End of training
Browse files- README.md +7 -7
- all_results.json +11 -11
- eval_results.json +6 -6
- runs/Aug20_11-34-00_DITEC2014063010/events.out.tfevents.1724865406.DITEC2014063010.27248.1 +3 -0
- train_results.json +6 -6
- trainer_state.json +1869 -241
README.md
CHANGED
|
@@ -5,7 +5,7 @@ base_model: openai/whisper-large-v3
|
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
| 7 |
datasets:
|
| 8 |
-
- common_voice_18_0
|
| 9 |
metrics:
|
| 10 |
- wer
|
| 11 |
model-index:
|
|
@@ -15,15 +15,15 @@ model-index:
|
|
| 15 |
name: Automatic Speech Recognition
|
| 16 |
type: automatic-speech-recognition
|
| 17 |
dataset:
|
| 18 |
-
name: common_voice_18_0
|
| 19 |
-
type: common_voice_18_0
|
| 20 |
config: pt
|
| 21 |
split: None
|
| 22 |
args: pt
|
| 23 |
metrics:
|
| 24 |
- name: Wer
|
| 25 |
type: wer
|
| 26 |
-
value: 0.
|
| 27 |
---
|
| 28 |
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -31,10 +31,10 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 31 |
|
| 32 |
# whisper-large-v3-pt-3000h-3
|
| 33 |
|
| 34 |
-
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the common_voice_18_0 dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
-
- Loss: 0.
|
| 37 |
-
- Wer: 0.
|
| 38 |
|
| 39 |
## Model description
|
| 40 |
|
|
|
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
| 7 |
datasets:
|
| 8 |
+
- fsicoli/common_voice_18_0
|
| 9 |
metrics:
|
| 10 |
- wer
|
| 11 |
model-index:
|
|
|
|
| 15 |
name: Automatic Speech Recognition
|
| 16 |
type: automatic-speech-recognition
|
| 17 |
dataset:
|
| 18 |
+
name: fsicoli/common_voice_18_0 pt
|
| 19 |
+
type: fsicoli/common_voice_18_0
|
| 20 |
config: pt
|
| 21 |
split: None
|
| 22 |
args: pt
|
| 23 |
metrics:
|
| 24 |
- name: Wer
|
| 25 |
type: wer
|
| 26 |
+
value: 0.10736707238949392
|
| 27 |
---
|
| 28 |
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 31 |
|
| 32 |
# whisper-large-v3-pt-3000h-3
|
| 33 |
|
| 34 |
+
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the fsicoli/common_voice_18_0 pt dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
+
- Loss: 0.1501
|
| 37 |
+
- Wer: 0.1074
|
| 38 |
|
| 39 |
## Model description
|
| 40 |
|
all_results.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 9494,
|
| 6 |
-
"eval_samples_per_second": 0.
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
-
"total_flos":
|
| 10 |
-
"train_loss": 0.
|
| 11 |
-
"train_runtime":
|
| 12 |
"train_samples": 22116,
|
| 13 |
-
"train_samples_per_second":
|
| 14 |
-
"train_steps_per_second":
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.99638336347197,
|
| 3 |
+
"eval_loss": 0.1500573754310608,
|
| 4 |
+
"eval_runtime": 50421.184,
|
| 5 |
"eval_samples": 9494,
|
| 6 |
+
"eval_samples_per_second": 0.188,
|
| 7 |
+
"eval_steps_per_second": 0.024,
|
| 8 |
+
"eval_wer": 0.10736707238949392,
|
| 9 |
+
"total_flos": 7.51132515607511e+20,
|
| 10 |
+
"train_loss": 0.05691683961323822,
|
| 11 |
+
"train_runtime": 650193.1322,
|
| 12 |
"train_samples": 22116,
|
| 13 |
+
"train_samples_per_second": 0.34,
|
| 14 |
+
"train_steps_per_second": 0.011
|
| 15 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 9494,
|
| 6 |
-
"eval_samples_per_second": 0.
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.99638336347197,
|
| 3 |
+
"eval_loss": 0.1500573754310608,
|
| 4 |
+
"eval_runtime": 50421.184,
|
| 5 |
"eval_samples": 9494,
|
| 6 |
+
"eval_samples_per_second": 0.188,
|
| 7 |
+
"eval_steps_per_second": 0.024,
|
| 8 |
+
"eval_wer": 0.10736707238949392
|
| 9 |
}
|
runs/Aug20_11-34-00_DITEC2014063010/events.out.tfevents.1724865406.DITEC2014063010.27248.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1e41817bbf37be9c0b8489bfb806165aa585307cc8379119344925e309d89fc8
|
| 3 |
+
size 406
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 22116,
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.99638336347197,
|
| 3 |
+
"total_flos": 7.51132515607511e+20,
|
| 4 |
+
"train_loss": 0.05691683961323822,
|
| 5 |
+
"train_runtime": 650193.1322,
|
| 6 |
"train_samples": 22116,
|
| 7 |
+
"train_samples_per_second": 0.34,
|
| 8 |
+
"train_steps_per_second": 0.011
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,421 +1,2049 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint":
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
-
"epoch": 0.
|
| 13 |
-
"grad_norm": 5.
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"loss": 0.
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch": 0.
|
| 20 |
-
"grad_norm":
|
| 21 |
-
"learning_rate":
|
| 22 |
-
"loss": 1.
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm":
|
| 28 |
-
"learning_rate":
|
| 29 |
-
"loss": 0.
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm":
|
| 35 |
-
"learning_rate":
|
| 36 |
-
"loss": 0.
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"grad_norm":
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"grad_norm":
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm": 2.
|
| 56 |
-
"learning_rate":
|
| 57 |
-
"loss": 0.
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm":
|
| 63 |
-
"learning_rate":
|
| 64 |
-
"loss": 0.
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm": 2.
|
| 70 |
-
"learning_rate":
|
| 71 |
-
"loss": 0.
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm":
|
| 77 |
-
"learning_rate":
|
| 78 |
-
"loss": 0.
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm":
|
| 84 |
-
"learning_rate":
|
| 85 |
-
"loss": 0.
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm":
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss": 0.
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm": 2.
|
| 98 |
-
"learning_rate":
|
| 99 |
-
"loss": 0.
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm":
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm":
|
| 112 |
-
"learning_rate":
|
| 113 |
-
"loss": 0.
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"loss": 0.
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"grad_norm": 2.
|
| 126 |
-
"learning_rate":
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm": 6.
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"grad_norm": 2.
|
| 140 |
-
"learning_rate":
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch": 0.
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"grad_norm": 2.
|
| 154 |
-
"learning_rate":
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
-
"epoch": 0.
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate":
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
-
"epoch": 0.
|
| 167 |
-
"grad_norm":
|
| 168 |
-
"learning_rate":
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm":
|
| 175 |
-
"learning_rate":
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 0.
|
| 181 |
-
"grad_norm": 2.
|
| 182 |
-
"learning_rate":
|
| 183 |
-
"loss": 0.
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
-
"epoch": 0.
|
| 188 |
-
"grad_norm": 5.
|
| 189 |
-
"learning_rate":
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch": 0.
|
| 195 |
-
"grad_norm": 3.
|
| 196 |
-
"learning_rate":
|
| 197 |
-
"loss": 0.
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch": 0.
|
| 202 |
-
"eval_loss": 0.
|
| 203 |
-
"eval_runtime":
|
| 204 |
-
"eval_samples_per_second": 0.
|
| 205 |
-
"eval_steps_per_second": 0.
|
| 206 |
-
"eval_wer": 0.
|
| 207 |
"step": 691
|
| 208 |
},
|
| 209 |
{
|
| 210 |
-
"epoch": 1.
|
| 211 |
-
"grad_norm":
|
| 212 |
-
"learning_rate":
|
| 213 |
-
"loss": 0.
|
| 214 |
"step": 700
|
| 215 |
},
|
| 216 |
{
|
| 217 |
-
"epoch": 1.
|
| 218 |
-
"grad_norm": 2.
|
| 219 |
-
"learning_rate":
|
| 220 |
-
"loss": 0.
|
| 221 |
"step": 725
|
| 222 |
},
|
| 223 |
{
|
| 224 |
-
"epoch": 1.
|
| 225 |
-
"grad_norm": 1.
|
| 226 |
-
"learning_rate":
|
| 227 |
-
"loss": 0.
|
| 228 |
"step": 750
|
| 229 |
},
|
| 230 |
{
|
| 231 |
-
"epoch": 1.
|
| 232 |
-
"grad_norm":
|
| 233 |
-
"learning_rate":
|
| 234 |
-
"loss": 0.
|
| 235 |
"step": 775
|
| 236 |
},
|
| 237 |
{
|
| 238 |
-
"epoch": 1.
|
| 239 |
-
"grad_norm":
|
| 240 |
-
"learning_rate":
|
| 241 |
-
"loss": 0.
|
| 242 |
"step": 800
|
| 243 |
},
|
| 244 |
{
|
| 245 |
-
"epoch": 1.
|
| 246 |
-
"grad_norm":
|
| 247 |
-
"learning_rate":
|
| 248 |
-
"loss": 0.
|
| 249 |
"step": 825
|
| 250 |
},
|
| 251 |
{
|
| 252 |
-
"epoch": 1.
|
| 253 |
-
"grad_norm": 1.
|
| 254 |
-
"learning_rate":
|
| 255 |
-
"loss": 0.
|
| 256 |
"step": 850
|
| 257 |
},
|
| 258 |
{
|
| 259 |
-
"epoch": 1.
|
| 260 |
-
"grad_norm":
|
| 261 |
-
"learning_rate":
|
| 262 |
-
"loss": 0.
|
| 263 |
"step": 875
|
| 264 |
},
|
| 265 |
{
|
| 266 |
-
"epoch": 1.
|
| 267 |
-
"grad_norm": 1.
|
| 268 |
-
"learning_rate":
|
| 269 |
-
"loss": 0.
|
| 270 |
"step": 900
|
| 271 |
},
|
| 272 |
{
|
| 273 |
-
"epoch": 1.
|
| 274 |
-
"grad_norm": 2.
|
| 275 |
-
"learning_rate":
|
| 276 |
-
"loss": 0.
|
| 277 |
"step": 925
|
| 278 |
},
|
| 279 |
{
|
| 280 |
-
"epoch": 1.
|
| 281 |
-
"grad_norm":
|
| 282 |
-
"learning_rate":
|
| 283 |
-
"loss": 0.
|
| 284 |
"step": 950
|
| 285 |
},
|
| 286 |
{
|
| 287 |
-
"epoch": 1.
|
| 288 |
-
"grad_norm":
|
| 289 |
-
"learning_rate":
|
| 290 |
-
"loss": 0.
|
| 291 |
"step": 975
|
| 292 |
},
|
| 293 |
{
|
| 294 |
-
"epoch": 1.
|
| 295 |
-
"grad_norm": 2.
|
| 296 |
-
"learning_rate":
|
| 297 |
-
"loss": 0.
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
-
"epoch": 1.
|
| 302 |
-
"grad_norm": 2.
|
| 303 |
-
"learning_rate": 9.
|
| 304 |
-
"loss": 0.
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
-
"epoch": 1.
|
| 309 |
-
"grad_norm":
|
| 310 |
-
"learning_rate":
|
| 311 |
-
"loss": 0.
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
-
"epoch": 1.
|
| 316 |
-
"grad_norm": 2.
|
| 317 |
-
"learning_rate":
|
| 318 |
-
"loss": 0.
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
-
"epoch": 1.
|
| 323 |
-
"grad_norm": 1.
|
| 324 |
-
"learning_rate":
|
| 325 |
-
"loss": 0.
