End of training
Browse files- README.md +5 -4
- all_results.json +12 -11
- eval_results.json +7 -6
- runs/Nov29_01-15-14_19115951741f/events.out.tfevents.1732848929.19115951741f.2594.1 +3 -0
- train_results.json +6 -5
- trainer_state.json +801 -579
README.md
CHANGED
|
@@ -3,19 +3,20 @@ library_name: transformers
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: openai/whisper-medium
|
| 5 |
tags:
|
|
|
|
| 6 |
- generated_from_trainer
|
| 7 |
datasets:
|
| 8 |
- audiofolder
|
| 9 |
metrics:
|
| 10 |
- wer
|
| 11 |
model-index:
|
| 12 |
-
- name:
|
| 13 |
results:
|
| 14 |
- task:
|
| 15 |
name: Automatic Speech Recognition
|
| 16 |
type: automatic-speech-recognition
|
| 17 |
dataset:
|
| 18 |
-
name: audiofolder
|
| 19 |
type: audiofolder
|
| 20 |
config: nan-tw
|
| 21 |
split: test
|
|
@@ -29,9 +30,9 @@ model-index:
|
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 30 |
should probably proofread and complete it, then remove this comment. -->
|
| 31 |
|
| 32 |
-
#
|
| 33 |
|
| 34 |
-
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the audiofolder dataset.
|
| 35 |
It achieves the following results on the evaluation set:
|
| 36 |
- Loss: 0.0141
|
| 37 |
- Model Preparation Time: 0.0121
|
|
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: openai/whisper-medium
|
| 5 |
tags:
|
| 6 |
+
- whisper-event
|
| 7 |
- generated_from_trainer
|
| 8 |
datasets:
|
| 9 |
- audiofolder
|
| 10 |
metrics:
|
| 11 |
- wer
|
| 12 |
model-index:
|
| 13 |
+
- name: Whisper medium nan-tw common voice
|
| 14 |
results:
|
| 15 |
- task:
|
| 16 |
name: Automatic Speech Recognition
|
| 17 |
type: automatic-speech-recognition
|
| 18 |
dataset:
|
| 19 |
+
name: audiofolder nan-tw
|
| 20 |
type: audiofolder
|
| 21 |
config: nan-tw
|
| 22 |
split: test
|
|
|
|
| 30 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 31 |
should probably proofread and complete it, then remove this comment. -->
|
| 32 |
|
| 33 |
+
# Whisper medium nan-tw common voice
|
| 34 |
|
| 35 |
+
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the audiofolder nan-tw dataset.
|
| 36 |
It achieves the following results on the evaluation set:
|
| 37 |
- Loss: 0.0141
|
| 38 |
- Model Preparation Time: 0.0121
|
all_results.json
CHANGED
|
@@ -1,14 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_cer":
|
| 4 |
-
"eval_loss":
|
| 5 |
"eval_model_preparation_time": 0.0121,
|
| 6 |
-
"eval_runtime":
|
| 7 |
-
"eval_samples_per_second": 2.
|
| 8 |
-
"eval_steps_per_second": 1.
|
| 9 |
-
"eval_wer":
|
| 10 |
-
"
|
| 11 |
-
"
|
| 12 |
-
"
|
| 13 |
-
"
|
|
|
|
| 14 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.2776,
|
| 3 |
+
"eval_cer": 0.9523809523809524,
|
| 4 |
+
"eval_loss": 0.014132725074887276,
|
| 5 |
"eval_model_preparation_time": 0.0121,
|
| 6 |
+
"eval_runtime": 117.6465,
|
| 7 |
+
"eval_samples_per_second": 2.55,
|
| 8 |
+
"eval_steps_per_second": 1.275,
|
| 9 |
+
"eval_wer": 0.9615384615384616,
|
| 10 |
+
"total_flos": 1.02060490752e+19,
|
| 11 |
+
"train_loss": 0.5477467903137208,
|
| 12 |
+
"train_runtime": 5409.143,
|
| 13 |
+
"train_samples_per_second": 1.849,
|
| 14 |
+
"train_steps_per_second": 0.924
|
| 15 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
|
|
|
| 4 |
"eval_model_preparation_time": 0.0121,
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second": 2.
|
| 7 |
-
"eval_steps_per_second": 1.
|
| 8 |
-
"eval_wer":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.2776,
|
| 3 |
+
"eval_cer": 0.9523809523809524,
|
| 4 |
+
"eval_loss": 0.014132725074887276,
|
| 5 |
"eval_model_preparation_time": 0.0121,
|
| 6 |
+
"eval_runtime": 117.6465,
|
| 7 |
+
"eval_samples_per_second": 2.55,
|
| 8 |
+
"eval_steps_per_second": 1.275,
|
| 9 |
+
"eval_wer": 0.9615384615384616
|
| 10 |
}
|
runs/Nov29_01-15-14_19115951741f/events.out.tfevents.1732848929.19115951741f.2594.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0d4f0b1c0b3df21e47fca5d383f121ca6e36b0441045ee54496e4495233100aa
|
| 3 |
+
size 519
|
train_results.json
CHANGED
|
@@ -1,7 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
|
|
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.2776,
|
| 3 |
+
"total_flos": 1.02060490752e+19,
|
| 4 |
+
"train_loss": 0.5477467903137208,
|
| 5 |
+
"train_runtime": 5409.143,
|
| 6 |
+
"train_samples_per_second": 1.849,
|
| 7 |
+
"train_steps_per_second": 0.924
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1275 +1,1497 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
|
|
|
| 5 |
"global_step": 5000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
-
"epoch": 0.
|
|
|
|
| 12 |
"learning_rate": 4.2000000000000006e-07,
|
| 13 |
-
"loss": 2.
|
| 14 |
"step": 25
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.01,
|
|
|
|
| 18 |
"learning_rate": 9.200000000000001e-07,
|
| 19 |
-
"loss": 2.
|
| 20 |
"step": 50
|
| 21 |
},
|
| 22 |
{
|
| 23 |
-
"epoch": 0.
|
|
|
|
| 24 |
"learning_rate": 1.42e-06,
|
| 25 |
-
"loss": 1.
|
| 26 |
"step": 75
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.02,
|
|
|
|
| 30 |
"learning_rate": 1.9200000000000003e-06,
|
| 31 |
-
"loss": 1.
|
| 32 |
"step": 100
|
| 33 |
},
|
| 34 |
{
|
| 35 |
-
"epoch": 0.
|
| 36 |
-
"
|
| 37 |
-
"
|
|
|
|
| 38 |
"step": 125
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.03,
|
| 42 |
-
"
|
| 43 |
-
"
|
|
|
|
| 44 |
"step": 150
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"
|
| 49 |
-
"
|
|
|
|
| 50 |
"step": 175
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.04,
|
| 54 |
-
"
|
| 55 |
-
"
|
|
|
|
| 56 |
"step": 200
|
| 57 |
},
|
| 58 |
{
|
| 59 |
-
"epoch": 0.
|
| 60 |
-
"
|
| 61 |
-
"
|
|
|
|
| 62 |
"step": 225
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.05,
|
| 66 |
-
"
|
| 67 |
-
"
|
|
|
|
| 68 |
"step": 250
|
| 69 |
},
|
| 70 |
{
|
| 71 |
-
"epoch": 0.
|
| 72 |
-
"
|
| 73 |
-
"
|
|
|
|
| 74 |
"step": 275
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 0.06,
|
| 78 |
-
"
|
| 79 |
-
"
|
|
|
|
| 80 |
"step": 300
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch": 0.
|
|
|
|
| 84 |
"learning_rate": 6.4000000000000006e-06,
|
| 85 |
-
"loss": 1.
|
| 86 |
"step": 325
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.07,
|
|
|
|
| 90 |
"learning_rate": 6.9e-06,
|
| 91 |
-
"loss": 1.
|
| 92 |
"step": 350
|
| 93 |
},
|
| 94 |
{
|
| 95 |
-
"epoch": 0.
|
|
|
|
| 96 |
"learning_rate": 7.4e-06,
|
| 97 |
-
"loss": 1.
|
| 98 |
"step": 375
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.08,
|
|
|
|
| 102 |
"learning_rate": 7.9e-06,
|
| 103 |
-
"loss": 1.
|
| 104 |
"step": 400
|
| 105 |
},
|
| 106 |
{
|
| 107 |
-
"epoch": 0.
|
|
|
|
| 108 |
"learning_rate": 8.400000000000001e-06,
|
| 109 |
-
"loss": 1.
|
| 110 |
"step": 425
|
| 111 |
},
|
| 112 |
{
|
| 113 |
"epoch": 0.09,
|
|
|
|
| 114 |
"learning_rate": 8.900000000000001e-06,
|
| 115 |
-
"loss": 1.
|
| 116 |
"step": 450
|
| 117 |
},
|
| 118 |
{
|
| 119 |
-
"epoch": 0.
|
|
|
|
| 120 |
"learning_rate": 9.4e-06,
|
| 121 |
-
"loss": 1.
|
| 122 |
"step": 475
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.1,
|
|
|
|
| 126 |
"learning_rate": 9.9e-06,
|
| 127 |
-
"loss": 1.
|
| 128 |
"step": 500
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
|
|
|
| 132 |
"learning_rate": 9.955555555555556e-06,
|
| 133 |
-
"loss": 1.
|
| 134 |
"step": 525
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 0.11,
|
|
|
|
| 138 |
"learning_rate": 9.9e-06,
|
| 139 |
-
"loss": 1.
|
| 140 |
"step": 550
|
| 141 |
},
|
| 142 |
{
|
| 143 |
-
"epoch": 0.
|
|
|
|
| 144 |
"learning_rate": 9.844444444444446e-06,
|
| 145 |
-
"loss": 1.
|
| 146 |
"step": 575
|
| 147 |
},
|
| 148 |
{
|
| 149 |
"epoch": 0.12,
|
|
|
|
| 150 |
"learning_rate": 9.78888888888889e-06,
|
| 151 |
-
"loss": 1.
|
| 152 |
"step": 600
|
| 153 |
},
|
| 154 |
{
|
| 155 |
-
"epoch": 0.
|
|
|
|
| 156 |
"learning_rate": 9.733333333333334e-06,
|
| 157 |
-
"loss": 1.
|
| 158 |
"step": 625
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 0.13,
|
|
|
|
| 162 |
"learning_rate": 9.677777777777778e-06,
|
| 163 |
-
"loss": 1.
|
| 164 |
"step": 650
|
| 165 |
},
|
| 166 |
{
|
| 167 |
-
"epoch": 0.
