| 1291 |
"step": 4450
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 3.16590024761231,
|
| 1295 |
+
"grad_norm": 1.2225841283798218,
|
| 1296 |
+
"learning_rate": 1.171111111111111e-06,
|
| 1297 |
+
"loss": 0.0419,
|
| 1298 |
"step": 4475
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 3.1835868411743897,
|
| 1302 |
+
"grad_norm": 1.5350502729415894,
|
| 1303 |
+
"learning_rate": 1.1155555555555558e-06,
|
| 1304 |
+
"loss": 0.033,
|
| 1305 |
"step": 4500
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 3.2012734347364695,
|
| 1309 |
+
"grad_norm": 1.7746856212615967,
|
| 1310 |
+
"learning_rate": 1.06e-06,
|
| 1311 |
+
"loss": 0.0314,
|
| 1312 |
"step": 4525
|
| 1313 |
},
|
| 1314 |
{
|
| 1315 |
"epoch": 3.2189600282985498,
|
| 1316 |
+
"grad_norm": 0.9352214932441711,
|
| 1317 |
+
"learning_rate": 1.0044444444444445e-06,
|
| 1318 |
+
"loss": 0.0309,
|
| 1319 |
"step": 4550
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 3.2366466218606296,
|
| 1323 |
+
"grad_norm": 1.1960421800613403,
|
| 1324 |
+
"learning_rate": 9.488888888888889e-07,
|
| 1325 |
+
"loss": 0.0432,
|
| 1326 |
"step": 4575
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 3.25433321542271,
|
| 1330 |
+
"grad_norm": 2.057264566421509,
|
| 1331 |
+
"learning_rate": 8.933333333333334e-07,
|
| 1332 |
+
"loss": 0.0361,
|
| 1333 |
"step": 4600
|
| 1334 |
},
|
| 1335 |
{
|
| 1336 |
"epoch": 3.2720198089847896,
|
| 1337 |
+
"grad_norm": 1.3137741088867188,
|
| 1338 |
+
"learning_rate": 8.37777777777778e-07,
|
| 1339 |
+
"loss": 0.0274,
|
| 1340 |
"step": 4625
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 3.2897064025468694,
|
| 1344 |
+
"grad_norm": 2.9971253871917725,
|
| 1345 |
+
"learning_rate": 7.822222222222223e-07,
|
| 1346 |
+
"loss": 0.0334,
|
| 1347 |
"step": 4650
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 3.307392996108949,
|
| 1351 |
+
"grad_norm": 1.1959739923477173,
|
| 1352 |
+
"learning_rate": 7.266666666666668e-07,
|
| 1353 |
+
"loss": 0.0321,
|
| 1354 |
"step": 4675
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
"epoch": 3.3250795896710295,
|
| 1358 |
+
"grad_norm": 1.4771987199783325,
|
| 1359 |
+
"learning_rate": 6.711111111111111e-07,
|
| 1360 |
+
"loss": 0.0304,
|
| 1361 |
"step": 4700
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 3.3427661832331093,
|
| 1365 |
+
"grad_norm": 1.407965898513794,
|
| 1366 |
+
"learning_rate": 6.155555555555556e-07,
|
| 1367 |
+
"loss": 0.0358,
|
| 1368 |
"step": 4725
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 3.360452776795189,
|
| 1372 |
+
"grad_norm": 1.603348970413208,
|
| 1373 |
+
"learning_rate": 5.6e-07,
|
| 1374 |
+
"loss": 0.0253,
|
| 1375 |
"step": 4750
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 3.3781393703572693,
|
| 1379 |
+
"grad_norm": 0.6914874911308289,
|
| 1380 |
+
"learning_rate": 5.044444444444445e-07,
|
| 1381 |
+
"loss": 0.0314,
|
| 1382 |
"step": 4775
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 3.395825963919349,
|
| 1386 |
+
"grad_norm": 0.9917079210281372,
|
| 1387 |
+
"learning_rate": 4.488888888888889e-07,
|
| 1388 |
+
"loss": 0.0366,
|
| 1389 |
"step": 4800
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 3.413512557481429,
|
| 1393 |
+
"grad_norm": 1.6925675868988037,
|
| 1394 |
+
"learning_rate": 3.9333333333333336e-07,
|
| 1395 |
+
"loss": 0.0325,
|
| 1396 |
"step": 4825
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 3.431199151043509,
|
| 1400 |
+
"grad_norm": 1.986444115638733,
|
| 1401 |
+
"learning_rate": 3.3777777777777777e-07,
|
| 1402 |
+
"loss": 0.0281,
|
| 1403 |
"step": 4850
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 3.448885744605589,
|
| 1407 |
+
"grad_norm": 0.704007089138031,
|
| 1408 |
+
"learning_rate": 2.822222222222222e-07,
|
| 1409 |
+
"loss": 0.0313,
|
| 1410 |
"step": 4875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 3.4665723381676687,
|
| 1414 |
+
"grad_norm": 1.0404950380325317,
|
| 1415 |
+
"learning_rate": 2.266666666666667e-07,
|
| 1416 |
+
"loss": 0.0311,
|
| 1417 |
"step": 4900
|
| 1418 |
},
|
| 1419 |
{
|
| 1420 |
"epoch": 3.484258931729749,
|
| 1421 |
+
"grad_norm": 1.7548632621765137,
|
| 1422 |
+
"learning_rate": 1.7111111111111114e-07,
|
| 1423 |
+
"loss": 0.029,
|
| 1424 |
"step": 4925
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 3.501945525291829,
|
| 1428 |
+
"grad_norm": 2.3233606815338135,
|
| 1429 |
+
"learning_rate": 1.1555555555555556e-07,
|
| 1430 |
+
"loss": 0.0364,
|
| 1431 |
"step": 4950
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 3.5196321188539086,
|
| 1435 |
+
"grad_norm": 1.7240501642227173,
|
| 1436 |
+
"learning_rate": 6.000000000000001e-08,
|
| 1437 |
+
"loss": 0.035,
|
| 1438 |
"step": 4975
|
| 1439 |
},
|
| 1440 |
{
|
| 1441 |
"epoch": 3.537318712415989,
|
| 1442 |
+
"grad_norm": 0.9116590023040771,
|
| 1443 |
+
"learning_rate": 4.444444444444445e-09,
|
| 1444 |
+
"loss": 0.0303,
|
| 1445 |
"step": 5000
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 3.537318712415989,
|
| 1449 |
+
"eval_loss": 0.27435705065727234,
|
| 1450 |
+
"eval_runtime": 4637.7857,
|
| 1451 |
+
"eval_samples_per_second": 2.321,
|
| 1452 |
+
"eval_steps_per_second": 0.145,
|
| 1453 |
+
"eval_wer": 0.14744191504341703,
|
| 1454 |
"step": 5000
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
"epoch": 3.537318712415989,
|
| 1458 |
"step": 5000,
|
| 1459 |
"total_flos": 5.435079965953229e+20,
|
| 1460 |
+
"train_loss": 0.26693406739234926,
|
| 1461 |
+
"train_runtime": 74036.7422,
|
| 1462 |
+
"train_samples_per_second": 2.161,
|
| 1463 |
+
"train_steps_per_second": 0.068
|
| 1464 |
}
|
| 1465 |
],
|
| 1466 |
"logging_steps": 25,
|