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
-
"epoch": 1.
|
| 330 |
-
"grad_norm":
|
| 331 |
-
"learning_rate":
|
| 332 |
-
"loss": 0.
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
-
"epoch": 1.
|
| 337 |
-
"grad_norm": 1.
|
| 338 |
-
"learning_rate":
|
| 339 |
-
"loss": 0.
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
-
"epoch": 1.
|
| 344 |
-
"grad_norm": 3.
|
| 345 |
-
"learning_rate":
|
| 346 |
-
"loss": 0.
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
-
"epoch": 1.
|
| 351 |
-
"grad_norm":
|
| 352 |
-
"learning_rate":
|
| 353 |
-
"loss": 0.
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
-
"epoch": 1.
|
| 358 |
-
"grad_norm": 2.
|
| 359 |
-
"learning_rate":
|
| 360 |
-
"loss": 0.
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
-
"epoch": 1.
|
| 365 |
-
"grad_norm": 1.
|
| 366 |
-
"learning_rate":
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
-
"epoch": 1.
|
| 372 |
-
"grad_norm":
|
| 373 |
-
"learning_rate":
|
| 374 |
-
"loss": 0.
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
-
"epoch": 1.
|
| 379 |
-
"grad_norm":
|
| 380 |
-
"learning_rate":
|
| 381 |
-
"loss": 0.
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
-
"epoch": 1.
|
| 386 |
-
"grad_norm":
|
| 387 |
-
"learning_rate":
|
| 388 |
-
"loss": 0.
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
-
"epoch": 1.
|
| 393 |
-
"grad_norm": 1.
|
| 394 |
-
"learning_rate":
|
| 395 |
-
"loss": 0.
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
-
"epoch": 1.
|
| 400 |
-
"grad_norm":
|
| 401 |
-
"learning_rate":
|
| 402 |
-
"loss": 0.
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
-
"epoch": 1.
|
| 407 |
-
"
|
| 408 |
-
"
|
| 409 |
-
"
|
| 410 |
-
"
|
| 411 |
-
"
|
| 412 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 413 |
}
|
| 414 |
],
|
| 415 |
"logging_steps": 25,
|
| 416 |
-
"max_steps":
|
| 417 |
"num_input_tokens_seen": 0,
|
| 418 |
-
"num_train_epochs":
|
| 419 |
"save_steps": 500,
|
| 420 |
"stateful_callbacks": {
|
| 421 |
"TrainerControl": {
|
|
@@ -429,8 +2057,8 @@
|
|
| 429 |
"attributes": {}
|
| 430 |
}
|
| 431 |
},
|
| 432 |
-
"total_flos":
|
| 433 |
-
"train_batch_size":
|
| 434 |
"trial_name": null,
|
| 435 |
"trial_params": null
|
| 436 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.1500573754310608,
|
| 3 |
+
"best_model_checkpoint": "d:\\\\whisper-large-v3-pt-3000h-3\\checkpoint-691",
|
| 4 |
+
"epoch": 9.99638336347197,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 6910,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.03616636528028933,
|
| 13 |
+
"grad_norm": 5.811695098876953,
|
| 14 |
+
"learning_rate": 2.5000000000000004e-07,
|
| 15 |
+
"loss": 0.7755,
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.07233273056057866,
|
| 20 |
+
"grad_norm": Infinity,
|
| 21 |
+
"learning_rate": 4.900000000000001e-07,
|
| 22 |
+
"loss": 1.2265,
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.10849909584086799,
|
| 27 |
+
"grad_norm": 3.2242519855499268,
|
| 28 |
+
"learning_rate": 7.4e-07,
|
| 29 |
+
"loss": 0.3546,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.14466546112115733,
|
| 34 |
+
"grad_norm": 11.40543270111084,
|
| 35 |
+
"learning_rate": 9.9e-07,
|
| 36 |
+
"loss": 0.2988,
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.18083182640144665,
|
| 41 |
+
"grad_norm": 2.457425117492676,
|
| 42 |
+
"learning_rate": 1.2400000000000002e-06,
|
| 43 |
+
"loss": 0.174,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.21699819168173598,
|
| 48 |
+
"grad_norm": 10.275086402893066,
|
| 49 |
+
"learning_rate": 1.4900000000000001e-06,
|
| 50 |
+
"loss": 0.2317,
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.25316455696202533,
|
| 55 |
+
"grad_norm": 2.2328078746795654,
|
| 56 |
+
"learning_rate": 1.74e-06,
|
| 57 |
+
"loss": 0.1487,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.28933092224231466,
|
| 62 |
+
"grad_norm": 8.097187995910645,
|
| 63 |
+
"learning_rate": 1.9900000000000004e-06,
|
| 64 |
+
"loss": 0.205,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.325497287522604,
|
| 69 |
+
"grad_norm": 2.458644390106201,
|
| 70 |
+
"learning_rate": 2.24e-06,
|
| 71 |
+
"loss": 0.1509,
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.3616636528028933,
|
| 76 |
+
"grad_norm": 8.181588172912598,
|
| 77 |
+
"learning_rate": 2.4900000000000003e-06,
|
| 78 |
+
"loss": 0.2057,
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.39783001808318263,
|
| 83 |
+
"grad_norm": 2.0923259258270264,
|
| 84 |
+
"learning_rate": 2.7400000000000004e-06,
|
| 85 |
+
"loss": 0.1579,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.43399638336347196,
|
| 90 |
+
"grad_norm": 5.871222496032715,
|
| 91 |
+
"learning_rate": 2.99e-06,
|
| 92 |
+
"loss": 0.188,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.4701627486437613,
|
| 97 |
+
"grad_norm": 2.132258653640747,
|
| 98 |
+
"learning_rate": 3.2400000000000003e-06,
|
| 99 |
+
"loss": 0.1513,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.5063291139240507,
|
| 104 |
+
"grad_norm": 9.259086608886719,
|
| 105 |
+
"learning_rate": 3.49e-06,
|
| 106 |
+
"loss": 0.1773,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.5424954792043399,
|
| 111 |
+
"grad_norm": 2.059004783630371,
|
| 112 |
+
"learning_rate": 3.74e-06,
|
| 113 |
+
"loss": 0.1469,
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.5786618444846293,
|
| 118 |
+
"grad_norm": 6.2740302085876465,
|
| 119 |
+
"learning_rate": 3.990000000000001e-06,
|
| 120 |
+
"loss": 0.1765,
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.6148282097649186,
|
| 125 |
+
"grad_norm": 2.315974473953247,
|
| 126 |
+
"learning_rate": 4.24e-06,
|
| 127 |
+
"loss": 0.1364,
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.650994575045208,
|
| 132 |
+
"grad_norm": 6.675321102142334,
|
| 133 |
+
"learning_rate": 4.49e-06,
|
| 134 |
+
"loss": 0.1921,
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.6871609403254972,
|
| 139 |
+
"grad_norm": 2.4726405143737793,
|
| 140 |
+
"learning_rate": 4.74e-06,
|
| 141 |
+
"loss": 0.1365,
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 0.7233273056057866,
|
| 146 |
+
"grad_norm": 7.037698745727539,
|
| 147 |
+
"learning_rate": 4.9900000000000005e-06,
|
| 148 |
+
"loss": 0.1792,
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.759493670886076,
|
| 153 |
+
"grad_norm": 2.1146976947784424,
|
| 154 |
+
"learning_rate": 5.240000000000001e-06,
|
| 155 |
+
"loss": 0.1347,
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
+
"epoch": 0.7956600361663653,
|
| 160 |
+
"grad_norm": 4.760638236999512,
|
| 161 |
+
"learning_rate": 5.490000000000001e-06,
|
| 162 |
+
"loss": 0.1702,
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 0.8318264014466547,
|
| 167 |
+
"grad_norm": 3.498173952102661,
|
| 168 |
+
"learning_rate": 5.74e-06,
|
| 169 |
+
"loss": 0.132,
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.8679927667269439,
|
| 174 |
+
"grad_norm": 6.879450798034668,
|
| 175 |
+
"learning_rate": 5.99e-06,
|
| 176 |
+
"loss": 0.1589,
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 0.9041591320072333,
|
| 181 |
+
"grad_norm": 2.3761587142944336,
|
| 182 |
+
"learning_rate": 6.24e-06,
|
| 183 |
+
"loss": 0.15,
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
+
"epoch": 0.9403254972875226,
|
| 188 |
+
"grad_norm": 5.599714756011963,
|
| 189 |
+
"learning_rate": 6.4900000000000005e-06,
|
| 190 |
+
"loss": 0.1801,
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 0.976491862567812,
|
| 195 |
+
"grad_norm": 3.146120071411133,
|
| 196 |
+
"learning_rate": 6.740000000000001e-06,
|
| 197 |
+
"loss": 0.1388,
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 0.9996383363471971,
|
| 202 |
+
"eval_loss": 0.1500573754310608,
|
| 203 |
+
"eval_runtime": 40633.1057,
|
| 204 |
+
"eval_samples_per_second": 0.234,
|
| 205 |
+
"eval_steps_per_second": 0.029,
|
| 206 |
+
"eval_wer": 0.10736707238949392,
|
| 207 |
"step": 691
|
| 208 |
},
|
| 209 |
{
|
| 210 |
+
"epoch": 1.0126582278481013,
|
| 211 |
+
"grad_norm": 1.3991518020629883,
|
| 212 |
+
"learning_rate": 6.99e-06,
|
| 213 |
+
"loss": 0.1468,
|
| 214 |
"step": 700
|
| 215 |
},
|
| 216 |
{
|
| 217 |
+
"epoch": 1.0488245931283906,
|
| 218 |
+
"grad_norm": 2.3540124893188477,
|
| 219 |
+
"learning_rate": 7.24e-06,
|
| 220 |
+
"loss": 0.0971,
|
| 221 |
"step": 725
|
| 222 |
},
|
| 223 |
{
|
| 224 |
+
"epoch": 1.0849909584086799,
|
| 225 |
+
"grad_norm": 1.6329786777496338,
|
| 226 |
+
"learning_rate": 7.49e-06,
|
| 227 |
+
"loss": 0.1155,
|
| 228 |
"step": 750
|
| 229 |
},
|
| 230 |
{
|
| 231 |
+
"epoch": 1.1211573236889691,
|
| 232 |
+
"grad_norm": 2.5000386238098145,
|
| 233 |
+
"learning_rate": 7.74e-06,
|
| 234 |
+
"loss": 0.0903,
|
| 235 |
"step": 775
|
| 236 |
},
|
| 237 |
{
|
| 238 |
+
"epoch": 1.1573236889692586,
|
| 239 |
+
"grad_norm": 1.3056849241256714,
|
| 240 |
+
"learning_rate": 7.990000000000001e-06,
|
| 241 |
+
"loss": 0.1255,
|
| 242 |
"step": 800
|
| 243 |
},
|
| 244 |
{
|
| 245 |
+
"epoch": 1.193490054249548,
|
| 246 |
+
"grad_norm": 2.1964187622070312,
|
| 247 |
+
"learning_rate": 8.24e-06,
|
| 248 |
+
"loss": 0.0987,
|
| 249 |
"step": 825
|
| 250 |
},
|
| 251 |
{
|
| 252 |
+
"epoch": 1.2296564195298372,
|
| 253 |
+
"grad_norm": 1.5802552700042725,
|
| 254 |
+
"learning_rate": 8.49e-06,
|
| 255 |
+
"loss": 0.1393,
|
| 256 |
"step": 850
|
| 257 |
},
|
| 258 |
{
|
| 259 |
+
"epoch": 1.2658227848101267,
|
| 260 |
+
"grad_norm": 1.7570548057556152,
|
| 261 |
+
"learning_rate": 8.740000000000001e-06,
|
| 262 |
+
"loss": 0.1094,
|
| 263 |
"step": 875
|
| 264 |
},
|
| 265 |
{
|
| 266 |
+