|
|
|
|
| 168 |
"learning_rate": 9.622222222222222e-06,
|
| 169 |
-
"loss": 1.
|
| 170 |
"step": 675
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.14,
|
|
|
|
| 174 |
"learning_rate": 9.566666666666668e-06,
|
| 175 |
-
"loss": 1.
|
| 176 |
"step": 700
|
| 177 |
},
|
| 178 |
{
|
| 179 |
-
"epoch": 0.
|
|
|
|
| 180 |
"learning_rate": 9.511111111111112e-06,
|
| 181 |
-
"loss": 1.
|
| 182 |
"step": 725
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 0.15,
|
|
|
|
| 186 |
"learning_rate": 9.455555555555557e-06,
|
| 187 |
-
"loss":
|
| 188 |
"step": 750
|
| 189 |
},
|
| 190 |
{
|
| 191 |
-
"epoch": 0.
|
|
|
|
| 192 |
"learning_rate": 9.4e-06,
|
| 193 |
-
"loss": 1.
|
| 194 |
"step": 775
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch":
|
|
|
|
| 198 |
"learning_rate": 9.344444444444446e-06,
|
| 199 |
-
"loss":
|
| 200 |
"step": 800
|
| 201 |
},
|
| 202 |
{
|
| 203 |
-
"epoch":
|
|
|
|
| 204 |
"learning_rate": 9.28888888888889e-06,
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 825
|
| 207 |
},
|
| 208 |
{
|
| 209 |
-
"epoch":
|
|
|
|
| 210 |
"learning_rate": 9.233333333333334e-06,
|
| 211 |
-
"loss":
|
| 212 |
"step": 850
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch":
|
|
|
|
| 216 |
"learning_rate": 9.17777777777778e-06,
|
| 217 |
-
"loss":
|
| 218 |
"step": 875
|
| 219 |
},
|
| 220 |
{
|
| 221 |
-
"epoch":
|
|
|
|
| 222 |
"learning_rate": 9.122222222222223e-06,
|
| 223 |
-
"loss":
|
| 224 |
"step": 900
|
| 225 |
},
|
| 226 |
{
|
| 227 |
-
"epoch":
|
|
|
|
| 228 |
"learning_rate": 9.066666666666667e-06,
|
| 229 |
-
"loss": 0.
|
| 230 |
"step": 925
|
| 231 |
},
|
| 232 |
{
|
| 233 |
-
"epoch":
|
|
|
|
| 234 |
"learning_rate": 9.011111111111111e-06,
|
| 235 |
-
"loss":
|
| 236 |
"step": 950
|
| 237 |
},
|
| 238 |
{
|
| 239 |
-
"epoch":
|
|
|
|
| 240 |
"learning_rate": 8.955555555555555e-06,
|
| 241 |
-
"loss":
|
| 242 |
"step": 975
|
| 243 |
},
|
| 244 |
{
|
| 245 |
-
"epoch":
|
|
|
|
| 246 |
"learning_rate": 8.900000000000001e-06,
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 1000
|
| 249 |
},
|
| 250 |
{
|
| 251 |
-
"epoch":
|
| 252 |
-
"eval_cer":
|
| 253 |
-
"eval_loss":
|
| 254 |
-
"
|
| 255 |
-
"
|
| 256 |
-
"
|
| 257 |
-
"
|
|
|
|
| 258 |
"step": 1000
|
| 259 |
},
|
| 260 |
{
|
| 261 |
-
"epoch":
|
|
|
|
| 262 |
"learning_rate": 8.844444444444445e-06,
|
| 263 |
-
"loss":
|
| 264 |
"step": 1025
|
| 265 |
},
|
| 266 |
{
|
| 267 |
-
"epoch":
|
|
|
|
| 268 |
"learning_rate": 8.788888888888891e-06,
|
| 269 |
-
"loss":
|
| 270 |
"step": 1050
|
| 271 |
},
|
| 272 |
{
|
| 273 |
-
"epoch":
|
|
|
|
| 274 |
"learning_rate": 8.733333333333333e-06,
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 1075
|
| 277 |
},
|
| 278 |
{
|
| 279 |
-
"epoch":
|
|
|
|
| 280 |
"learning_rate": 8.677777777777779e-06,
|
| 281 |
-
"loss":
|
| 282 |
"step": 1100
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch":
|
|
|
|
| 286 |
"learning_rate": 8.622222222222223e-06,
|
| 287 |
-
"loss": 0.
|
| 288 |
"step": 1125
|
| 289 |
},
|
| 290 |
{
|
| 291 |
-
"epoch":
|
|
|
|
| 292 |
"learning_rate": 8.566666666666667e-06,
|
| 293 |
-
"loss": 0.
|
| 294 |
"step": 1150
|
| 295 |
},
|
| 296 |
{
|
| 297 |
-
"epoch":
|
|
|
|
| 298 |
"learning_rate": 8.511111111111113e-06,
|
| 299 |
-
"loss":
|
| 300 |
"step": 1175
|
| 301 |
},
|
| 302 |
{
|
| 303 |
-
"epoch":
|
|
|
|
| 304 |
"learning_rate": 8.455555555555555e-06,
|
| 305 |
-
"loss": 0.
|
| 306 |
"step": 1200
|
| 307 |
},
|
| 308 |
{
|
| 309 |
-
"epoch":
|
|
|
|
| 310 |
"learning_rate": 8.400000000000001e-06,
|
| 311 |
-
"loss":
|
| 312 |
"step": 1225
|
| 313 |
},
|
| 314 |
{
|
| 315 |
-
"epoch":
|
|
|
|
| 316 |
"learning_rate": 8.344444444444445e-06,
|
| 317 |
-
"loss":
|
| 318 |
"step": 1250
|
| 319 |
},
|
| 320 |
{
|
| 321 |
-
"epoch":
|
|
|
|
| 322 |
"learning_rate": 8.288888888888889e-06,
|
| 323 |
-
"loss":
|
| 324 |
"step": 1275
|
| 325 |
},
|
| 326 |
{
|
| 327 |
-
"epoch":
|
|
|
|
| 328 |
"learning_rate": 8.233333333333335e-06,
|
| 329 |
-
"loss":
|
| 330 |
"step": 1300
|
| 331 |
},
|
| 332 |
{
|
| 333 |
-
"epoch":
|
|
|
|
| 334 |
"learning_rate": 8.177777777777779e-06,
|
| 335 |
-
"loss": 0.
|
| 336 |
"step": 1325
|
| 337 |
},
|
| 338 |
{
|
| 339 |
-
"epoch":
|
|
|
|
| 340 |
"learning_rate": 8.122222222222223e-06,
|
| 341 |
-
"loss": 0.
|
| 342 |
"step": 1350
|
| 343 |
},
|
| 344 |
{
|
| 345 |
-
"epoch":
|
|
|
|
| 346 |
"learning_rate": 8.066666666666667e-06,
|
| 347 |
-
"loss": 0.
|
| 348 |
"step": 1375
|
| 349 |
},
|
| 350 |
{
|
| 351 |
-
"epoch":
|
|
|
|
| 352 |
"learning_rate": 8.011111111111113e-06,
|
| 353 |
-
"loss": 0.
|
| 354 |
"step": 1400
|
| 355 |
},
|
| 356 |
{
|
| 357 |
-
"epoch":
|
|
|
|
| 358 |
"learning_rate": 7.955555555555557e-06,
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 1425
|
| 361 |
},
|
| 362 |
{
|
| 363 |
-
"epoch":
|
| 364 |
-
"
|
| 365 |
-
"
|
|
|
|
| 366 |
"step": 1450
|
| 367 |
},
|
| 368 |
{
|
| 369 |
-
"epoch":
|
| 370 |
-
"
|
| 371 |
-
"
|
|
|
|
| 372 |
"step": 1475
|
| 373 |
},
|
| 374 |
{
|
| 375 |
-
"epoch":
|
| 376 |
-
"
|
| 377 |
-
"
|
|
|
|
| 378 |
"step": 1500
|
| 379 |
},
|
| 380 |
{
|
| 381 |
-
"epoch":
|
| 382 |
-
"
|
| 383 |
-
"
|
|
|
|
| 384 |
"step": 1525
|
| 385 |
},
|
| 386 |
{
|
| 387 |
-
"epoch":
|
| 388 |
-
"
|
| 389 |
-
"
|
|
|
|
| 390 |
"step": 1550
|