"epoch": 1.301989150090416,
|
| 267 |
+
"grad_norm": 1.5590579509735107,
|
| 268 |
+
"learning_rate": 8.99e-06,
|
| 269 |
+
"loss": 0.1276,
|
| 270 |
"step": 900
|
| 271 |
},
|
| 272 |
{
|
| 273 |
+
"epoch": 1.3381555153707052,
|
| 274 |
+
"grad_norm": 2.429781913757324,
|
| 275 |
+
"learning_rate": 9.240000000000001e-06,
|
| 276 |
+
"loss": 0.1023,
|
| 277 |
"step": 925
|
| 278 |
},
|
| 279 |
{
|
| 280 |
+
"epoch": 1.3743218806509945,
|
| 281 |
+
"grad_norm": 2.3684589862823486,
|
| 282 |
+
"learning_rate": 9.49e-06,
|
| 283 |
+
"loss": 0.1372,
|
| 284 |
"step": 950
|
| 285 |
},
|
| 286 |
{
|
| 287 |
+
"epoch": 1.410488245931284,
|
| 288 |
+
"grad_norm": 5.570235252380371,
|
| 289 |
+
"learning_rate": 9.74e-06,
|
| 290 |
+
"loss": 0.1122,
|
| 291 |
"step": 975
|
| 292 |
},
|
| 293 |
{
|
| 294 |
+
"epoch": 1.4466546112115732,
|
| 295 |
+
"grad_norm": 2.2822372913360596,
|
| 296 |
+
"learning_rate": 9.990000000000001e-06,
|
| 297 |
+
"loss": 0.1463,
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
+
"epoch": 1.4828209764918625,
|
| 302 |
+
"grad_norm": 2.185589075088501,
|
| 303 |
+
"learning_rate": 9.959390862944164e-06,
|
| 304 |
+
"loss": 0.1055,
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
+
"epoch": 1.518987341772152,
|
| 309 |
+
"grad_norm": 2.054370164871216,
|
| 310 |
+
"learning_rate": 9.917089678511e-06,
|
| 311 |
+
"loss": 0.1426,
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
+
"epoch": 1.5551537070524413,
|
| 316 |
+
"grad_norm": 2.4509952068328857,
|
| 317 |
+
"learning_rate": 9.874788494077835e-06,
|
| 318 |
+
"loss": 0.0985,
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
+
"epoch": 1.5913200723327305,
|
| 323 |
+
"grad_norm": 1.7492316961288452,
|
| 324 |
+
"learning_rate": 9.83248730964467e-06,
|
| 325 |
+
"loss": 0.1553,
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
+
"epoch": 1.6274864376130198,
|
| 330 |
+
"grad_norm": 2.5610837936401367,
|
| 331 |
+
"learning_rate": 9.790186125211507e-06,
|
| 332 |
+
"loss": 0.1235,
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
+
"epoch": 1.663652802893309,
|
| 337 |
+
"grad_norm": 1.8614284992218018,
|
| 338 |
+
"learning_rate": 9.747884940778342e-06,
|
| 339 |
+
"loss": 0.1524,
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
+
"epoch": 1.6998191681735986,
|
| 344 |
+
"grad_norm": 3.679600954055786,
|
| 345 |
+
"learning_rate": 9.705583756345178e-06,
|
| 346 |
+
"loss": 0.1237,
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
+
"epoch": 1.7359855334538878,
|
| 351 |
+
"grad_norm": 1.617661952972412,
|
| 352 |
+
"learning_rate": 9.663282571912015e-06,
|
| 353 |
+
"loss": 0.1341,
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
+
"epoch": 1.7721518987341773,
|
| 358 |
+
"grad_norm": 2.3510446548461914,
|
| 359 |
+
"learning_rate": 9.62098138747885e-06,
|
| 360 |
+
"loss": 0.1232,
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
+
"epoch": 1.8083182640144666,
|
| 365 |
+
"grad_norm": 1.9561399221420288,
|
| 366 |
+
"learning_rate": 9.578680203045685e-06,
|
| 367 |
+
"loss": 0.1294,
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
+
"epoch": 1.8444846292947559,
|
| 372 |
+
"grad_norm": 3.439204216003418,
|
| 373 |
+
"learning_rate": 9.536379018612522e-06,
|
| 374 |
+
"loss": 0.1119,
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
+
"epoch": 1.8806509945750451,
|
| 379 |
+
"grad_norm": 2.9221997261047363,
|
| 380 |
+
"learning_rate": 9.494077834179358e-06,
|
| 381 |
+
"loss": 0.1494,
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
+
"epoch": 1.9168173598553344,
|
| 386 |
+
"grad_norm": 2.623828411102295,
|
| 387 |
+
"learning_rate": 9.451776649746193e-06,
|
| 388 |
+
"loss": 0.1109,
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
+
"epoch": 1.952983725135624,
|
| 393 |
+
"grad_norm": 1.5950006246566772,
|
| 394 |
+
"learning_rate": 9.40947546531303e-06,
|
| 395 |
+
"loss": 0.134,
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
+
"epoch": 1.9891500904159132,
|
| 400 |
+
"grad_norm": 2.305098533630371,
|
| 401 |
+
"learning_rate": 9.367174280879865e-06,
|
| 402 |
+
"loss": 0.108,
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
+
"epoch": 1.9992766726943942,
|
| 407 |
+
"eval_loss": 0.16191039979457855,
|
| 408 |
+
"eval_runtime": 48873.7373,
|
| 409 |
+
"eval_samples_per_second": 0.194,
|
| 410 |
+
"eval_steps_per_second": 0.024,
|
| 411 |
+
"eval_wer": 0.11531069827033953,
|
| 412 |
+
"step": 1382
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 2.0253164556962027,
|
| 416 |
+
"grad_norm": 1.2765154838562012,
|
| 417 |
+
"learning_rate": 9.324873096446702e-06,
|
| 418 |
+
"loss": 0.1064,
|
| 419 |
+
"step": 1400
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 2.061482820976492,
|
| 423 |
+
"grad_norm": 2.579493999481201,
|
| 424 |
+
"learning_rate": 9.282571912013538e-06,
|
| 425 |
+
"loss": 0.0676,
|
| 426 |
+
"step": 1425
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 2.097649186256781,
|
| 430 |
+
"grad_norm": 1.4855302572250366,
|
| 431 |
+
"learning_rate": 9.240270727580373e-06,
|
| 432 |
+
"loss": 0.0733,
|
| 433 |
+
"step": 1450
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 2.1338155515370705,
|
| 437 |
+
"grad_norm": 1.7606139183044434,
|
| 438 |
+
"learning_rate": 9.19796954314721e-06,
|
| 439 |
+
"loss": 0.069,
|
| 440 |
+
"step": 1475
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.1699819168173597,
|
| 444 |
+
"grad_norm": 1.383319616317749,
|
| 445 |
+
"learning_rate": 9.155668358714045e-06,
|
| 446 |
+
"loss": 0.069,
|
| 447 |
+
"step": 1500
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 2.206148282097649,
|
| 451 |
+
"grad_norm": 3.2015483379364014,
|
| 452 |
+
"learning_rate": 9.11336717428088e-06,
|
| 453 |
+
"loss": 0.0725,
|
| 454 |
+
"step": 1525
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 2.2423146473779383,
|
| 458 |
+
"grad_norm": 2.2851688861846924,
|
| 459 |
+
"learning_rate": 9.071065989847716e-06,
|
| 460 |
+
"loss": 0.0743,
|
| 461 |
+
"step": 1550
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 2.278481012658228,
|
| 465 |
+
"grad_norm": 3.721728801727295,
|
| 466 |
+
"learning_rate": 9.028764805414553e-06,
|
| 467 |
+
"loss": 0.0704,
|
| 468 |
+
"step": 1575
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 2.3146473779385173,
|
| 472 |
+
"grad_norm": 1.077505111694336,
|
| 473 |
+
"learning_rate": 8.986463620981388e-06,
|
| 474 |
+
"loss": 0.0743,
|
| 475 |
+
"step": 1600
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 2.3508137432188065,
|
| 479 |
+
"grad_norm": 2.2788355350494385,
|
| 480 |
+
"learning_rate": 8.944162436548224e-06,
|
| 481 |
+
"loss": 0.071,
|
| 482 |
+
"step": 1625
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.386980108499096,
|
| 486 |
+
"grad_norm": 3.0096864700317383,
|
| 487 |
+
"learning_rate": 8.901861252115061e-06,
|
| 488 |
+
"loss": 0.0808,
|
| 489 |
+
"step": 1650
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 2.423146473779385,
|
| 493 |
+
"grad_norm": 3.3069839477539062,
|
| 494 |
+
"learning_rate": 8.859560067681896e-06,
|
| 495 |
+
"loss": 0.0731,
|
| 496 |
+
"step": 1675
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 2.4593128390596743,
|
| 500 |
+
"grad_norm": 1.791975498199463,
|
| 501 |
+
"learning_rate": 8.817258883248732e-06,
|
| 502 |
+
"loss": 0.0788,
|
| 503 |
+
"step": 1700
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 2.495479204339964,
|
| 507 |
+
"grad_norm": 1.9154149293899536,
|
| 508 |
+
"learning_rate": 8.774957698815567e-06,
|
| 509 |
+
"loss": 0.0737,
|
| 510 |
+
"step": 1725
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 2.5316455696202533,
|
| 514 |
+
"grad_norm": 1.7026249170303345,
|
| 515 |
+
"learning_rate": 8.732656514382404e-06,
|
| 516 |
+
"loss": 0.0821,
|
| 517 |
+
"step": 1750
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 2.5678119349005426,
|
| 521 |
+
"grad_norm": 1.5672781467437744,
|
| 522 |
+
"learning_rate": 8.69035532994924e-06,
|
| 523 |
+
"loss": 0.0708,
|
| 524 |
+
"step": 1775
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.603978300180832,
|
| 528 |
+
"grad_norm": 1.1428169012069702,
|
| 529 |
+
"learning_rate": 8.648054145516075e-06,
|
| 530 |
+
"loss": 0.0825,
|
| 531 |
+
"step": 1800
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 2.640144665461121,
|
| 535 |
+
"grad_norm": 4.432084560394287,
|
| 536 |
+
"learning_rate": 8.60575296108291e-06,
|
| 537 |
+
"loss": 0.0749,
|
| 538 |
+
"step": 1825
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 2.6763110307414104,
|
| 542 |
+
"grad_norm": 2.4532995223999023,
|
| 543 |
+
"learning_rate": 8.563451776649747e-06,
|
| 544 |
+
"loss": 0.0856,
|
| 545 |
+
"step": 1850
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 2.7124773960216997,
|
| 549 |
+
"grad_norm": 2.286349296569824,
|
| 550 |
+
"learning_rate": 8.521150592216582e-06,
|
| 551 |
+
"loss": 0.075,
|
| 552 |
+
"step": 1875
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 2.748643761301989,
|
| 556 |
+
"grad_norm": 1.524868130683899,
|
| 557 |
+
"learning_rate": 8.478849407783418e-06,
|
| 558 |
+
"loss": 0.0829,
|
| 559 |
+
"step": 1900
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 2.7848101265822782,
|
| 563 |
+
"grad_norm": 3.2034952640533447,
|
| 564 |
+
"learning_rate": 8.436548223350255e-06,
|
| 565 |
+
"loss": 0.0749,
|
| 566 |
+
"step": 1925
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.820976491862568,
|
| 570 |
+
"grad_norm": 2.6012070178985596,
|
| 571 |
+
"learning_rate": 8.39424703891709e-06,
|
| 572 |
+
"loss": 0.089,
|
| 573 |
+
"step": 1950
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 2.857142857142857,
|
| 577 |
+
"grad_norm": 2.747398614883423,
|
| 578 |
+
"learning_rate": 8.351945854483925e-06,
|
| 579 |
+
"loss": 0.0774,
|
| 580 |
+
"step": 1975
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 2.8933092224231465,
|
| 584 |
+
"grad_norm": 1.3775649070739746,
|
| 585 |
+
"learning_rate": 8.309644670050762e-06,
|
| 586 |
+
"loss": 0.0795,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 2.9294755877034357,
|
| 591 |
+
"grad_norm": 6.074094295501709,
|
| 592 |
+
"learning_rate": 8.267343485617598e-06,
|
| 593 |
+
"loss": 0.0658,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 2.965641952983725,
|
| 598 |
+
"grad_norm": 2.2065982818603516,
|
| 599 |
+
"learning_rate": 8.225042301184433e-06,
|
| 600 |
+
"loss": 0.091,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 2.9989150090415913,
|
| 605 |
+
"eval_loss": 0.1697043776512146,
|
| 606 |
+
"eval_runtime": 49669.5944,
|
| 607 |
+
"eval_samples_per_second": 0.191,
|
| 608 |
+
"eval_steps_per_second": 0.024,
|
| 609 |
+
"eval_wer": 0.11237988468930173,
|
| 610 |
+
"step": 2073
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 3.0018083182640143,
|
| 614 |
+
"grad_norm": 1.5902258157730103,
|
| 615 |
+
"learning_rate": 8.18274111675127e-06,
|
| 616 |
+
"loss": 0.0754,
|
| 617 |
+
"step": 2075
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 3.037974683544304,
|
| 621 |
+
"grad_norm": 1.6310546398162842,
|
| 622 |
+
"learning_rate": 8.140439932318105e-06,
|
| 623 |
+
"loss": 0.0409,
|
| 624 |
+
"step": 2100
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 3.0741410488245933,
|
| 628 |
+
"grad_norm": 1.9858086109161377,
|
| 629 |
+
"learning_rate": 8.098138747884942e-06,
|
| 630 |
+
"loss": 0.0529,
|
| 631 |
+
"step": 2125
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 3.1103074141048825,
|
| 635 |
+
"grad_norm": 0.8269912004470825,
|
| 636 |
+
"learning_rate": 8.055837563451778e-06,
|
| 637 |
+
"loss": 0.0376,
|
| 638 |
+
"step": 2150
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.146473779385172,
|
| 642 |
+
"grad_norm": 0.8812470436096191,
|
| 643 |
+
"learning_rate": 8.013536379018613e-06,
|
| 644 |
+
"loss": 0.0522,
|
| 645 |
+
"step": 2175
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.182640144665461,
|
| 649 |
+
"grad_norm": 1.5196110010147095,
|
| 650 |
+
"learning_rate": 7.97123519458545e-06,
|
| 651 |
+
"loss": 0.043,
|
| 652 |
+
"step": 2200
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"epoch": 3.2188065099457503,
|
| 656 |
+
"grad_norm": 1.6323350667953491,
|
| 657 |
+
"learning_rate": 7.928934010152285e-06,
|
| 658 |
+
"loss": 0.0577,
|
| 659 |
+
"step": 2225
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 3.2549728752260396,
|
| 663 |
+
"grad_norm": 1.475954532623291,
|
| 664 |
+
"learning_rate": 7.88663282571912e-06,
|
| 665 |
+
"loss": 0.0396,
|
| 666 |
+
"step": 2250
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 3.291139240506329,
|
| 670 |
+
"grad_norm": 1.9558379650115967,
|
| 671 |
+
"learning_rate": 7.844331641285958e-06,
|
| 672 |
+
"loss": 0.0433,
|
| 673 |
+
"step": 2275
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 3.3273056057866186,
|
| 677 |
+
"grad_norm": 1.1998658180236816,
|
| 678 |
+
"learning_rate": 7.802030456852793e-06,
|
| 679 |
+
"loss": 0.0383,
|
| 680 |
+
"step": 2300
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 3.363471971066908,
|
| 684 |
+
"grad_norm": 1.4807426929473877,
|
| 685 |
+
"learning_rate": 7.759729272419628e-06,
|
| 686 |
+
"loss": 0.06,
|
| 687 |
+
"step": 2325
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 3.399638336347197,
|
| 691 |
+
"grad_norm": 0.8246263265609741,
|
| 692 |
+
"learning_rate": 7.717428087986464e-06,
|
| 693 |
+
"loss": 0.0443,
|
| 694 |
+
"step": 2350
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 3.4358047016274864,
|
| 698 |
+
"grad_norm": 0.9691349267959595,
|
| 699 |
+
"learning_rate": 7.675126903553301e-06,
|
| 700 |
+
"loss": 0.0535,
|
| 701 |
+
"step": 2375
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 3.4719710669077757,
|
| 705 |
+
"grad_norm": 0.7182671427726746,
|
| 706 |
+
"learning_rate": 7.632825719120136e-06,
|
| 707 |
+
"loss": 0.0417,
|
| 708 |
+
"step": 2400
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 3.508137432188065,
|
| 712 |
+
"grad_norm": 1.2362239360809326,
|
| 713 |
+
"learning_rate": 7.5905245346869715e-06,
|
| 714 |
+
"loss": 0.0529,
|
| 715 |
+
"step": 2425
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 3.5443037974683547,
|
| 719 |
+
"grad_norm": 1.6033586263656616,
|
| 720 |
+
"learning_rate": 7.5482233502538086e-06,
|
| 721 |
+
"loss": 0.0377,
|
| 722 |
+
"step": 2450
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 3.580470162748644,
|
| 726 |
+
"grad_norm": 1.3885324001312256,
|
| 727 |
+
"learning_rate": 7.505922165820644e-06,
|
| 728 |
+
"loss": 0.056,
|
| 729 |
+
"step": 2475
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 3.616636528028933,
|
| 733 |
+
"grad_norm": 0.925649881362915,
|
| 734 |
+
"learning_rate": 7.463620981387479e-06,
|
| 735 |
+
"loss": 0.0399,
|
| 736 |
+
"step": 2500
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 3.6528028933092225,
|
| 740 |
+
"grad_norm": 2.046786308288574,
|
| 741 |
+
"learning_rate": 7.4213197969543146e-06,
|
| 742 |
+
"loss": 0.064,
|
| 743 |
+
"step": 2525
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 3.6889692585895117,
|
| 747 |
+
"grad_norm": 2.5204615592956543,
|
| 748 |
+
"learning_rate": 7.379018612521152e-06,
|
| 749 |
+
"loss": 0.0392,
|
| 750 |
+
"step": 2550
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.725135623869801,
|
| 754 |
+
"grad_norm": 1.6770093441009521,
|
| 755 |
+
"learning_rate": 7.336717428087987e-06,
|
| 756 |
+
"loss": 0.0508,
|
| 757 |
+
"step": 2575
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 3.7613019891500903,
|
| 761 |
+
"grad_norm": 1.5132670402526855,
|
| 762 |
+
"learning_rate": 7.294416243654823e-06,
|
| 763 |
+
"loss": 0.0442,
|
| 764 |
+
"step": 2600
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 3.7974683544303796,
|
| 768 |
+
"grad_norm": 1.3414489030838013,
|
| 769 |
+
"learning_rate": 7.2521150592216584e-06,
|
| 770 |
+
"loss": 0.0551,
|
| 771 |
+
"step": 2625
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 3.833634719710669,
|
| 775 |
+
"grad_norm": 1.41124427318573,
|
| 776 |
+
"learning_rate": 7.209813874788495e-06,
|
| 777 |
+
"loss": 0.0382,
|
| 778 |
+
"step": 2650
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 3.8698010849909585,
|
| 782 |
+
"grad_norm": 1.133798360824585,
|
| 783 |
+
"learning_rate": 7.167512690355331e-06,
|
| 784 |
+
"loss": 0.0585,
|
| 785 |
+
"step": 2675
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 3.905967450271248,
|
| 789 |
+
"grad_norm": 2.0255684852600098,
|
| 790 |
+
"learning_rate": 7.125211505922166e-06,
|
| 791 |
+
"loss": 0.0417,
|
| 792 |
+
"step": 2700
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 3.942133815551537,
|
| 796 |
+
"grad_norm": 1.0678452253341675,
|
| 797 |
+
"learning_rate": 7.082910321489002e-06,
|
| 798 |
+
"loss": 0.0578,
|
| 799 |
+
"step": 2725
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 3.9783001808318263,
|
| 803 |
+
"grad_norm": 1.9355206489562988,
|
| 804 |
+
"learning_rate": 7.0406091370558385e-06,
|
| 805 |
+
"loss": 0.0461,
|
| 806 |
+
"step": 2750
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.0,
|
| 810 |
+
"eval_loss": 0.17637395858764648,
|
| 811 |
+
"eval_runtime": 49043.1261,
|
| 812 |
+
"eval_samples_per_second": 0.194,
|
| 813 |
+
"eval_steps_per_second": 0.024,
|
| 814 |
+
"eval_wer": 0.1119795003203075,
|
| 815 |
+
"step": 2765
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.014466546112116,
|
| 819 |
+
"grad_norm": 0.5305593013763428,
|
| 820 |
+
"learning_rate": 6.998307952622674e-06,
|
| 821 |
+
"loss": 0.0582,
|
| 822 |
+
"step": 2775
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 4.050632911392405,
|
| 826 |
+
"grad_norm": 1.1978957653045654,
|
| 827 |
+
"learning_rate": 6.956006768189509e-06,
|
| 828 |
+
"loss": 0.0205,
|
| 829 |
+
"step": 2800
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 4.086799276672695,
|
| 833 |
+
"grad_norm": 1.0628869533538818,
|
| 834 |
+
"learning_rate": 6.913705583756346e-06,
|
| 835 |
+
"loss": 0.024,
|
| 836 |
+
"step": 2825
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.122965641952984,
|
| 840 |
+
"grad_norm": 3.1139473915100098,
|
| 841 |
+
"learning_rate": 6.8714043993231816e-06,
|
| 842 |
+
"loss": 0.0242,
|
| 843 |
+
"step": 2850
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 4.159132007233273,
|
| 847 |
+
"grad_norm": 0.7646371126174927,
|
| 848 |
+
"learning_rate": 6.829103214890017e-06,
|
| 849 |
+
"loss": 0.0363,
|
| 850 |
+
"step": 2875
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 4.195298372513562,
|
| 854 |
+
"grad_norm": 1.3536927700042725,
|
| 855 |
+
"learning_rate": 6.786802030456854e-06,
|
| 856 |
+
"loss": 0.0254,
|
| 857 |
+
"step": 2900
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 4.231464737793852,
|
| 861 |
+
"grad_norm": 1.744105339050293,
|
| 862 |
+
"learning_rate": 6.744500846023689e-06,
|
| 863 |
+
"loss": 0.0362,
|
| 864 |
+
"step": 2925
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 4.267631103074141,
|
| 868 |
+
"grad_norm": 0.8838434815406799,
|
| 869 |
+
"learning_rate": 6.702199661590525e-06,
|
| 870 |
+
"loss": 0.0236,
|
| 871 |
+
"step": 2950
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 4.30379746835443,
|
| 875 |
+
"grad_norm": 1.056674838066101,
|
| 876 |
+
"learning_rate": 6.659898477157361e-06,
|
| 877 |
+
"loss": 0.0311,
|