| 391 |
},
|
| 392 |
{
|
| 393 |
-
"epoch":
|
| 394 |
-
"
|
| 395 |
-
"
|
|
|
|
| 396 |
"step": 1575
|
| 397 |
},
|
| 398 |
{
|
| 399 |
-
"epoch":
|
| 400 |
-
"
|
| 401 |
-
"
|
|
|
|
| 402 |
"step": 1600
|
| 403 |
},
|
| 404 |
{
|
| 405 |
-
"epoch":
|
| 406 |
-
"
|
| 407 |
-
"
|
|
|
|
| 408 |
"step": 1625
|
| 409 |
},
|
| 410 |
{
|
| 411 |
-
"epoch":
|
| 412 |
-
"
|
| 413 |
-
"
|
|
|
|
| 414 |
"step": 1650
|
| 415 |
},
|
| 416 |
{
|
| 417 |
-
"epoch":
|
| 418 |
-
"
|
| 419 |
-
"
|
|
|
|
| 420 |
"step": 1675
|
| 421 |
},
|
| 422 |
{
|
| 423 |
-
"epoch":
|
| 424 |
-
"
|
| 425 |
-
"
|
|
|
|
| 426 |
"step": 1700
|
| 427 |
},
|
| 428 |
{
|
| 429 |
-
"epoch":
|
| 430 |
-
"
|
| 431 |
-
"
|
|
|
|
| 432 |
"step": 1725
|
| 433 |
},
|
| 434 |
{
|
| 435 |
-
"epoch":
|
| 436 |
-
"
|
| 437 |
-
"
|
|
|
|
| 438 |
"step": 1750
|
| 439 |
},
|
| 440 |
{
|
| 441 |
-
"epoch":
|
| 442 |
-
"
|
| 443 |
-
"
|
|
|
|
| 444 |
"step": 1775
|
| 445 |
},
|
| 446 |
{
|
| 447 |
-
"epoch":
|
| 448 |
-
"
|
| 449 |
-
"
|
|
|
|
| 450 |
"step": 1800
|
| 451 |
},
|
| 452 |
{
|
| 453 |
-
"epoch":
|
| 454 |
-
"
|
| 455 |
-
"
|
|
|
|
| 456 |
"step": 1825
|
| 457 |
},
|
| 458 |
{
|
| 459 |
-
"epoch":
|
| 460 |
-
"
|
| 461 |
-
"
|
|
|
|
| 462 |
"step": 1850
|
| 463 |
},
|
| 464 |
{
|
| 465 |
-
"epoch":
|
| 466 |
-
"
|
| 467 |
-
"
|
|
|
|
| 468 |
"step": 1875
|
| 469 |
},
|
| 470 |
{
|
| 471 |
-
"epoch":
|
| 472 |
-
"
|
| 473 |
-
"
|
|
|
|
| 474 |
"step": 1900
|
| 475 |
},
|
| 476 |
{
|
| 477 |
-
"epoch":
|
| 478 |
-
"
|
| 479 |
-
"
|
|
|
|
| 480 |
"step": 1925
|
| 481 |
},
|
| 482 |
{
|
| 483 |
-
"epoch":
|
| 484 |
-
"
|
| 485 |
-
"
|
|
|
|
| 486 |
"step": 1950
|
| 487 |
},
|
| 488 |
{
|
| 489 |
-
"epoch":
|
| 490 |
-
"
|
| 491 |
-
"
|
|
|
|
| 492 |
"step": 1975
|
| 493 |
},
|
| 494 |
{
|
| 495 |
-
"epoch":
|
| 496 |
-
"
|
| 497 |
-
"
|
|
|
|
| 498 |
"step": 2000
|
| 499 |
},
|
| 500 |
{
|
| 501 |
-
"epoch":
|
| 502 |
-
"eval_cer":
|
| 503 |
-
"eval_loss":
|
| 504 |
-
"
|
| 505 |
-
"
|
| 506 |
-
"
|
| 507 |
-
"
|
|
|
|
| 508 |
"step": 2000
|
| 509 |
},
|
| 510 |
{
|
| 511 |
-
"epoch":
|
| 512 |
-
"
|
| 513 |
-
"
|
|
|
|
| 514 |
"step": 2025
|
| 515 |
},
|
| 516 |
{
|
| 517 |
-
"epoch":
|
| 518 |
-
"
|
| 519 |
-
"
|
|
|
|
| 520 |
"step": 2050
|
| 521 |
},
|
| 522 |
{
|
| 523 |
-
"epoch":
|
| 524 |
-
"
|
| 525 |
-
"
|
|
|
|
| 526 |
"step": 2075
|
| 527 |
},
|
| 528 |
{
|
| 529 |
-
"epoch":
|
| 530 |
-
"
|
| 531 |
-
"
|
|
|
|
| 532 |
"step": 2100
|
| 533 |
},
|
| 534 |
{
|
| 535 |
-
"epoch":
|
| 536 |
-
"
|
| 537 |
-
"
|
|
|
|
| 538 |
"step": 2125
|
| 539 |
},
|
| 540 |
{
|
| 541 |
-
"epoch":
|
| 542 |
-
"
|
| 543 |
-
"
|
|
|
|
| 544 |
"step": 2150
|
| 545 |
},
|
| 546 |
{
|
| 547 |
-
"epoch":
|
| 548 |
-
"
|
| 549 |
-
"
|
|
|
|
| 550 |
"step": 2175
|
| 551 |
},
|
| 552 |
{
|
| 553 |
-
"epoch":
|
| 554 |
-
"
|
| 555 |
-
"
|
|
|
|
| 556 |
"step": 2200
|
| 557 |
},
|
| 558 |
{
|
| 559 |
-
"epoch":
|
| 560 |
-
"
|
| 561 |
-
"
|
|
|
|
| 562 |
"step": 2225
|
| 563 |
},
|
| 564 |
{
|
| 565 |
-
"epoch":
|
| 566 |
-
"
|
| 567 |
-
"
|
|
|
|
| 568 |
"step": 2250
|
| 569 |
},
|
| 570 |
{
|
| 571 |
-
"epoch":
|
| 572 |
-
"
|
| 573 |
-
"
|
|
|
|
| 574 |
"step": 2275
|
| 575 |
},
|
| 576 |
{
|
| 577 |
-
"epoch":
|
| 578 |
-
"
|
| 579 |
-
"
|
|
|
|
| 580 |
"step": 2300
|
| 581 |
},
|
| 582 |
{
|
| 583 |
-
"epoch":
|
| 584 |
-
"
|
| 585 |
-
"
|
|
|
|
| 586 |
"step": 2325
|
| 587 |
},
|
| 588 |
{
|
| 589 |
-
"epoch":
|
| 590 |
-
"
|
| 591 |
-
"
|
|
|
|
| 592 |
"step": 2350
|
| 593 |
},
|
| 594 |
{
|
| 595 |
-
"epoch":
|
| 596 |
-
"
|
| 597 |
-
"
|
|
|
|
| 598 |
"step": 2375
|
| 599 |
},
|
| 600 |
{
|
| 601 |
-
"epoch":
|
| 602 |
-
"
|
| 603 |
-
"
|
|
|
|
| 604 |
"step": 2400
|
| 605 |
},
|
| 606 |
{
|
| 607 |
-
"epoch":
|
| 608 |
-
"
|
| 609 |
-
"
|
|
|
|
| 610 |
"step": 2425
|
| 611 |
},
|
| 612 |
{
|
| 613 |
-
"epoch":
|
| 614 |
-
"
|
| 615 |
-
"
|
|
|
|
| 616 |
"step": 2450
|
| 617 |
},
|
| 618 |
{
|
| 619 |
-
"epoch":
|
| 620 |
-
"
|
| 621 |
-
"
|
|
|
|
| 622 |
"step": 2475
|
| 623 |
},
|
| 624 |
{
|
| 625 |
-
"epoch":
|
| 626 |
-
"
|
| 627 |
-
"
|
|
|
|
| 628 |
"step": 2500
|
| 629 |
},
|
| 630 |
{
|
| 631 |
-
"epoch":
|
| 632 |
-
"
|
| 633 |
-
"
|
|
|
|
| 634 |
"step": 2525
|
| 635 |
},
|
| 636 |
{
|
| 637 |
-
"epoch":
|
| 638 |
-
"
|
| 639 |
-
"
|
|
|
|
| 640 |
"step": 2550
|
| 641 |
},
|
| 642 |
{
|
| 643 |
-
"epoch":
|
| 644 |
-
"
|
| 645 |
-
"
|
|
|
|
| 646 |
"step": 2575
|
| 647 |
},
|
| 648 |
{
|
| 649 |
-
"epoch":
|
| 650 |
-
"
|
| 651 |
-
"
|
|
|
|
| 652 |
"step": 2600
|
| 653 |
},
|
| 654 |
{
|
| 655 |
-
"epoch":
|
| 656 |
-
"
|
| 657 |
-
"
|
|
|
|
| 658 |
"step": 2625
|
| 659 |
},
|
| 660 |
{
|
| 661 |
-
"epoch":
|
| 662 |
-
"
|
| 663 |
-
"
|
|
|
|
| 664 |
"step": 2650
|
| 665 |
},
|
| 666 |
{
|
| 667 |
-
"epoch":
|
| 668 |
-
"
|
| 669 |
-
"
|
|
|
|
| 670 |
"step": 2675
|
| 671 |
},
|
| 672 |
{
|
| 673 |
-
"epoch":
|
| 674 |
-
"
|
| 675 |
-
"
|
|
|
|
| 676 |
"step": 2700
|
| 677 |
},
|
| 678 |
{
|
| 679 |
-
"epoch":
|
| 680 |
-
"
|
| 681 |
-
"
|
|
|
|
| 682 |
"step": 2725
|
| 683 |
},
|
| 684 |
{
|
| 685 |
-
"epoch":
|
| 686 |
-
"
|
| 687 |
-
"
|
|
|
|
| 688 |
"step": 2750
|
| 689 |
},
|
| 690 |
{
|
| 691 |
-
"epoch":
|
| 692 |
-
"
|
| 693 |
-
"
|
|
|
|
| 694 |
"step": 2775
|
| 695 |
},
|
| 696 |
{
|
| 697 |
-
"epoch":
|
| 698 |
-
"