| 878 |
+
"step": 2975
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 4.3399638336347195,
|
| 882 |
+
"grad_norm": 1.42670476436615,
|
| 883 |
+
"learning_rate": 6.617597292724197e-06,
|
| 884 |
+
"loss": 0.0277,
|
| 885 |
+
"step": 3000
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 4.376130198915009,
|
| 889 |
+
"grad_norm": 1.0685067176818848,
|
| 890 |
+
"learning_rate": 6.575296108291032e-06,
|
| 891 |
+
"loss": 0.0346,
|
| 892 |
+
"step": 3025
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 4.412296564195298,
|
| 896 |
+
"grad_norm": 0.6285632252693176,
|
| 897 |
+
"learning_rate": 6.5329949238578685e-06,
|
| 898 |
+
"loss": 0.0271,
|
| 899 |
+
"step": 3050
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 4.448462929475587,
|
| 903 |
+
"grad_norm": 1.1495152711868286,
|
| 904 |
+
"learning_rate": 6.490693739424705e-06,
|
| 905 |
+
"loss": 0.0339,
|
| 906 |
+
"step": 3075
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 4.484629294755877,
|
| 910 |
+
"grad_norm": 1.4339568614959717,
|
| 911 |
+
"learning_rate": 6.448392554991541e-06,
|
| 912 |
+
"loss": 0.0218,
|
| 913 |
+
"step": 3100
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 4.520795660036167,
|
| 917 |
+
"grad_norm": 0.9594871997833252,
|
| 918 |
+
"learning_rate": 6.406091370558376e-06,
|
| 919 |
+
"loss": 0.029,
|
| 920 |
+
"step": 3125
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 4.556962025316456,
|
| 924 |
+
"grad_norm": 1.1738754510879517,
|
| 925 |
+
"learning_rate": 6.3637901861252115e-06,
|
| 926 |
+
"loss": 0.0312,
|
| 927 |
+
"step": 3150
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 4.593128390596745,
|
| 931 |
+
"grad_norm": 1.202968716621399,
|
| 932 |
+
"learning_rate": 6.3214890016920485e-06,
|
| 933 |
+
"loss": 0.032,
|
| 934 |
+
"step": 3175
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 4.6292947558770345,
|
| 938 |
+
"grad_norm": 1.722782850265503,
|
| 939 |
+
"learning_rate": 6.279187817258884e-06,
|
| 940 |
+
"loss": 0.0327,
|
| 941 |
+
"step": 3200
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 4.665461121157324,
|
| 945 |
+
"grad_norm": 0.8393648862838745,
|
| 946 |
+
"learning_rate": 6.236886632825719e-06,
|
| 947 |
+
"loss": 0.0356,
|
| 948 |
+
"step": 3225
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 4.701627486437613,
|
| 952 |
+
"grad_norm": 2.255413770675659,
|
| 953 |
+
"learning_rate": 6.194585448392556e-06,
|
| 954 |
+
"loss": 0.0298,
|
| 955 |
+
"step": 3250
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 4.737793851717902,
|
| 959 |
+
"grad_norm": 1.082092523574829,
|
| 960 |
+
"learning_rate": 6.152284263959392e-06,
|
| 961 |
+
"loss": 0.035,
|
| 962 |
+
"step": 3275
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 4.773960216998192,
|
| 966 |
+
"grad_norm": 1.9351752996444702,
|
| 967 |
+
"learning_rate": 6.109983079526227e-06,
|
| 968 |
+
"loss": 0.025,
|
| 969 |
+
"step": 3300
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 4.810126582278481,
|
| 973 |
+
"grad_norm": 1.7391856908798218,
|
| 974 |
+
"learning_rate": 6.067681895093063e-06,
|
| 975 |
+
"loss": 0.0398,
|
| 976 |
+
"step": 3325
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 4.84629294755877,
|
| 980 |
+
"grad_norm": 1.332651972770691,
|
| 981 |
+
"learning_rate": 6.025380710659899e-06,
|
| 982 |
+
"loss": 0.0269,
|
| 983 |
+
"step": 3350
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 4.882459312839059,
|
| 987 |
+
"grad_norm": 1.1322509050369263,
|
| 988 |
+
"learning_rate": 5.983079526226735e-06,
|
| 989 |
+
"loss": 0.0296,
|
| 990 |
+
"step": 3375
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 4.918625678119349,
|
| 994 |
+
"grad_norm": 1.640367865562439,
|
| 995 |
+
"learning_rate": 5.940778341793571e-06,
|
| 996 |
+
"loss": 0.0245,
|
| 997 |
+
"step": 3400
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 4.954792043399638,
|
| 1001 |
+
"grad_norm": 1.5555723905563354,
|
| 1002 |
+
"learning_rate": 5.898477157360406e-06,
|
| 1003 |
+
"loss": 0.0455,
|
| 1004 |
+
"step": 3425
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 4.990958408679928,
|
| 1008 |
+
"grad_norm": 2.142411947250366,
|
| 1009 |
+
"learning_rate": 5.856175972927242e-06,
|
| 1010 |
+
"loss": 0.0264,
|
| 1011 |
+
"step": 3450
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 4.999638336347197,
|
| 1015 |
+
"eval_loss": 0.20242072641849518,
|
| 1016 |
+
"eval_runtime": 49328.761,
|
| 1017 |
+
"eval_samples_per_second": 0.192,
|
| 1018 |
+
"eval_steps_per_second": 0.024,
|
| 1019 |
+
"eval_wer": 0.11330877642536835,
|
| 1020 |
+
"step": 3456
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 5.027124773960217,
|
| 1024 |
+
"grad_norm": 2.6065480709075928,
|
| 1025 |
+
"learning_rate": 5.8138747884940785e-06,
|
| 1026 |
+
"loss": 0.0223,
|
| 1027 |
+
"step": 3475
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 5.063291139240507,
|
| 1031 |
+
"grad_norm": 6.660514831542969,
|
| 1032 |
+
"learning_rate": 5.771573604060914e-06,
|
| 1033 |
+
"loss": 0.0196,
|
| 1034 |
+
"step": 3500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 5.099457504520796,
|
| 1038 |
+
"grad_norm": 0.8771660923957825,
|
| 1039 |
+
"learning_rate": 5.729272419627751e-06,
|
| 1040 |
+
"loss": 0.0211,
|
| 1041 |
+
"step": 3525
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 5.135623869801085,
|
| 1045 |
+
"grad_norm": 5.249636650085449,
|
| 1046 |
+
"learning_rate": 5.686971235194586e-06,
|
| 1047 |
+
"loss": 0.0212,
|
| 1048 |
+
"step": 3550
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 5.1717902350813745,
|
| 1052 |
+
"grad_norm": 0.7272325754165649,
|
| 1053 |
+
"learning_rate": 5.6446700507614216e-06,
|
| 1054 |
+
"loss": 0.018,
|
| 1055 |
+
"step": 3575
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 5.207956600361664,
|
| 1059 |
+
"grad_norm": 1.247758388519287,
|
| 1060 |
+
"learning_rate": 5.602368866328257e-06,
|
| 1061 |
+
"loss": 0.0174,
|
| 1062 |
+
"step": 3600
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 5.244122965641953,
|
| 1066 |
+
"grad_norm": 1.6683036088943481,
|
| 1067 |
+
"learning_rate": 5.560067681895094e-06,
|
| 1068 |
+
"loss": 0.0208,
|
| 1069 |
+
"step": 3625
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 5.280289330922242,
|
| 1073 |
+
"grad_norm": 0.9653752446174622,
|
| 1074 |
+
"learning_rate": 5.517766497461929e-06,
|
| 1075 |
+
"loss": 0.0158,
|
| 1076 |
+
"step": 3650
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 5.3164556962025316,
|
| 1080 |
+
"grad_norm": 2.857499361038208,
|
| 1081 |
+
"learning_rate": 5.475465313028765e-06,
|
| 1082 |
+
"loss": 0.0215,
|
| 1083 |
+
"step": 3675
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 5.352622061482821,
|
| 1087 |
+
"grad_norm": 2.953227996826172,
|
| 1088 |
+
"learning_rate": 5.433164128595602e-06,
|
| 1089 |
+
"loss": 0.017,
|
| 1090 |
+
"step": 3700
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 5.38878842676311,
|
| 1094 |
+
"grad_norm": 2.159450054168701,
|
| 1095 |
+
"learning_rate": 5.390862944162437e-06,
|
| 1096 |
+
"loss": 0.0199,
|
| 1097 |
+
"step": 3725
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 5.424954792043399,
|
| 1101 |
+
"grad_norm": 2.423708200454712,
|
| 1102 |
+
"learning_rate": 5.348561759729273e-06,
|
| 1103 |
+
"loss": 0.0185,
|
| 1104 |
+
"step": 3750
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 5.461121157323689,
|
| 1108 |
+
"grad_norm": 1.3024959564208984,
|
| 1109 |
+
"learning_rate": 5.3062605752961085e-06,
|
| 1110 |
+
"loss": 0.0212,
|
| 1111 |
+
"step": 3775
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 5.497287522603978,
|
| 1115 |
+
"grad_norm": 1.6463954448699951,
|
| 1116 |
+
"learning_rate": 5.263959390862945e-06,
|
| 1117 |
+
"loss": 0.018,
|
| 1118 |
+
"step": 3800
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 5.533453887884267,
|
| 1122 |
+
"grad_norm": 0.8385624885559082,
|
| 1123 |
+
"learning_rate": 5.221658206429781e-06,
|
| 1124 |
+
"loss": 0.0191,
|
| 1125 |
+
"step": 3825
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 5.569620253164557,
|
| 1129 |
+
"grad_norm": 1.0305800437927246,
|
| 1130 |
+
"learning_rate": 5.179357021996616e-06,
|
| 1131 |
+
"loss": 0.0179,
|
| 1132 |
+
"step": 3850
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 5.605786618444847,
|
| 1136 |
+
"grad_norm": 1.5330723524093628,
|
| 1137 |
+
"learning_rate": 5.137055837563452e-06,
|
| 1138 |
+
"loss": 0.0203,
|
| 1139 |
+
"step": 3875
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 5.641952983725136,
|
| 1143 |
+
"grad_norm": 1.5121113061904907,
|
| 1144 |
+
"learning_rate": 5.0947546531302885e-06,
|
| 1145 |
+
"loss": 0.0187,
|
| 1146 |
+
"step": 3900
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 5.678119349005425,
|
| 1150 |
+
"grad_norm": 0.41953858733177185,
|
| 1151 |
+
"learning_rate": 5.052453468697124e-06,
|
| 1152 |
+
"loss": 0.0207,
|
| 1153 |
+
"step": 3925
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 5.714285714285714,
|
| 1157 |
+
"grad_norm": 3.4192609786987305,
|
| 1158 |
+
"learning_rate": 5.010152284263959e-06,
|
| 1159 |
+
"loss": 0.0176,
|
| 1160 |
+
"step": 3950
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 5.750452079566004,
|
| 1164 |
+
"grad_norm": 3.3937628269195557,
|
| 1165 |
+
"learning_rate": 4.967851099830795e-06,
|
| 1166 |
+
"loss": 0.0207,
|
| 1167 |
+
"step": 3975
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 5.786618444846293,
|
| 1171 |
+
"grad_norm": 1.6074724197387695,
|
| 1172 |
+
"learning_rate": 4.925549915397632e-06,
|
| 1173 |
+
"loss": 0.0207,
|
| 1174 |
+
"step": 4000