|
| 699 |
-
"
|
|
|
|
| 700 |
"step": 2800
|
| 701 |
},
|
| 702 |
{
|
| 703 |
-
"epoch":
|
| 704 |
-
"
|
| 705 |
-
"
|
|
|
|
| 706 |
"step": 2825
|
| 707 |
},
|
| 708 |
{
|
| 709 |
-
"epoch":
|
| 710 |
-
"
|
| 711 |
-
"
|
|
|
|
| 712 |
"step": 2850
|
| 713 |
},
|
| 714 |
{
|
| 715 |
-
"epoch":
|
| 716 |
-
"
|
| 717 |
-
"
|
|
|
|
| 718 |
"step": 2875
|
| 719 |
},
|
| 720 |
{
|
| 721 |
-
"epoch":
|
| 722 |
-
"
|
| 723 |
-
"
|
|
|
|
| 724 |
"step": 2900
|
| 725 |
},
|
| 726 |
{
|
| 727 |
-
"epoch":
|
| 728 |
-
"
|
| 729 |
-
"
|
|
|
|
| 730 |
"step": 2925
|
| 731 |
},
|
| 732 |
{
|
| 733 |
-
"epoch":
|
| 734 |
-
"
|
| 735 |
-
"
|
|
|
|
| 736 |
"step": 2950
|
| 737 |
},
|
| 738 |
{
|
| 739 |
-
"epoch":
|
| 740 |
-
"
|
| 741 |
-
"
|
|
|
|
| 742 |
"step": 2975
|
| 743 |
},
|
| 744 |
{
|
| 745 |
-
"epoch":
|
| 746 |
-
"
|
| 747 |
-
"
|
|
|
|
| 748 |
"step": 3000
|
| 749 |
},
|
| 750 |
{
|
| 751 |
-
"epoch":
|
| 752 |
-
"eval_cer":
|
| 753 |
-
"eval_loss":
|
| 754 |
-
"
|
| 755 |
-
"
|
| 756 |
-
"
|
| 757 |
-
"
|
|
|
|
| 758 |
"step": 3000
|
| 759 |
},
|
| 760 |
{
|
| 761 |
-
"epoch":
|
| 762 |
-
"
|
| 763 |
-
"
|
|
|
|
| 764 |
"step": 3025
|
| 765 |
},
|
| 766 |
{
|
| 767 |
-
"epoch":
|
| 768 |
-
"
|
| 769 |
-
"
|
|
|
|
| 770 |
"step": 3050
|
| 771 |
},
|
| 772 |
{
|
| 773 |
-
"epoch":
|
| 774 |
-
"
|
| 775 |
-
"
|
|
|
|
| 776 |
"step": 3075
|
| 777 |
},
|
| 778 |
{
|
| 779 |
-
"epoch":
|
| 780 |
-
"
|
| 781 |
-
"
|
|
|
|
| 782 |
"step": 3100
|
| 783 |
},
|
| 784 |
{
|
| 785 |
-
"epoch":
|
| 786 |
-
"
|
| 787 |
-
"
|
|
|
|
| 788 |
"step": 3125
|
| 789 |
},
|
| 790 |
{
|
| 791 |
-
"epoch":
|
| 792 |
-
"
|
| 793 |
-
"
|
|
|
|
| 794 |
"step": 3150
|
| 795 |
},
|
| 796 |
{
|
| 797 |
-
"epoch":
|
| 798 |
-
"
|
| 799 |
-
"
|
|
|
|
| 800 |
"step": 3175
|
| 801 |
},
|
| 802 |
{
|
| 803 |
-
"epoch":
|
| 804 |
-
"
|
| 805 |
-
"
|
|
|
|
| 806 |
"step": 3200
|
| 807 |
},
|
| 808 |
{
|
| 809 |
-
"epoch":
|
| 810 |
-
"
|
| 811 |
-
"
|
|
|
|
| 812 |
"step": 3225
|
| 813 |
},
|
| 814 |
{
|
| 815 |
-
"epoch":
|
| 816 |
-
"
|
| 817 |
-
"
|
|
|
|
| 818 |
"step": 3250
|
| 819 |
},
|
| 820 |
{
|
| 821 |
-
"epoch":
|
| 822 |
-
"
|
| 823 |
-
"
|
|
|
|
| 824 |
"step": 3275
|
| 825 |
},
|
| 826 |
{
|
| 827 |
-
"epoch":
|
| 828 |
-
"
|
| 829 |
-
"
|
|
|
|
| 830 |
"step": 3300
|
| 831 |
},
|
| 832 |
{
|
| 833 |
-
"epoch":
|
| 834 |
-
"
|
| 835 |
-
"
|
|
|
|
| 836 |
"step": 3325
|
| 837 |
},
|
| 838 |
{
|
| 839 |
-
"epoch":
|
| 840 |
-
"
|
| 841 |
-
"
|
|
|
|
| 842 |
"step": 3350
|
| 843 |
},
|
| 844 |
{
|
| 845 |
-
"epoch":
|
| 846 |
-
"
|
| 847 |
-
"
|
|
|
|
| 848 |
"step": 3375
|
| 849 |
},
|
| 850 |
{
|
| 851 |
-
"epoch":
|
| 852 |
-
"
|
| 853 |
-
"
|
|
|
|
| 854 |
"step": 3400
|
| 855 |
},
|
| 856 |
{
|
| 857 |
-
"epoch":
|
| 858 |
-
"
|
| 859 |
-
"
|
|
|
|
| 860 |
"step": 3425
|
| 861 |
},
|
| 862 |
{
|
| 863 |
-
"epoch":
|
| 864 |
-
"
|
| 865 |
-
"
|
|
|
|
| 866 |
"step": 3450
|
| 867 |
},
|
| 868 |
{
|
| 869 |
-
"epoch":
|
| 870 |
-
"
|
| 871 |
-
"
|
|
|
|
| 872 |
"step": 3475
|
| 873 |
},
|
| 874 |
{
|
| 875 |
-
"epoch":
|
| 876 |
-
"
|
| 877 |
-
"
|
|
|
|
| 878 |
"step": 3500
|
| 879 |
},
|
| 880 |
{
|
| 881 |
-
"epoch":
|
| 882 |
-
"
|
| 883 |
-
"
|
|
|
|
| 884 |
"step": 3525
|
| 885 |
},
|
| 886 |
{
|
| 887 |
-
"epoch":
|
| 888 |
-
"
|
| 889 |
-
"
|
|
|
|
| 890 |
"step": 3550
|
| 891 |
},
|
| 892 |
{
|
| 893 |
-
"epoch":
|
| 894 |
-
"
|
| 895 |
-
"
|
|
|
|
| 896 |
"step": 3575
|
| 897 |
},
|
| 898 |
{
|
| 899 |
-
"epoch":
|
| 900 |
-
"
|
| 901 |
-
"
|
|
|
|
| 902 |
"step": 3600
|
| 903 |
},
|
| 904 |
{
|
| 905 |
-
"epoch":
|
| 906 |
-
"
|
| 907 |
-
"
|
|
|
|
| 908 |
"step": 3625
|
| 909 |
},
|
| 910 |
{
|
| 911 |
-
"epoch":
|
| 912 |
-
"
|
| 913 |
-
"
|
|
|
|
| 914 |
"step": 3650
|
| 915 |
},
|
| 916 |
{
|
| 917 |
-
"epoch":
|
| 918 |
-
"
|
| 919 |
-
"
|
|
|
|
| 920 |
"step": 3675
|
| 921 |
},
|
| 922 |
{
|
| 923 |
-
"epoch":
|
| 924 |
-
"
|
| 925 |
-
"
|
|
|
|
| 926 |
"step": 3700
|
| 927 |
},
|
| 928 |
{
|
| 929 |
-
"epoch":
|
| 930 |
-
"
|
| 931 |
-
"
|
|
|
|
| 932 |
"step": 3725
|
| 933 |
},
|
| 934 |
{
|
| 935 |
-
"epoch":
|
| 936 |
-
"
|
| 937 |
-
"
|
|
|
|
| 938 |
"step": 3750
|
| 939 |
},
|
| 940 |
{
|
| 941 |
-
"epoch":
|
| 942 |
-
"
|
| 943 |
-
"
|
|
|
|
| 944 |
"step": 3775
|
| 945 |
},
|
| 946 |
{
|
| 947 |
-
"epoch":
|
| 948 |
-
"
|
| 949 |
-
"
|
|
|
|
| 950 |
"step": 3800
|
| 951 |
},
|
| 952 |
{
|
| 953 |
-
"epoch":
|
| 954 |
-
"
|
| 955 |
-
"
|
|
|
|
| 956 |
"step": 3825
|
| 957 |
},
|
| 958 |
{
|
| 959 |
-
"epoch":
|
| 960 |
-
"
|
| 961 |
-
"
|
|
|
|
| 962 |
"step": 3850
|
| 963 |
},
|
| 964 |
{
|
| 965 |
-
"epoch":
|
| 966 |
-
"
|
| 967 |
-
"
|
|
|
|
| 968 |
"step": 3875
|
| 969 |
},
|
| 970 |
{
|
| 971 |
-
"epoch":
|
| 972 |
-
"
|
| 973 |
-
"
|
|
|
|
| 974 |
"step": 3900
|
| 975 |
},
|
| 976 |
{
|
| 977 |
-
"epoch":
|
| 978 |
-
"
|
| 979 |
-
"
|
|
|
|
| 980 |
"step": 3925
|
| 981 |
},
|
| 982 |
{
|
| 983 |
-
"epoch":
|
| 984 |
-
"
|
| 985 |
-
"
|
|
|
|
| 986 |
"step": 3950
|
| 987 |
},
|
| 988 |
{
|
| 989 |
-
"epoch":
|
| 990 |
-
"
|
| 991 |
-
"
|
|
|
|
| 992 |
"step": 3975
|
| 993 |
},
|
| 994 |
{
|
| 995 |
-
"epoch":
|
| 996 |
-
"
|
| 997 |
-
"
|
|
|
|
| 998 |
"step": 4000
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
-
"epoch":
|
| 1002 |
-
"eval_cer":
|
| 1003 |
-
"eval_loss": 0.
|
| 1004 |
-
"
|
| 1005 |
-
"
|
| 1006 |
-
"
|
| 1007 |
-
"
|
|
|
|
| 1008 |
"step": 4000
|
| 1009 |
},
|
| 1010 |
{
|
| 1011 |
-
"epoch":
|
| 1012 |
-
"
|
| 1013 |
-
"
|
|
|
|
| 1014 |
"step": 4025
|
| 1015 |
},
|
| 1016 |
{
|
| 1017 |
-
"epoch":
|
| 1018 |
-
"
|
| 1019 |
-
"
|
|
|
|
| 1020 |
"step": 4050
|
| 1021 |
},
|
| 1022 |
{
|
| 1023 |
-
"epoch":
|
| 1024 |
-
"
|
| 1025 |
-
"
|
|
|
|
| 1026 |
"step": 4075
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
-
"epoch":
|
| 1030 |
-
"
|
| 1031 |
-
"
|
|
|
|
| 1032 |
"step": 4100
|
| 1033 |
},
|
| 1034 |
{
|
| 1035 |
-
"epoch":
|
| 1036 |
-
"
|
| 1037 |
-
"
|
|
|
|
| 1038 |
"step": 4125
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
-
"epoch":
|
| 1042 |
-
"
|
| 1043 |
-
"
|
|
|
|
| 1044 |
"step": 4150
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
-
"epoch":
|
| 1048 |
-
"
|
| 1049 |
-
"
|
|
|
|
| 1050 |
"step": 4175
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
-
"epoch":
|
| 1054 |
-
"
|
| 1055 |
-
"
|
|
|
|
| 1056 |
"step": 4200
|
| 1057 |
},
|
| 1058 |
{
|
| 1059 |
-
"epoch":
|
| 1060 |
-
"
|
| 1061 |
-
"
|
|
|
|
| 1062 |
"step": 4225
|
| 1063 |
},
|
| 1064 |
{
|
| 1065 |
-
"epoch":
|
| 1066 |
-
"
|
| 1067 |
-
"
|
|
|
|
| 1068 |
"step": 4250
|
| 1069 |
},
|
| 1070 |
{
|
| 1071 |
-
"epoch":
|
| 1072 |
-
"
|
| 1073 |
-
"
|
|
|
|
| 1074 |
"step": 4275
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
-
"epoch":
|
| 1078 |
-
"
|
| 1079 |
-
"
|
|
|
|
| 1080 |
"step": 4300
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
-
"epoch":
|
| 1084 |
-
"
|
| 1085 |
-
"
|
|
|
|
| 1086 |
"step": 4325
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
-
"epoch":
|
| 1090 |
-
"
|
| 1091 |
-
"
|
|
|
|
| 1092 |
"step": 4350
|
| 1093 |
},
|
| 1094 |
{
|
| 1095 |
-
"epoch":
|
| 1096 |
-
"
|
| 1097 |
-
"
|
|
|
|
| 1098 |
"step": 4375
|
| 1099 |
},
|
| 1100 |
{
|
| 1101 |
-
"epoch":
|
| 1102 |
-
"
|
| 1103 |
-
"
|
|
|
|
| 1104 |
"step": 4400
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
-
"epoch":
|
| 1108 |
-
"
|
| 1109 |
-
"
|
|
|
|
| 1110 |
"step": 4425
|
| 1111 |
},
|
| 1112 |
{
|
| 1113 |
-
"epoch":
|
| 1114 |
-
"
|
| 1115 |
-
"
|
|
|
|
| 1116 |
"step": 4450
|
| 1117 |
},
|
| 1118 |
{
|
| 1119 |
-
"epoch":
|
| 1120 |
-
"
|
| 1121 |
-
"
|
|
|
|
| 1122 |
"step": 4475
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
-
"epoch":
|
| 1126 |
-
"
|
| 1127 |
-
"
|
|
|
|
| 1128 |
"step": 4500
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
-
"epoch":
|
| 1132 |
-
"
|
| 1133 |
-
"
|
|
|
|
| 1134 |
"step": 4525
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
-
"epoch":
|
| 1138 |
-
"
|
| 1139 |
-
"
|
|
|
|
| 1140 |
"step": 4550
|
| 1141 |
},
|
| 1142 |
{
|
| 1143 |
-
"epoch":
|
| 1144 |
-
"
|
| 1145 |
-
"
|
|
|
|
| 1146 |
"step": 4575
|
| 1147 |
},
|
| 1148 |
{
|
| 1149 |
-
"epoch":
|
| 1150 |
-
"
|
| 1151 |
-
"
|
|
|
|
| 1152 |
"step": 4600
|
| 1153 |
},
|
| 1154 |
{
|
| 1155 |
-
"epoch":
|
| 1156 |
-
"
|
| 1157 |
-
"
|
|
|
|
| 1158 |
"step": 4625
|
| 1159 |
},
|
| 1160 |
{
|
| 1161 |
-
"epoch":
|
| 1162 |
-
"
|
| 1163 |
-
"
|
|
|
|
| 1164 |
"step": 4650
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
-
"epoch":
|
| 1168 |
-
"
|
| 1169 |
-
"
|
|
|
|
| 1170 |
"step": 4675
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
-
"epoch":
|
| 1174 |
-
"
|
| 1175 |
-
"
|
|
|
|
| 1176 |
"step": 4700
|
| 1177 |
},
|
| 1178 |
{
|
| 1179 |
-
"epoch":
|
| 1180 |
-
"
|
| 1181 |
-
"
|
|
|
|
| 1182 |
"step": 4725
|
| 1183 |
},
|
| 1184 |
{
|
| 1185 |
-
"epoch":
|
| 1186 |
-
"
|
| 1187 |
-
"
|
|
|
|
| 1188 |
"step": 4750
|
| 1189 |
},
|
| 1190 |
{
|
| 1191 |
-
"epoch":
|
| 1192 |
-
"
|
| 1193 |
-
"
|
|
|
|
| 1194 |
"step": 4775
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
-
"epoch":
|
| 1198 |
-
"
|
| 1199 |
-
"
|
|
|
|
| 1200 |
"step": 4800
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
-
"epoch":
|
| 1204 |
-
"
|
| 1205 |
-
"
|
|
|
|
| 1206 |
"step": 4825
|
| 1207 |
},
|
| 1208 |
{
|
| 1209 |
-
"epoch":
|
| 1210 |
-
"
|
| 1211 |
-
"
|
|
|
|
| 1212 |
"step": 4850
|
| 1213 |
},
|
| 1214 |
{
|
| 1215 |
-
"epoch":
|
| 1216 |
-
"
|
| 1217 |
-
"
|
|
|
|
| 1218 |
"step": 4875
|
| 1219 |
},
|
| 1220 |
{
|
| 1221 |
-
"epoch":
|
| 1222 |
-
"
|
| 1223 |
-
"
|
|
|
|
| 1224 |
"step": 4900
|
| 1225 |
},
|
| 1226 |
{
|
| 1227 |
-
"epoch":
|
| 1228 |
-
"
|
| 1229 |
-
"
|
|
|
|
| 1230 |
"step": 4925
|
| 1231 |
},
|
| 1232 |
{
|
| 1233 |
-
"epoch":
|
| 1234 |
-
"
|
| 1235 |
-
"
|
|
|
|
| 1236 |
"step": 4950
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
-
"epoch":
|
| 1240 |
-
"
|
| 1241 |
-
"
|
|
|
|
| 1242 |
"step": 4975
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
-
"epoch":
|
| 1246 |
-
"
|
| 1247 |
-
"
|
|
|
|
| 1248 |
"step": 5000
|
| 1249 |
},
|
| 1250 |
{
|
| 1251 |
-
"epoch":
|
| 1252 |
-
"eval_cer":
|
| 1253 |
-
"eval_loss": 0.
|
| 1254 |
-
"
|
| 1255 |
-
"
|
| 1256 |
-
"
|
| 1257 |
-
"
|
|
|
|
| 1258 |
"step": 5000
|
| 1259 |
},
|
| 1260 |
{
|
| 1261 |
-
"epoch":
|
| 1262 |
"step": 5000,
|
| 1263 |
-
"total_flos": 1.
|
| 1264 |
-
"train_loss": 0.
|
| 1265 |
-
"train_runtime":
|
| 1266 |
-
"train_samples_per_second":
|
| 1267 |
-
"train_steps_per_second":
|
| 1268 |
}
|
| 1269 |
],
|
|
|
|
| 1270 |
"max_steps": 5000,
|
|
|
|
| 1271 |
"num_train_epochs": 9223372036854775807,
|
| 1272 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1273 |
"trial_name": null,
|
| 1274 |
"trial_params": null
|
| 1275 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.2776,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
"global_step": 5000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.005,
|
| 13 |
+
"grad_norm": 88.54288482666016,
|
| 14 |
"learning_rate": 4.2000000000000006e-07,
|
| 15 |
+
"loss": 2.5968,
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.01,
|
| 20 |
+
"grad_norm": 71.25262451171875,
|
| 21 |
"learning_rate": 9.200000000000001e-07,
|
| 22 |
+
"loss": 2.3534,
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.015,
|
| 27 |
+
"grad_norm": 80.52667999267578,
|
| 28 |
"learning_rate": 1.42e-06,
|
| 29 |
+
"loss": 1.9642,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.02,
|
| 34 |
+
"grad_norm": 57.064151763916016,
|
| 35 |
"learning_rate": 1.9200000000000003e-06,
|
| 36 |
+
"loss": 1.6321,
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.025,
|
| 41 |
+
"grad_norm": 50.983245849609375,
|
| 42 |
+
"learning_rate": 2.42e-06,
|
| 43 |
+
"loss": 1.5842,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.03,
|
| 48 |
+
"grad_norm": 52.22635269165039,
|
| 49 |
+
"learning_rate": 2.92e-06,
|
| 50 |
+
"loss": 1.4106,
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.035,
|
| 55 |
+
"grad_norm": 53.9218864440918,
|
| 56 |
+
"learning_rate": 3.4200000000000007e-06,
|
| 57 |
+
"loss": 1.417,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.04,
|
| 62 |
+
"grad_norm": 48.82598876953125,
|
| 63 |
+
"learning_rate": 3.920000000000001e-06,
|
| 64 |
+
"loss": 1.4496,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.045,
|
| 69 |
+
"grad_norm": 63.08296585083008,
|
| 70 |
+
"learning_rate": 4.42e-06,
|
| 71 |
+
"loss": 1.3191,
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.05,
|
| 76 |
+
"grad_norm": 41.98011016845703,
|
| 77 |
+
"learning_rate": 4.92e-06,
|
| 78 |
+
"loss": 1.0756,
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.055,
|
| 83 |
+
"grad_norm": 53.777217864990234,
|
| 84 |
+
"learning_rate": 5.420000000000001e-06,
|
| 85 |
+