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 5.822784810126582,
|
| 1178 |
+
"grad_norm": 0.9468584656715393,
|
| 1179 |
+
"learning_rate": 4.883248730964467e-06,
|
| 1180 |
+
"loss": 0.018,
|
| 1181 |
+
"step": 4025
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 5.8589511754068715,
|
| 1185 |
+
"grad_norm": 1.2002973556518555,
|
| 1186 |
+
"learning_rate": 4.840947546531303e-06,
|
| 1187 |
+
"loss": 0.0151,
|
| 1188 |
+
"step": 4050
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 5.895117540687161,
|
| 1192 |
+
"grad_norm": 1.0356348752975464,
|
| 1193 |
+
"learning_rate": 4.798646362098139e-06,
|
| 1194 |
+
"loss": 0.0201,
|
| 1195 |
+
"step": 4075
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 5.93128390596745,
|
| 1199 |
+
"grad_norm": 1.657277226448059,
|
| 1200 |
+
"learning_rate": 4.756345177664975e-06,
|
| 1201 |
+
"loss": 0.0202,
|
| 1202 |
+
"step": 4100
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 5.967450271247739,
|
| 1206 |
+
"grad_norm": 1.1826387643814087,
|
| 1207 |
+
"learning_rate": 4.714043993231811e-06,
|
| 1208 |
+
"loss": 0.0203,
|
| 1209 |
+
"step": 4125
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 5.9992766726943945,
|
| 1213 |
+
"eval_loss": 0.22004875540733337,
|
| 1214 |
+
"eval_runtime": 49062.4528,
|
| 1215 |
+
"eval_samples_per_second": 0.194,
|
| 1216 |
+
"eval_steps_per_second": 0.024,
|
| 1217 |
+
"eval_wer": 0.10991351697629724,
|
| 1218 |
+
"step": 4147
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 6.003616636528029,
|
| 1222 |
+
"grad_norm": 0.2818962335586548,
|
| 1223 |
+
"learning_rate": 4.671742808798647e-06,
|
| 1224 |
+
"loss": 0.0216,
|
| 1225 |
+
"step": 4150
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 6.039783001808318,
|
| 1229 |
+
"grad_norm": 0.8754785656929016,
|
| 1230 |
+
"learning_rate": 4.631133671742809e-06,
|
| 1231 |
+
"loss": 0.0117,
|
| 1232 |
+
"step": 4175
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 6.075949367088608,
|
| 1236 |
+
"grad_norm": 0.6040045619010925,
|
| 1237 |
+
"learning_rate": 4.588832487309645e-06,
|
| 1238 |
+
"loss": 0.0185,
|
| 1239 |
+
"step": 4200
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 6.112115732368897,
|
| 1243 |
+
"grad_norm": 0.5590214729309082,
|
| 1244 |
+
"learning_rate": 4.546531302876481e-06,
|
| 1245 |
+
"loss": 0.0095,
|
| 1246 |
+
"step": 4225
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 6.1482820976491865,
|
| 1250 |
+
"grad_norm": 1.0092438459396362,
|
| 1251 |
+
"learning_rate": 4.504230118443317e-06,
|
| 1252 |
+
"loss": 0.0132,
|
| 1253 |
+
"step": 4250
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 6.184448462929476,
|
| 1257 |
+
"grad_norm": 0.68355393409729,
|
| 1258 |
+
"learning_rate": 4.461928934010153e-06,
|
| 1259 |
+
"loss": 0.0073,
|
| 1260 |
+
"step": 4275
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 6.220614828209765,
|
| 1264 |
+
"grad_norm": 0.854202926158905,
|
| 1265 |
+
"learning_rate": 4.4196277495769884e-06,
|
| 1266 |
+
"loss": 0.0208,
|
| 1267 |
+
"step": 4300
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 6.256781193490054,
|
| 1271 |
+
"grad_norm": 0.3510783314704895,
|
| 1272 |
+
"learning_rate": 4.377326565143825e-06,
|
| 1273 |
+
"loss": 0.0112,
|
| 1274 |
+
"step": 4325
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 6.292947558770344,
|
| 1278 |
+
"grad_norm": 0.36134082078933716,
|
| 1279 |
+
"learning_rate": 4.33502538071066e-06,
|
| 1280 |
+
"loss": 0.0136,
|
| 1281 |
+
"step": 4350
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 6.329113924050633,
|
| 1285 |
+
"grad_norm": 1.4613643884658813,
|
| 1286 |
+
"learning_rate": 4.292724196277496e-06,
|
| 1287 |
+
"loss": 0.0109,
|
| 1288 |
+
"step": 4375
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 6.365280289330922,
|
| 1292 |
+
"grad_norm": 0.7109003067016602,
|
| 1293 |
+
"learning_rate": 4.2504230118443315e-06,
|
| 1294 |
+
"loss": 0.0179,
|
| 1295 |
+
"step": 4400
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 6.401446654611211,
|
| 1299 |
+
"grad_norm": 0.505111813545227,
|
| 1300 |
+
"learning_rate": 4.208121827411168e-06,
|
| 1301 |
+
"loss": 0.0094,
|
| 1302 |
+
"step": 4425
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 6.437613019891501,
|
| 1306 |
+
"grad_norm": 1.2527493238449097,
|
| 1307 |
+
"learning_rate": 4.165820642978004e-06,
|
| 1308 |
+
"loss": 0.0195,
|
| 1309 |
+
"step": 4450
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 6.47377938517179,
|
| 1313 |
+
"grad_norm": 1.6427637338638306,
|
| 1314 |
+
"learning_rate": 4.123519458544839e-06,
|
| 1315 |
+
"loss": 0.0102,
|
| 1316 |
+
"step": 4475
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 6.509945750452079,
|
| 1320 |
+
"grad_norm": 0.24043454229831696,
|
| 1321 |
+
"learning_rate": 4.081218274111675e-06,
|
| 1322 |
+
"loss": 0.0181,
|
| 1323 |
+
"step": 4500
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 6.5461121157323685,
|
| 1327 |
+
"grad_norm": 0.9318906664848328,
|
| 1328 |
+
"learning_rate": 4.0389170896785115e-06,
|
| 1329 |
+
"loss": 0.011,
|
| 1330 |
+
"step": 4525
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 6.582278481012658,
|
| 1334 |
+
"grad_norm": 0.46116721630096436,
|
| 1335 |
+
"learning_rate": 3.996615905245348e-06,
|
| 1336 |
+
"loss": 0.0208,
|
| 1337 |
+
"step": 4550
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 6.618444846292948,
|
| 1341 |
+
"grad_norm": 1.371081829071045,
|
| 1342 |
+
"learning_rate": 3.954314720812183e-06,
|
| 1343 |
+
"loss": 0.0092,
|
| 1344 |
+
"step": 4575
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 6.654611211573237,
|
| 1348 |
+
"grad_norm": 1.1018704175949097,
|
| 1349 |
+
"learning_rate": 3.912013536379019e-06,
|
| 1350 |
+
"loss": 0.014,
|
| 1351 |
+
"step": 4600
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 6.6907775768535265,
|
| 1355 |
+
"grad_norm": 1.1697708368301392,
|
| 1356 |
+
"learning_rate": 3.869712351945855e-06,
|
| 1357 |
+
"loss": 0.0126,
|
| 1358 |
+
"step": 4625
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 6.726943942133816,
|
| 1362 |
+
"grad_norm": 0.5917466878890991,
|
| 1363 |
+
"learning_rate": 3.827411167512691e-06,
|
| 1364 |
+
"loss": 0.0168,
|
| 1365 |
+
"step": 4650
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 6.763110307414105,
|
| 1369 |
+
"grad_norm": 0.7262076735496521,
|
| 1370 |
+
"learning_rate": 3.785109983079526e-06,
|
| 1371 |
+
"loss": 0.0077,
|
| 1372 |
+
"step": 4675
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 6.799276672694394,
|
| 1376 |
+
"grad_norm": 0.16528570652008057,
|
| 1377 |
+
"learning_rate": 3.7428087986463623e-06,
|
| 1378 |
+
"loss": 0.0142,
|
| 1379 |
+
"step": 4700
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 6.8354430379746836,
|
| 1383 |
+
"grad_norm": 2.227794885635376,
|
| 1384 |
+
"learning_rate": 3.7005076142131985e-06,
|
| 1385 |
+
"loss": 0.0077,
|
| 1386 |
+
"step": 4725
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 6.871609403254973,
|
| 1390 |
+
"grad_norm": 0.8085231184959412,
|
| 1391 |
+
"learning_rate": 3.6582064297800342e-06,
|
| 1392 |
+
"loss": 0.0154,
|
| 1393 |
+
"step": 4750
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 6.907775768535262,
|
| 1397 |
+
"grad_norm": 1.447237253189087,
|
| 1398 |
+
"learning_rate": 3.61590524534687e-06,
|
| 1399 |
+
"loss": 0.0087,
|
| 1400 |
+
"step": 4775
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 6.943942133815551,
|
| 1404 |
+
"grad_norm": 0.4593859910964966,
|
| 1405 |
+
"learning_rate": 3.5736040609137058e-06,
|
| 1406 |
+
"loss": 0.0131,
|
| 1407 |
+
"step": 4800
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 6.980108499095841,
|
| 1411 |
+
"grad_norm": 1.0502936840057373,
|
| 1412 |
+
"learning_rate": 3.531302876480542e-06,
|
| 1413 |
+
"loss": 0.0129,
|
| 1414 |
+
"step": 4825
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 6.998915009041592,
|
| 1418 |
+
"eval_loss": 0.22768299281597137,
|
| 1419 |
+
"eval_runtime": 48990.2447,
|
| 1420 |
+
"eval_samples_per_second": 0.194,
|
| 1421 |
+
"eval_steps_per_second": 0.024,
|
| 1422 |
+
"eval_wer": 0.11143497757847534,
|
| 1423 |
+
"step": 4838
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 7.01627486437613,
|
| 1427 |
+
"grad_norm": 0.29863396286964417,
|
| 1428 |
+
"learning_rate": 3.4890016920473773e-06,
|
| 1429 |
+
"loss": 0.0131,
|
| 1430 |
+
"step": 4850
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 7.052441229656419,
|
| 1434 |
+
"grad_norm": 1.6121262311935425,
|
| 1435 |
+
"learning_rate": 3.4467005076142135e-06,
|
| 1436 |
+
"loss": 0.0067,
|
| 1437 |
+
"step": 4875
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 7.0886075949367084,
|
| 1441 |
+
"grad_norm": 0.682396411895752,
|
| 1442 |
+
"learning_rate": 3.4043993231810496e-06,
|
| 1443 |
+
"loss": 0.0104,
|
| 1444 |
+
"step": 4900
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 7.124773960216999,
|
| 1448 |
+
"grad_norm": 1.451640009880066,
|
| 1449 |
+
"learning_rate": 3.362098138747885e-06,
|
| 1450 |
+
"loss": 0.0061,
|
| 1451 |
+
"step": 4925
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 7.160940325497288,
|
| 1455 |
+
"grad_norm": 1.0819129943847656,
|
| 1456 |
+
"learning_rate": 3.319796954314721e-06,
|
| 1457 |
+
"loss": 0.0098,
|
| 1458 |
+
"step": 4950
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 7.197106690777577,
|
| 1462 |
+
"grad_norm": 0.8002307415008545,
|
| 1463 |
+
"learning_rate": 3.277495769881557e-06,
|
| 1464 |
+
"loss": 0.0076,
|
| 1465 |
+
"step": 4975
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 7.233273056057866,
|
| 1469 |
+
"grad_norm": 0.34655046463012695,
|
| 1470 |
+
"learning_rate": 3.235194585448393e-06,
|
| 1471 |
+
"loss": 0.0119,
|
| 1472 |
+
"step": 5000
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 7.269439421338156,
|
| 1476 |
+
"grad_norm": 0.7425087690353394,
|
| 1477 |
+
"learning_rate": 3.1928934010152284e-06,
|
| 1478 |
+
"loss": 0.0051,
|
| 1479 |
+
"step": 5025
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 7.305605786618445,
|
| 1483 |
+
"grad_norm": 0.48938363790512085,
|
| 1484 |
+
"learning_rate": 3.1505922165820646e-06,
|
| 1485 |
+
"loss": 0.0103,
|
| 1486 |
+
"step": 5050
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 7.341772151898734,
|
| 1490 |
+
"grad_norm": 1.4168592691421509,
|
| 1491 |
+
"learning_rate": 3.1082910321489e-06,
|
| 1492 |
+
"loss": 0.0053,
|
| 1493 |
+
"step": 5075
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 7.3779385171790235,
|
| 1497 |
+
"grad_norm": 0.7530602812767029,
|
| 1498 |
+
"learning_rate": 3.065989847715736e-06,
|
| 1499 |
+
"loss": 0.0124,
|
| 1500 |
+
"step": 5100
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 7.414104882459313,
|
| 1504 |
+
"grad_norm": 0.81525719165802,
|
| 1505 |
+
"learning_rate": 3.0236886632825723e-06,
|
| 1506 |
+
"loss": 0.0047,
|
| 1507 |
+
"step": 5125
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 7.450271247739602,
|
| 1511 |
+
"grad_norm": 0.5934809446334839,
|
| 1512 |
+
"learning_rate": 2.981387478849408e-06,
|
| 1513 |
+
"loss": 0.0073,
|
| 1514 |
+
"step": 5150
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 7.486437613019891,
|
| 1518 |
+
"grad_norm": 1.5411005020141602,
|
| 1519 |
+
"learning_rate": 2.939086294416244e-06,
|
| 1520 |
+
"loss": 0.0076,
|
| 1521 |
+
"step": 5175
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 7.522603978300181,
|
| 1525 |
+
"grad_norm": 3.0440473556518555,
|
| 1526 |
+
"learning_rate": 2.8967851099830796e-06,
|
| 1527 |
+
"loss": 0.0119,
|
| 1528 |
+
"step": 5200
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 7.55877034358047,
|
| 1532 |
+
"grad_norm": 0.15616849064826965,
|
| 1533 |
+
"learning_rate": 2.8544839255499158e-06,
|
| 1534 |
+
"loss": 0.0071,
|
| 1535 |
+
"step": 5225
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 7.594936708860759,
|
| 1539 |
+
"grad_norm": 0.6418558955192566,
|
| 1540 |
+
"learning_rate": 2.812182741116751e-06,
|
| 1541 |
+
"loss": 0.0125,
|
| 1542 |
+
"step": 5250
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 7.631103074141048,
|
| 1546 |
+
"grad_norm": 1.245208978652954,
|
| 1547 |
+
"learning_rate": 2.7698815566835873e-06,
|
| 1548 |
+
"loss": 0.009,
|
| 1549 |
+
"step": 5275
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 7.6672694394213385,
|
| 1553 |
+
"grad_norm": 0.6857640743255615,
|
| 1554 |
+
"learning_rate": 2.7275803722504235e-06,
|
| 1555 |
+
"loss": 0.0143,
|
| 1556 |
+
"step": 5300
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 7.703435804701628,
|
| 1560 |
+
"grad_norm": 0.6764671802520752,
|
| 1561 |
+
"learning_rate": 2.6852791878172592e-06,
|
| 1562 |
+
"loss": 0.0081,
|
| 1563 |
+
"step": 5325
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 7.739602169981917,
|
| 1567 |
+
"grad_norm": 0.6338791251182556,
|
| 1568 |
+
"learning_rate": 2.642978003384095e-06,
|
| 1569 |
+
"loss": 0.0128,
|
| 1570 |
+
"step": 5350
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 7.775768535262206,
|
| 1574 |
+
"grad_norm": 1.4596633911132812,
|
| 1575 |
+
"learning_rate": 2.6006768189509308e-06,
|
| 1576 |
+
"loss": 0.0075,
|
| 1577 |
+
"step": 5375
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 7.811934900542496,
|
| 1581 |
+
"grad_norm": 0.14179235696792603,
|
| 1582 |
+
"learning_rate": 2.558375634517767e-06,
|
| 1583 |
+
"loss": 0.0142,
|
| 1584 |
+
"step": 5400
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 7.848101265822785,
|
| 1588 |
+
"grad_norm": 0.7647730112075806,
|
| 1589 |
+
"learning_rate": 2.5160744500846023e-06,
|
| 1590 |
+
"loss": 0.0068,
|
| 1591 |
+
"step": 5425
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 7.884267631103074,
|
| 1595 |
+
"grad_norm": 0.1921517550945282,
|
| 1596 |
+
"learning_rate": 2.4737732656514385e-06,
|
| 1597 |
+
"loss": 0.0078,
|
| 1598 |
+
"step": 5450
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 7.920433996383363,
|
| 1602 |
+
"grad_norm": 1.385634183883667,
|
| 1603 |
+
"learning_rate": 2.4314720812182742e-06,
|
| 1604 |
+
"loss": 0.0067,
|
| 1605 |
+
"step": 5475
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 7.956600361663653,
|
| 1609 |
+
"grad_norm": 0.08770379424095154,
|
| 1610 |
+
"learning_rate": 2.38917089678511e-06,
|
| 1611 |
+
"loss": 0.007,
|
| 1612 |
+
"step": 5500
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 7.992766726943942,
|
| 1616 |
+
"grad_norm": 2.0286593437194824,
|
| 1617 |
+
"learning_rate": 2.346869712351946e-06,
|
| 1618 |
+
"loss": 0.0091,
|
| 1619 |
+
"step": 5525
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 8.0,
|
| 1623 |
+
"eval_loss": 0.2551898956298828,
|
| 1624 |
+
"eval_runtime": 49008.1266,
|
| 1625 |
+
"eval_samples_per_second": 0.194,
|
| 1626 |
+
"eval_steps_per_second": 0.024,
|
| 1627 |
+
"eval_wer": 0.10667841127482383,
|
| 1628 |
+
"step": 5530
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 8.028933092224232,
|
| 1632 |
+
"grad_norm": 0.40499478578567505,
|
| 1633 |
+
"learning_rate": 2.304568527918782e-06,
|
| 1634 |
+
"loss": 0.011,
|
| 1635 |
+
"step": 5550
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 8.065099457504521,
|
| 1639 |
+
"grad_norm": 1.5098172426223755,
|
| 1640 |
+
"learning_rate": 2.2622673434856177e-06,
|
| 1641 |
+
"loss": 0.0074,
|
| 1642 |
+
"step": 5575
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 8.10126582278481,
|
| 1646 |
+
"grad_norm": 2.1074013710021973,
|
| 1647 |
+
"learning_rate": 2.219966159052454e-06,
|
| 1648 |
+
"loss": 0.0079,
|
| 1649 |
+
"step": 5600
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 8.1374321880651,
|
| 1653 |
+
"grad_norm": 1.3512495756149292,
|
| 1654 |
+
"learning_rate": 2.1776649746192896e-06,
|
| 1655 |
+
"loss": 0.0049,
|
| 1656 |
+
"step": 5625
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 8.17359855334539,
|
| 1660 |
+
"grad_norm": 0.27601808309555054,
|
| 1661 |
+
"learning_rate": 2.1353637901861254e-06,
|
| 1662 |
+
"loss": 0.0037,
|
| 1663 |
+
"step": 5650
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 8.209764918625678,
|
| 1667 |
+
"grad_norm": 0.38552117347717285,
|
| 1668 |
+
"learning_rate": 2.093062605752961e-06,
|
| 1669 |
+
"loss": 0.0054,
|
| 1670 |
+
"step": 5675
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 8.245931283905968,
|
| 1674 |
+
"grad_norm": 0.6290847063064575,
|
| 1675 |
+
"learning_rate": 2.050761421319797e-06,
|
| 1676 |
+
"loss": 0.0075,
|
| 1677 |
+
"step": 5700
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 8.282097649186257,
|
| 1681 |
+
"grad_norm": 0.34640949964523315,
|
| 1682 |
+
"learning_rate": 2.008460236886633e-06,
|
| 1683 |
+
"loss": 0.0038,
|
| 1684 |
+
"step": 5725
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 8.318264014466546,
|
| 1688 |
+
"grad_norm": 0.47082650661468506,
|
| 1689 |
+
"learning_rate": 1.966159052453469e-06,
|
| 1690 |
+
"loss": 0.0072,
|
| 1691 |
+
"step": 5750
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 8.354430379746836,
|
| 1695 |
+
"grad_norm": 0.7971699237823486,
|
| 1696 |
+
"learning_rate": 1.9238578680203046e-06,
|
| 1697 |
+
"loss": 0.0074,
|
| 1698 |
+
"step": 5775
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 8.390596745027125,
|
| 1702 |
+
"grad_norm": 1.3511334657669067,
|
| 1703 |
+
"learning_rate": 1.8815566835871408e-06,
|
| 1704 |
+
"loss": 0.0065,
|
| 1705 |
+
"step": 5800
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 8.426763110307414,
|
| 1709 |
+
"grad_norm": 0.04755210131406784,
|
| 1710 |
+
"learning_rate": 1.8392554991539766e-06,
|
| 1711 |
+
"loss": 0.0054,
|
| 1712 |
+
"step": 5825
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 8.462929475587703,
|
| 1716 |
+
"grad_norm": 0.7288948893547058,
|
| 1717 |
+
"learning_rate": 1.7969543147208123e-06,
|
| 1718 |
+
"loss": 0.0075,
|
| 1719 |
+
"step": 5850
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 8.499095840867993,
|
| 1723 |
+
"grad_norm": 2.372441291809082,
|
| 1724 |
+
"learning_rate": 1.7546531302876483e-06,
|
| 1725 |
+
"loss": 0.0071,
|
| 1726 |
+
"step": 5875
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 8.535262206148282,
|
| 1730 |
+
"grad_norm": 0.029580192640423775,
|
| 1731 |
+
"learning_rate": 1.712351945854484e-06,
|
| 1732 |
+
"loss": 0.0057,
|
| 1733 |
+
"step": 5900
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 8.571428571428571,
|
| 1737 |
+
"grad_norm": 3.4771149158477783,
|
| 1738 |
+
"learning_rate": 1.6700507614213198e-06,
|
| 1739 |
+
"loss": 0.0063,
|
| 1740 |
+
"step": 5925
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 8.60759493670886,
|
| 1744 |
+
"grad_norm": 0.47946012020111084,
|
| 1745 |
+
"learning_rate": 1.6277495769881558e-06,
|
| 1746 |
+
"loss": 0.0095,
|
| 1747 |
+
"step": 5950
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 8.64376130198915,
|
| 1751 |
+
"grad_norm": 1.6082369089126587,
|
| 1752 |
+
"learning_rate": 1.5854483925549915e-06,
|
| 1753 |
+
"loss": 0.0061,
|
| 1754 |
+
"step": 5975
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 8.679927667269439,
|
| 1758 |
+
"grad_norm": 0.8396530747413635,
|
| 1759 |
+
"learning_rate": 1.5431472081218277e-06,
|
| 1760 |
+
"loss": 0.005,
|
| 1761 |
+
"step": 6000
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 8.716094032549728,