"loss": 1.2538,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.06,
|
| 90 |
+
"grad_norm": 57.49655532836914,
|
| 91 |
+
"learning_rate": 5.92e-06,
|
| 92 |
+
"loss": 1.2225,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.065,
|
| 97 |
+
"grad_norm": 52.660003662109375,
|
| 98 |
"learning_rate": 6.4000000000000006e-06,
|
| 99 |
+
"loss": 1.2031,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.07,
|
| 104 |
+
"grad_norm": 45.03782272338867,
|
| 105 |
"learning_rate": 6.9e-06,
|
| 106 |
+
"loss": 1.2953,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.075,
|
| 111 |
+
"grad_norm": 49.226261138916016,
|
| 112 |
"learning_rate": 7.4e-06,
|
| 113 |
+
"loss": 1.0642,
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.08,
|
| 118 |
+
"grad_norm": 33.70058822631836,
|
| 119 |
"learning_rate": 7.9e-06,
|
| 120 |
+
"loss": 1.0962,
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.085,
|
| 125 |
+
"grad_norm": 47.71772766113281,
|
| 126 |
"learning_rate": 8.400000000000001e-06,
|
| 127 |
+
"loss": 1.3042,
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.09,
|
| 132 |
+
"grad_norm": 55.22367477416992,
|
| 133 |
"learning_rate": 8.900000000000001e-06,
|
| 134 |
+
"loss": 1.3794,
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.095,
|
| 139 |
+
"grad_norm": 62.34690475463867,
|
| 140 |
"learning_rate": 9.4e-06,
|
| 141 |
+
"loss": 1.1516,
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 0.1,
|
| 146 |
+
"grad_norm": 84.5876693725586,
|
| 147 |
"learning_rate": 9.9e-06,
|
| 148 |
+
"loss": 1.2527,
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.105,
|
| 153 |
+
"grad_norm": 53.2620735168457,
|
| 154 |
"learning_rate": 9.955555555555556e-06,
|
| 155 |
+
"loss": 1.1349,
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.11,
|
| 160 |
+
"grad_norm": 54.91654586791992,
|
| 161 |
"learning_rate": 9.9e-06,
|
| 162 |
+
"loss": 1.2107,
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 0.115,
|
| 167 |
+
"grad_norm": 43.851173400878906,
|
| 168 |
"learning_rate": 9.844444444444446e-06,
|
| 169 |
+
"loss": 1.4065,
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.12,
|
| 174 |
+
"grad_norm": 48.8751335144043,
|
| 175 |
"learning_rate": 9.78888888888889e-06,
|
| 176 |
+
"loss": 1.2255,
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 0.125,
|
| 181 |
+
"grad_norm": 58.15187454223633,
|
| 182 |
"learning_rate": 9.733333333333334e-06,
|
| 183 |
+
"loss": 1.1948,
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.13,
|
| 188 |
+
"grad_norm": 53.069602966308594,
|
| 189 |
"learning_rate": 9.677777777777778e-06,
|
| 190 |
+
"loss": 1.2878,
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 0.135,
|
| 195 |
+
"grad_norm": 45.524715423583984,
|
| 196 |
"learning_rate": 9.622222222222222e-06,
|
| 197 |
+
"loss": 1.1325,
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
"epoch": 0.14,
|
| 202 |
+
"grad_norm": 26.86398696899414,
|
| 203 |
"learning_rate": 9.566666666666668e-06,
|
| 204 |
+
"loss": 1.0762,
|
| 205 |
"step": 700
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 0.145,
|
| 209 |
+
"grad_norm": 48.63008499145508,
|
| 210 |
"learning_rate": 9.511111111111112e-06,
|
| 211 |
+
"loss": 1.279,
|
| 212 |
"step": 725
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.15,
|
| 216 |
+
"grad_norm": 50.58299255371094,
|
| 217 |
"learning_rate": 9.455555555555557e-06,
|
| 218 |
+
"loss": 1.0627,
|
| 219 |
"step": 750
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.155,
|
| 223 |
+
"grad_norm": 32.71992492675781,
|
| 224 |
"learning_rate": 9.4e-06,
|
| 225 |
+
"loss": 1.0438,
|
| 226 |
"step": 775
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.16,
|
| 230 |
+
"grad_norm": 42.12641143798828,
|
| 231 |
"learning_rate": 9.344444444444446e-06,
|
| 232 |
+
"loss": 1.1204,
|
| 233 |
"step": 800
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 0.165,
|
| 237 |
+
"grad_norm": 44.01963806152344,
|
| 238 |
"learning_rate": 9.28888888888889e-06,
|
| 239 |
+
"loss": 0.908,
|
| 240 |
"step": 825
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 0.17,
|
| 244 |
+
"grad_norm": 39.93177795410156,
|
| 245 |
"learning_rate": 9.233333333333334e-06,
|
| 246 |
+
"loss": 1.158,
|
| 247 |
"step": 850
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 0.175,
|
| 251 |
+
"grad_norm": 38.166259765625,
|
| 252 |
"learning_rate": 9.17777777777778e-06,
|
| 253 |
+
"loss": 1.2021,
|
| 254 |
"step": 875
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 0.18,
|
| 258 |
+
"grad_norm": 40.696678161621094,
|
| 259 |
"learning_rate": 9.122222222222223e-06,
|
| 260 |
+
"loss": 1.0054,
|
| 261 |
"step": 900
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 0.185,
|
| 265 |
+
"grad_norm": 48.79116439819336,
|
| 266 |
"learning_rate": 9.066666666666667e-06,
|
| 267 |
+
"loss": 0.9587,
|
| 268 |
"step": 925
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"epoch": 0.19,
|
| 272 |
+
"grad_norm": 39.57270431518555,
|
| 273 |
"learning_rate": 9.011111111111111e-06,
|
| 274 |
+
"loss": 1.1084,
|
| 275 |
"step": 950
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 0.195,
|
| 279 |
+
"grad_norm": 54.647216796875,
|
| 280 |
"learning_rate": 8.955555555555555e-06,
|
| 281 |
+
"loss": 1.0416,
|
| 282 |
"step": 975
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 0.2,
|
| 286 |
+
"grad_norm": 42.69215774536133,
|
| 287 |
"learning_rate": 8.900000000000001e-06,
|
| 288 |
+
"loss": 0.97,
|
| 289 |
"step": 1000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 0.2,
|
| 293 |
+
"eval_cer": 38.476190476190474,
|
| 294 |
+
"eval_loss": 0.7355929613113403,
|
| 295 |
+
"eval_model_preparation_time": 0.0121,
|
| 296 |
+
"eval_runtime": 116.5648,
|
| 297 |
+
"eval_samples_per_second": 2.574,
|
| 298 |
+
"eval_steps_per_second": 1.287,
|
| 299 |
+
"eval_wer": 38.17307692307692,
|
| 300 |
"step": 1000
|
| 301 |
},
|
| 302 |
{
|
| 303 |
+
"epoch": 0.205,
|
| 304 |
+
"grad_norm": 53.78874206542969,
|
| 305 |
"learning_rate": 8.844444444444445e-06,
|
| 306 |
+
"loss": 1.2455,
|
| 307 |
"step": 1025
|
| 308 |
},
|
| 309 |
{
|
| 310 |
+
"epoch": 0.21,
|
| 311 |
+
"grad_norm": 34.4659423828125,
|
| 312 |
"learning_rate": 8.788888888888891e-06,
|
| 313 |
+
"loss": 1.0515,
|
| 314 |
"step": 1050
|
| 315 |
},
|
| 316 |
{
|
| 317 |
+
"epoch": 0.215,
|
| 318 |
+
"grad_norm": 51.5352668762207,
|
| 319 |
"learning_rate": 8.733333333333333e-06,
|
| 320 |
+
"loss": 0.9412,
|
| 321 |
"step": 1075
|
| 322 |
},
|
| 323 |
{
|
| 324 |
+
"epoch": 0.22,
|
| 325 |
+
"grad_norm": 41.004310607910156,
|
| 326 |
"learning_rate": 8.677777777777779e-06,
|
| 327 |
+
"loss": 1.0831,
|
| 328 |
"step": 1100
|
| 329 |
},
|
| 330 |
{
|
| 331 |
+
"epoch": 0.225,
|
| 332 |
+
"grad_norm": 43.79686737060547,
|
| 333 |
"learning_rate": 8.622222222222223e-06,
|
| 334 |
+
"loss": 0.8895,
|
| 335 |
"step": 1125
|
| 336 |
},
|
| 337 |
{
|
| 338 |
+
"epoch": 0.23,
|
| 339 |
+
"grad_norm": 32.330204010009766,
|
| 340 |
"learning_rate": 8.566666666666667e-06,
|
| 341 |
+
"loss": 0.8984,
|
| 342 |
"step": 1150
|
| 343 |
},
|
| 344 |
{
|
| 345 |
+
"epoch": 0.235,
|
| 346 |
+
"grad_norm": 43.38914108276367,
|
| 347 |
"learning_rate": 8.511111111111113e-06,
|
| 348 |
+
"loss": 1.1215,
|
| 349 |
"step": 1175
|
| 350 |
},
|
| 351 |
{
|
| 352 |
+
"epoch": 0.24,
|
| 353 |
+
"grad_norm": 62.52465057373047,
|
| 354 |
"learning_rate": 8.455555555555555e-06,
|
| 355 |
+
"loss": 0.9912,
|
| 356 |
"step": 1200
|
| 357 |
},
|
| 358 |
{
|
| 359 |
+
"epoch": 0.245,
|