|
| 1765 |
+
"grad_norm": 0.683506965637207,
|
| 1766 |
+
"learning_rate": 1.5008460236886635e-06,
|
| 1767 |
+
"loss": 0.0042,
|
| 1768 |
+
"step": 6025
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 8.752260397830018,
|
| 1772 |
+
"grad_norm": 0.23053808510303497,
|
| 1773 |
+
"learning_rate": 1.4585448392554992e-06,
|
| 1774 |
+
"loss": 0.0085,
|
| 1775 |
+
"step": 6050
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 8.788426763110307,
|
| 1779 |
+
"grad_norm": 1.4277093410491943,
|
| 1780 |
+
"learning_rate": 1.4162436548223352e-06,
|
| 1781 |
+
"loss": 0.0057,
|
| 1782 |
+
"step": 6075
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 8.824593128390596,
|
| 1786 |
+
"grad_norm": 0.8418155908584595,
|
| 1787 |
+
"learning_rate": 1.373942470389171e-06,
|
| 1788 |
+
"loss": 0.0085,
|
| 1789 |
+
"step": 6100
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 8.860759493670885,
|
| 1793 |
+
"grad_norm": 0.16715645790100098,
|
| 1794 |
+
"learning_rate": 1.3316412859560067e-06,
|
| 1795 |
+
"loss": 0.0039,
|
| 1796 |
+
"step": 6125
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 8.896925858951175,
|
| 1800 |
+
"grad_norm": 1.1576683521270752,
|
| 1801 |
+
"learning_rate": 1.2893401015228427e-06,
|
| 1802 |
+
"loss": 0.0056,
|
| 1803 |
+
"step": 6150
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 8.933092224231466,
|
| 1807 |
+
"grad_norm": 4.466509819030762,
|
| 1808 |
+
"learning_rate": 1.2470389170896787e-06,
|
| 1809 |
+
"loss": 0.008,
|
| 1810 |
+
"step": 6175
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 8.969258589511753,
|
| 1814 |
+
"grad_norm": 0.09006431698799133,
|
| 1815 |
+
"learning_rate": 1.2047377326565144e-06,
|
| 1816 |
+
"loss": 0.0063,
|
| 1817 |
+
"step": 6200
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 8.999638336347196,
|
| 1821 |
+
"eval_loss": 0.25654470920562744,
|
| 1822 |
+
"eval_runtime": 49994.6568,
|
| 1823 |
+
"eval_samples_per_second": 0.19,
|
| 1824 |
+
"eval_steps_per_second": 0.024,
|
| 1825 |
+
"eval_wer": 0.10539718129404228,
|
| 1826 |
+
"step": 6221
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 9.005424954792044,
|
| 1830 |
+
"grad_norm": 0.04043040797114372,
|
| 1831 |
+
"learning_rate": 1.1624365482233504e-06,
|
| 1832 |
+
"loss": 0.0099,
|
| 1833 |
+
"step": 6225
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 9.041591320072333,
|
| 1837 |
+
"grad_norm": 0.31738966703414917,
|
| 1838 |
+
"learning_rate": 1.1201353637901864e-06,
|
| 1839 |
+
"loss": 0.0031,
|
| 1840 |
+
"step": 6250
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 9.077757685352623,
|
| 1844 |
+
"grad_norm": 0.8276641368865967,
|
| 1845 |
+
"learning_rate": 1.0778341793570221e-06,
|
| 1846 |
+
"loss": 0.0064,
|
| 1847 |
+
"step": 6275
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 9.113924050632912,
|
| 1851 |
+
"grad_norm": 0.3469080626964569,
|
| 1852 |
+
"learning_rate": 1.035532994923858e-06,
|
| 1853 |
+
"loss": 0.0032,
|
| 1854 |
+
"step": 6300
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 9.150090415913201,
|
| 1858 |
+
"grad_norm": 0.6140421628952026,
|
| 1859 |
+
"learning_rate": 9.932318104906939e-07,
|
| 1860 |
+
"loss": 0.0088,
|
| 1861 |
+
"step": 6325
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 9.18625678119349,
|
| 1865 |
+
"grad_norm": 1.1443873643875122,
|
| 1866 |
+
"learning_rate": 9.509306260575297e-07,
|
| 1867 |
+
"loss": 0.0032,
|
| 1868 |
+
"step": 6350
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 9.22242314647378,
|
| 1872 |
+
"grad_norm": 0.21027229726314545,
|
| 1873 |
+
"learning_rate": 9.086294416243656e-07,
|
| 1874 |
+
"loss": 0.0105,
|
| 1875 |
+
"step": 6375
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 9.258589511754069,
|
| 1879 |
+
"grad_norm": 0.2232620120048523,
|
| 1880 |
+
"learning_rate": 8.663282571912015e-07,
|
| 1881 |
+
"loss": 0.0047,
|
| 1882 |
+
"step": 6400
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 9.294755877034358,
|
| 1886 |
+
"grad_norm": 0.5159393548965454,
|
| 1887 |
+
"learning_rate": 8.240270727580372e-07,
|
| 1888 |
+
"loss": 0.0069,
|
| 1889 |
+
"step": 6425
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 9.330922242314648,
|
| 1893 |
+
"grad_norm": 0.5648090839385986,
|
| 1894 |
+
"learning_rate": 7.817258883248732e-07,
|
| 1895 |
+
"loss": 0.0031,
|
| 1896 |
+
"step": 6450
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 9.367088607594937,
|
| 1900 |
+
"grad_norm": 0.24160954356193542,
|
| 1901 |
+
"learning_rate": 7.394247038917091e-07,
|
| 1902 |
+
"loss": 0.0076,
|
| 1903 |
+
"step": 6475
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 9.403254972875226,
|
| 1907 |
+
"grad_norm": 0.21812845766544342,
|
| 1908 |
+
"learning_rate": 6.971235194585449e-07,
|
| 1909 |
+
"loss": 0.0036,
|
| 1910 |
+
"step": 6500
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 9.439421338155515,
|
| 1914 |
+
"grad_norm": 0.3133035898208618,
|
| 1915 |
+
"learning_rate": 6.548223350253807e-07,
|
| 1916 |
+
"loss": 0.0086,
|
| 1917 |
+
"step": 6525
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 9.475587703435805,
|
| 1921 |
+
"grad_norm": 0.16532699763774872,
|
| 1922 |
+
"learning_rate": 6.125211505922167e-07,
|
| 1923 |
+
"loss": 0.0029,
|
| 1924 |
+
"step": 6550
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 9.511754068716094,
|
| 1928 |
+
"grad_norm": 0.5517584681510925,
|
| 1929 |
+
"learning_rate": 5.702199661590524e-07,
|
| 1930 |
+
"loss": 0.0096,
|
| 1931 |
+
"step": 6575
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 9.547920433996383,
|
| 1935 |
+
"grad_norm": 0.2974016070365906,
|
| 1936 |
+
"learning_rate": 5.279187817258884e-07,
|
| 1937 |
+
"loss": 0.0018,
|
| 1938 |
+
"step": 6600
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 9.584086799276673,
|
| 1942 |
+
"grad_norm": 0.7380957007408142,
|
| 1943 |
+
"learning_rate": 4.856175972927242e-07,
|
| 1944 |
+
"loss": 0.0091,
|
| 1945 |
+
"step": 6625
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 9.620253164556962,
|
| 1949 |
+
"grad_norm": 0.2390100508928299,
|
| 1950 |
+
"learning_rate": 4.433164128595601e-07,
|
| 1951 |
+
"loss": 0.0024,
|
| 1952 |
+
"step": 6650
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 9.656419529837251,
|
| 1956 |
+
"grad_norm": 0.22240670025348663,
|
| 1957 |
+
"learning_rate": 4.0101522842639594e-07,
|
| 1958 |
+
"loss": 0.0054,
|
| 1959 |
+
"step": 6675
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 9.69258589511754,
|
| 1963 |
+
"grad_norm": 0.03914798051118851,
|
| 1964 |
+
"learning_rate": 3.5871404399323186e-07,
|
| 1965 |
+
"loss": 0.0029,
|
| 1966 |
+
"step": 6700
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 9.72875226039783,
|
| 1970 |
+
"grad_norm": 0.5454278588294983,
|
| 1971 |
+
"learning_rate": 3.1641285956006767e-07,
|
| 1972 |
+
"loss": 0.0053,
|
| 1973 |
+
"step": 6725
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 9.764918625678119,
|
| 1977 |
+
"grad_norm": 0.21154353022575378,
|
| 1978 |
+
"learning_rate": 2.741116751269036e-07,
|
| 1979 |
+
"loss": 0.0024,
|
| 1980 |
+
"step": 6750
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 9.801084990958408,
|
| 1984 |
+
"grad_norm": 0.47212332487106323,
|
| 1985 |
+
"learning_rate": 2.3181049069373945e-07,
|
| 1986 |
+
"loss": 0.0081,
|
| 1987 |
+
"step": 6775
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 9.837251356238697,
|
| 1991 |
+
"grad_norm": 0.09104613214731216,
|
| 1992 |
+
"learning_rate": 1.8950930626057532e-07,
|
| 1993 |
+
"loss": 0.0048,
|
| 1994 |
+
"step": 6800
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 9.873417721518987,
|
| 1998 |
+
"grad_norm": 0.35835790634155273,
|
| 1999 |
+
"learning_rate": 1.4720812182741118e-07,
|
| 2000 |
+
"loss": 0.0085,
|
| 2001 |
+
"step": 6825
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 9.909584086799276,
|
| 2005 |
+
"grad_norm": 0.13305124640464783,
|
| 2006 |
+
"learning_rate": 1.0490693739424705e-07,
|
| 2007 |
+
"loss": 0.0028,
|
| 2008 |
+
"step": 6850
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 9.945750452079565,
|
| 2012 |
+
"grad_norm": 0.4060196280479431,
|
| 2013 |
+
"learning_rate": 6.260575296108291e-08,
|
| 2014 |
+
"loss": 0.0096,
|
| 2015 |
+
"step": 6875
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 9.981916817359856,
|
| 2019 |
+
"grad_norm": 1.0367088317871094,
|
| 2020 |
+
"learning_rate": 2.0304568527918784e-08,
|
| 2021 |
+
"loss": 0.0019,
|
| 2022 |
+
"step": 6900
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 9.99638336347197,
|
| 2026 |
+
"eval_loss": 0.26707419753074646,
|
| 2027 |
+
"eval_runtime": 51487.4721,
|
| 2028 |
+
"eval_samples_per_second": 0.184,
|
| 2029 |
+
"eval_steps_per_second": 0.023,
|
| 2030 |
+
"eval_wer": 0.10419602818705957,
|
| 2031 |
+
"step": 6910
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 9.99638336347197,
|
| 2035 |
+
"step": 6910,
|
| 2036 |
+
"total_flos": 7.51132515607511e+20,
|
| 2037 |
+
"train_loss": 0.05691683961323822,
|
| 2038 |
+
"train_runtime": 650193.1322,
|
| 2039 |
+
"train_samples_per_second": 0.34,
|
| 2040 |
+
"train_steps_per_second": 0.011
|
| 2041 |
}
|
| 2042 |
],
|
| 2043 |
"logging_steps": 25,
|
| 2044 |
+
"max_steps": 6910,
|
| 2045 |
"num_input_tokens_seen": 0,
|
| 2046 |
+
"num_train_epochs": 10,
|
| 2047 |
"save_steps": 500,
|
| 2048 |
"stateful_callbacks": {
|
| 2049 |
"TrainerControl": {
|
|
|
|
| 2057 |
"attributes": {}
|
| 2058 |
}
|
| 2059 |
},
|
| 2060 |
+
"total_flos": 7.51132515607511e+20,
|
| 2061 |
+
"train_batch_size": 8,
|
| 2062 |
"trial_name": null,
|
| 2063 |
"trial_params": null
|
| 2064 |
}
|