| 360 |
+
"grad_norm": 39.1291618347168,
|
| 361 |
"learning_rate": 8.400000000000001e-06,
|
| 362 |
+
"loss": 1.0422,
|
| 363 |
"step": 1225
|
| 364 |
},
|
| 365 |
{
|
| 366 |
+
"epoch": 0.25,
|
| 367 |
+
"grad_norm": 53.39470291137695,
|
| 368 |
"learning_rate": 8.344444444444445e-06,
|
| 369 |
+
"loss": 1.1502,
|
| 370 |
"step": 1250
|
| 371 |
},
|
| 372 |
{
|
| 373 |
+
"epoch": 0.255,
|
| 374 |
+
"grad_norm": 42.28481674194336,
|
| 375 |
"learning_rate": 8.288888888888889e-06,
|
| 376 |
+
"loss": 1.112,
|
| 377 |
"step": 1275
|
| 378 |
},
|
| 379 |
{
|
| 380 |
+
"epoch": 0.26,
|
| 381 |
+
"grad_norm": 37.374542236328125,
|
| 382 |
"learning_rate": 8.233333333333335e-06,
|
| 383 |
+
"loss": 1.0152,
|
| 384 |
"step": 1300
|
| 385 |
},
|
| 386 |
{
|
| 387 |
+
"epoch": 0.265,
|
| 388 |
+
"grad_norm": 52.94354248046875,
|
| 389 |
"learning_rate": 8.177777777777779e-06,
|
| 390 |
+
"loss": 0.9147,
|
| 391 |
"step": 1325
|
| 392 |
},
|
| 393 |
{
|
| 394 |
+
"epoch": 0.27,
|
| 395 |
+
"grad_norm": 46.23354721069336,
|
| 396 |
"learning_rate": 8.122222222222223e-06,
|
| 397 |
+
"loss": 0.8429,
|
| 398 |
"step": 1350
|
| 399 |
},
|
| 400 |
{
|
| 401 |
+
"epoch": 0.275,
|
| 402 |
+
"grad_norm": 40.089210510253906,
|
| 403 |
"learning_rate": 8.066666666666667e-06,
|
| 404 |
+
"loss": 0.8026,
|
| 405 |
"step": 1375
|
| 406 |
},
|
| 407 |
{
|
| 408 |
+
"epoch": 0.28,
|
| 409 |
+
"grad_norm": 42.95212173461914,
|
| 410 |
"learning_rate": 8.011111111111113e-06,
|
| 411 |
+
"loss": 0.7804,
|
| 412 |
"step": 1400
|
| 413 |
},
|
| 414 |
{
|
| 415 |
+
"epoch": 0.285,
|
| 416 |
+
"grad_norm": 56.216678619384766,
|
| 417 |
"learning_rate": 7.955555555555557e-06,
|
| 418 |
+
"loss": 0.8045,
|
| 419 |
"step": 1425
|
| 420 |
},
|
| 421 |
{
|
| 422 |
+
"epoch": 0.29,
|
| 423 |
+
"grad_norm": 45.01803207397461,
|
| 424 |
+
"learning_rate": 7.902222222222223e-06,
|
| 425 |
+
"loss": 0.9807,
|
| 426 |
"step": 1450
|
| 427 |
},
|
| 428 |
{
|
| 429 |
+
"epoch": 0.295,
|
| 430 |
+
"grad_norm": 63.81135940551758,
|
| 431 |
+
"learning_rate": 7.846666666666667e-06,
|
| 432 |
+
"loss": 0.9447,
|
| 433 |
"step": 1475
|
| 434 |
},
|
| 435 |
{
|
| 436 |
+
"epoch": 0.3,
|
| 437 |
+
"grad_norm": 38.958457946777344,
|
| 438 |
+
"learning_rate": 7.791111111111111e-06,
|
| 439 |
+
"loss": 1.0144,
|
| 440 |
"step": 1500
|
| 441 |
},
|
| 442 |
{
|
| 443 |
+
"epoch": 0.305,
|
| 444 |
+
"grad_norm": 48.87187957763672,
|
| 445 |
+
"learning_rate": 7.735555555555557e-06,
|
| 446 |
+
"loss": 1.0261,
|
| 447 |
"step": 1525
|
| 448 |
},
|
| 449 |
{
|
| 450 |
+
"epoch": 0.31,
|
| 451 |
+
"grad_norm": 34.80329513549805,
|
| 452 |
+
"learning_rate": 7.680000000000001e-06,
|
| 453 |
+
"loss": 0.8152,
|
| 454 |
"step": 1550
|
| 455 |
},
|
| 456 |
{
|
| 457 |
+
"epoch": 0.315,
|
| 458 |
+
"grad_norm": 45.12413024902344,
|
| 459 |
+
"learning_rate": 7.624444444444445e-06,
|
| 460 |
+
"loss": 0.8943,
|
| 461 |
"step": 1575
|
| 462 |
},
|
| 463 |
{
|
| 464 |
+
"epoch": 0.32,
|
| 465 |
+
"grad_norm": 38.103729248046875,
|
| 466 |
+
"learning_rate": 7.56888888888889e-06,
|
| 467 |
+
"loss": 1.085,
|
| 468 |
"step": 1600
|
| 469 |
},
|
| 470 |
{
|
| 471 |
+
"epoch": 0.325,
|
| 472 |
+
"grad_norm": 38.715492248535156,
|
| 473 |
+
"learning_rate": 7.513333333333334e-06,
|
| 474 |
+
"loss": 0.9853,
|
| 475 |
"step": 1625
|
| 476 |
},
|
| 477 |
{
|
| 478 |
+
"epoch": 0.33,
|
| 479 |
+
"grad_norm": 42.01364517211914,
|
| 480 |
+
"learning_rate": 7.457777777777778e-06,
|
| 481 |
+
"loss": 0.8214,
|
| 482 |
"step": 1650
|
| 483 |
},
|
| 484 |
{
|
| 485 |
+
"epoch": 0.335,
|
| 486 |
+
"grad_norm": 37.475799560546875,
|
| 487 |
+
"learning_rate": 7.402222222222223e-06,
|
| 488 |
+
"loss": 0.7706,
|
| 489 |
"step": 1675
|
| 490 |
},
|
| 491 |
{
|
| 492 |
+
"epoch": 0.34,
|
| 493 |
+
"grad_norm": 40.387210845947266,
|
| 494 |
+
"learning_rate": 7.346666666666668e-06,
|
| 495 |
+
"loss": 1.081,
|
| 496 |
"step": 1700
|
| 497 |
},
|
| 498 |
{
|
| 499 |
+
"epoch": 0.345,
|
| 500 |
+
"grad_norm": 21.215946197509766,
|
| 501 |
+
"learning_rate": 7.291111111111112e-06,
|
| 502 |
+
"loss": 0.8047,
|
| 503 |
"step": 1725
|
| 504 |
},
|
| 505 |
{
|
| 506 |
+
"epoch": 0.35,
|
| 507 |
+
"grad_norm": 21.454147338867188,
|
| 508 |
+
"learning_rate": 7.235555555555556e-06,
|
| 509 |
+
"loss": 0.836,
|
| 510 |
"step": 1750
|
| 511 |
},
|
| 512 |
{
|
| 513 |
+
"epoch": 0.355,
|
| 514 |
+
"grad_norm": 46.05310821533203,
|
| 515 |
+
"learning_rate": 7.180000000000001e-06,
|
| 516 |
+
"loss": 0.8883,
|
| 517 |
"step": 1775
|
| 518 |
},
|
| 519 |
{
|
| 520 |
+
"epoch": 0.36,
|
| 521 |
+
"grad_norm": 61.988643646240234,
|
| 522 |
+
"learning_rate": 7.124444444444445e-06,
|
| 523 |
+
"loss": 0.8254,
|
| 524 |
"step": 1800
|
| 525 |
},
|
| 526 |
{
|
| 527 |
+
"epoch": 1.0038,
|
| 528 |
+
"grad_norm": 34.00222396850586,
|
| 529 |
+
"learning_rate": 7.06888888888889e-06,
|
| 530 |
+
"loss": 0.5202,
|
| 531 |
"step": 1825
|
| 532 |
},
|
| 533 |
{
|
| 534 |
+
"epoch": 1.0088,
|
| 535 |
+
"grad_norm": 23.966150283813477,
|
| 536 |
+
"learning_rate": 7.0133333333333345e-06,
|
| 537 |
+
"loss": 0.362,
|
| 538 |
"step": 1850
|
| 539 |
},
|
| 540 |
{
|
| 541 |
+
"epoch": 1.0138,
|
| 542 |
+
"grad_norm": 19.050518035888672,
|
| 543 |
+
"learning_rate": 6.9577777777777785e-06,
|
| 544 |
+
"loss": 0.2891,
|
| 545 |
"step": 1875
|
| 546 |
},
|
| 547 |
{
|
| 548 |
+
"epoch": 1.0188,
|
| 549 |
+
"grad_norm": 34.59785079956055,
|
| 550 |
+
"learning_rate": 6.902222222222223e-06,
|
| 551 |
+
"loss": 0.3438,
|
| 552 |
"step": 1900
|
| 553 |
},
|
| 554 |
{
|
| 555 |
+
"epoch": 1.0238,
|
| 556 |
+
"grad_norm": 29.40850257873535,
|
| 557 |
+
"learning_rate": 6.846666666666667e-06,
|
| 558 |
+
"loss": 0.2822,
|
| 559 |
"step": 1925
|
| 560 |
},
|
| 561 |
{
|
| 562 |
+
"epoch": 1.0288,
|
| 563 |
+
"grad_norm": 15.476716041564941,
|
| 564 |
+
"learning_rate": 6.7911111111111115e-06,
|
| 565 |
+
"loss": 0.4843,
|
| 566 |
"step": 1950
|
| 567 |
},
|
| 568 |
{
|
| 569 |
+
"epoch": 1.0338,
|
| 570 |
+
"grad_norm": 33.912174224853516,
|
| 571 |
+
"learning_rate": 6.735555555555556e-06,
|
| 572 |
+
"loss": 0.3741,
|
| 573 |
"step": 1975
|
| 574 |
},
|
| 575 |
{
|
| 576 |
+
"epoch": 1.0388,
|
| 577 |
+
"grad_norm": 30.411020278930664,
|
| 578 |
+
"learning_rate": 6.680000000000001e-06,
|
| 579 |
+
"loss": 0.3044,
|
| 580 |
"step": 2000
|
| 581 |
},
|
| 582 |
{
|
| 583 |
+
"epoch": 1.0388,
|
| 584 |
+
"eval_cer": 23.904761904761905,
|
| 585 |
+
"eval_loss": 0.309874564409256,
|
| 586 |
+
"eval_model_preparation_time": 0.0121,
|
| 587 |
+
"eval_runtime": 119.6464,
|
| 588 |
+
"eval_samples_per_second": 2.507,
|
| 589 |
+
"eval_steps_per_second": 1.254,
|
| 590 |
+
"eval_wer": 23.46153846153846,
|
| 591 |
"step": 2000
|
| 592 |
},
|
| 593 |
{
|
| 594 |
+
"epoch": 1.0438,
|
| 595 |
+
"grad_norm": 19.959854125976562,
|
| 596 |
+
"learning_rate": 6.6244444444444445e-06,
|
| 597 |
+
"loss": 0.2255,
|
| 598 |
"step": 2025
|
| 599 |
},
|
| 600 |
{
|
| 601 |
+
"epoch": 1.0488,
|
| 602 |
+
"grad_norm": 20.751602172851562,
|
| 603 |
+
"learning_rate": 6.568888888888889e-06,
|
| 604 |
+
"loss": 0.3341,
|
| 605 |
"step": 2050
|
| 606 |
},
|
| 607 |
{
|
| 608 |
+
"epoch": 1.0538,
|
| 609 |
+
"grad_norm": 24.52460479736328,
|
| 610 |
+
"learning_rate": 6.513333333333333e-06,
|
| 611 |
+
"loss": 0.4861,
|
| 612 |
"step": 2075
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 1.0588,
|
| 616 |
+
"grad_norm": 5.740904331207275,
|
| 617 |
+
"learning_rate": 6.457777777777778e-06,
|
| 618 |
+
"loss": 0.4165,
|
| 619 |
"step": 2100
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"epoch": 1.0638,
|
| 623 |
+
"grad_norm": 24.452116012573242,
|
| 624 |
+
"learning_rate": 6.402222222222223e-06,
|
| 625 |
+
"loss": 0.4478,
|
| 626 |
"step": 2125
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 1.0688,
|
| 630 |
+
"grad_norm": 29.230716705322266,
|
| 631 |
+
"learning_rate": 6.346666666666668e-06,
|
| 632 |
+
"loss": 0.4387,
|
| 633 |
"step": 2150
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"epoch": 1.0738,
|
| 637 |
+
"grad_norm": 41.07571792602539,
|
| 638 |
+
"learning_rate": 6.291111111111111e-06,
|
| 639 |
+
"loss": 0.2466,
|
| 640 |
"step": 2175
|
| 641 |
},
|
| 642 |
{
|
| 643 |
+
"epoch": 1.0788,
|
| 644 |
+
"grad_norm": 19.89525032043457,
|
| 645 |
+
"learning_rate": 6.235555555555556e-06,
|
| 646 |
+
"loss": 0.3156,
|
| 647 |
"step": 2200
|
| 648 |
},
|
| 649 |
{
|
| 650 |
+
"epoch": 1.0838,
|
| 651 |
+
"grad_norm": 33.628971099853516,
|
| 652 |
+
"learning_rate": 6.18e-06,
|
| 653 |
+
"loss": 0.3624,
|
| 654 |
"step": 2225
|
| 655 |
},
|
| 656 |
{
|
| 657 |
+
"epoch": 1.0888,
|
| 658 |
+
"grad_norm": 23.34870147705078,
|
| 659 |
+
"learning_rate": 6.124444444444445e-06,
|
| 660 |
+
"loss": 0.4263,
|
| 661 |
"step": 2250
|
| 662 |
},
|
| 663 |
{
|
| 664 |
+
"epoch": 1.0937999999999999,
|
| 665 |
+
"grad_norm": 30.75408172607422,
|
| 666 |
+
"learning_rate": 6.06888888888889e-06,
|
| 667 |
+
"loss": 0.4153,
|
| 668 |
"step": 2275
|
| 669 |
},
|
| 670 |
{
|
| 671 |
+
"epoch": 1.0988,
|
| 672 |
+
"grad_norm": 7.229944705963135,
|
| 673 |
+
"learning_rate": 6.013333333333335e-06,
|
| 674 |
+
"loss": 0.3068,
|
| 675 |
"step": 2300
|
| 676 |
},
|
| 677 |
{
|
| 678 |
+
"epoch": 1.1038000000000001,
|
| 679 |
+
"grad_norm": 37.87436294555664,
|
| 680 |
+
"learning_rate": 5.957777777777778e-06,
|
| 681 |
+
"loss": 0.3049,
|
| 682 |
"step": 2325
|
| 683 |
},
|
| 684 |
{
|
| 685 |
+
"epoch": 1.1088,
|
| 686 |
+
"grad_norm": 33.34481430053711,
|
| 687 |
+
"learning_rate": 5.902222222222223e-06,
|
| 688 |
+
"loss": 0.2759,
|
| 689 |
"step": 2350
|
| 690 |
},
|
| 691 |
{
|
| 692 |
+
"epoch": 1.1138,
|
| 693 |
+
"grad_norm": 24.211904525756836,
|
| 694 |
+
"learning_rate": 5.846666666666667e-06,
|
| 695 |
+
"loss": 0.35,
|
| 696 |
"step": 2375
|
| 697 |
},
|
| 698 |
{
|
| 699 |
+
"epoch": 1.1188,
|
| 700 |
+
"grad_norm": 41.01383590698242,
|
| 701 |
+
"learning_rate": 5.791111111111112e-06,
|
| 702 |
+
"loss": 0.3992,
|
| 703 |
"step": 2400
|
| 704 |
},
|
| 705 |
{
|
| 706 |
+
"epoch": 1.1238,
|
| 707 |
+
"grad_norm": 6.543262004852295,
|
| 708 |
+
"learning_rate": 5.735555555555557e-06,
|
| 709 |
+
"loss": 0.3356,
|
| 710 |
"step": 2425
|
| 711 |
},
|
| 712 |
{
|
| 713 |
+
"epoch": 1.1288,
|
| 714 |
+
"grad_norm": 34.344913482666016,
|
| 715 |
+
"learning_rate": 5.68e-06,
|
| 716 |
+
"loss": 0.3364,
|
| 717 |
"step": 2450
|
| 718 |
},
|
| 719 |
{
|
| 720 |
+
"epoch": 1.1338,
|
| 721 |
+
"grad_norm": 9.35561752319336,
|
| 722 |
+
"learning_rate": 5.624444444444445e-06,
|
| 723 |
+
"loss": 0.3606,
|
| 724 |
"step": 2475
|
| 725 |
},
|
| 726 |
{
|
| 727 |
+
"epoch": 1.1388,
|
| 728 |
+
"grad_norm": 21.762096405029297,
|
| 729 |
+
"learning_rate": 5.56888888888889e-06,
|
| 730 |
+
"loss": 0.3437,
|
| 731 |
"step": 2500
|
| 732 |
},
|
| 733 |
{
|
| 734 |
+
"epoch": 1.1438,
|
| 735 |
+
"grad_norm": 29.119796752929688,
|
| 736 |
+
"learning_rate": 5.513333333333334e-06,
|
| 737 |
+
"loss": 0.428,
|
| 738 |
"step": 2525
|
| 739 |
},
|
| 740 |
{
|
| 741 |
+
"epoch": 1.1488,
|
| 742 |
+
"grad_norm": 46.66371536254883,
|
| 743 |
+
"learning_rate": 5.4577777777777785e-06,
|
| 744 |
+
"loss": 0.4566,
|
| 745 |
"step": 2550
|
| 746 |
},
|
| 747 |
{
|
| 748 |
+
"epoch": 1.1538,
|
| 749 |
+
"grad_norm": 15.0108642578125,
|
| 750 |
+
"learning_rate": 5.402222222222223e-06,
|
| 751 |
+
"loss": 0.3162,
|
| 752 |
"step": 2575
|
| 753 |
},
|
| 754 |
{
|
| 755 |
+
"epoch": 1.1588,
|
| 756 |
+
"grad_norm": 60.40862274169922,
|
| 757 |
+
"learning_rate": 5.346666666666667e-06,
|
| 758 |
+
"loss": 0.3696,
|
| 759 |
"step": 2600
|
| 760 |
},
|
| 761 |
{
|
| 762 |
+
"epoch": 1.1638,
|
| 763 |
+
"grad_norm": 26.4654598236084,
|
| 764 |
+
"learning_rate": 5.2911111111111115e-06,
|
| 765 |
+
"loss": 0.2831,
|
| 766 |
"step": 2625
|
| 767 |
},
|
| 768 |
{
|
| 769 |
+
"epoch": 1.1688,
|
| 770 |
+
"grad_norm": 23.651691436767578,
|
| 771 |
+
"learning_rate": 5.235555555555556e-06,
|
| 772 |
+
"loss": 0.4484,
|
| 773 |
"step": 2650
|
| 774 |
},
|
| 775 |
{
|
| 776 |
+
"epoch": 1.1738,
|
| 777 |
+
"grad_norm": 33.675167083740234,
|
| 778 |
+
"learning_rate": 5.18e-06,
|
| 779 |
+
"loss": 0.3062,
|
| 780 |
"step": 2675
|
| 781 |
},
|
| 782 |
{
|
| 783 |
+
"epoch": 1.1788,
|
| 784 |
+
"grad_norm": 27.336896896362305,
|
| 785 |
+
"learning_rate": 5.124444444444445e-06,
|
| 786 |
+
"loss": 0.3738,
|
| 787 |
"step": 2700
|
| 788 |
},
|
| 789 |
{
|
| 790 |
+
"epoch": 1.1838,
|
| 791 |
+
"grad_norm": 17.323768615722656,
|
| 792 |
+
"learning_rate": 5.06888888888889e-06,
|
| 793 |
+
"loss": 0.3182,
|
| 794 |
"step": 2725
|
| 795 |
},
|
| 796 |
{
|
| 797 |
+
"epoch": 1.1888,
|
| 798 |
+
"grad_norm": 18.657014846801758,
|
| 799 |
+
"learning_rate": 5.013333333333333e-06,
|
| 800 |
+
"loss": 0.3589,
|
| 801 |
"step": 2750
|
| 802 |
},
|
| 803 |
{
|
| 804 |
+
"epoch": 1.1938,
|
| 805 |
+
"grad_norm": 39.54949188232422,
|
| 806 |
+
"learning_rate": 4.957777777777778e-06,
|
| 807 |
+
"loss": 0.4187,
|
| 808 |
"step": 2775
|
| 809 |
},
|
| 810 |
{
|
| 811 |
+
"epoch": 1.1988,
|
| 812 |
+
"grad_norm": 57.646602630615234,
|
| 813 |
+
"learning_rate": 4.902222222222222e-06,
|
| 814 |
+
"loss": 0.3374,
|
| 815 |
"step": 2800
|
| 816 |
},
|
| 817 |
{
|
| 818 |
+
"epoch": 1.2038,
|
| 819 |
+
"grad_norm": 24.029294967651367,
|
| 820 |
+
"learning_rate": 4.846666666666667e-06,
|
| 821 |
+
"loss": 0.2373,
|
| 822 |
"step": 2825
|
| 823 |
},
|
| 824 |
{
|
| 825 |
+
"epoch": 1.2088,
|
| 826 |
+
"grad_norm": 34.673709869384766,
|
| 827 |
+
"learning_rate": 4.791111111111111e-06,
|
| 828 |
+
"loss": 0.3274,
|
| 829 |
"step": 2850
|
| 830 |
},
|
| 831 |
{
|
| 832 |
+
"epoch": 1.2138,
|
| 833 |
+
"grad_norm": 40.28670120239258,
|
| 834 |
+
"learning_rate": 4.735555555555556e-06,
|
| 835 |
+
"loss": 0.3105,
|
| 836 |
"step": 2875
|
| 837 |
},
|
| 838 |
{
|
| 839 |
+
"epoch": 1.2187999999999999,
|
| 840 |
+
"grad_norm": 22.120824813842773,
|
| 841 |
+
"learning_rate": 4.680000000000001e-06,
|
| 842 |
+
"loss": 0.2678,
|
| 843 |
"step": 2900
|
| 844 |
},
|
| 845 |
{
|
| 846 |
+
"epoch": 1.2238,
|
| 847 |
+
"grad_norm": 21.942718505859375,
|
| 848 |
+
"learning_rate": 4.624444444444445e-06,
|
| 849 |
+
"loss": 0.5025,
|
| 850 |
"step": 2925
|
| 851 |
},
|
| 852 |
{
|
| 853 |
+
"epoch": 1.2288000000000001,
|
| 854 |
+
"grad_norm": 31.645906448364258,
|
| 855 |
+
"learning_rate": 4.568888888888889e-06,
|
| 856 |
+
"loss": 0.3753,
|
| 857 |
"step": 2950
|
| 858 |
},
|
| 859 |
{
|
| 860 |
+
"epoch": 1.2338,
|
| 861 |
+
"grad_norm": 19.00498390197754,
|
| 862 |
+
"learning_rate": 4.513333333333333e-06,
|
| 863 |
+
"loss": 0.3421,
|
| 864 |
"step": 2975
|
| 865 |
},
|
| 866 |
{
|
| 867 |
+
"epoch": 1.2388,
|
| 868 |
+
"grad_norm": 42.73635482788086,
|
| 869 |
+
"learning_rate": 4.457777777777778e-06,
|
| 870 |
+
"loss": 0.3108,
|
| 871 |
"step": 3000
|
| 872 |
},
|
| 873 |
{
|
| 874 |
+
"epoch": 1.2388,
|
| 875 |
+
"eval_cer": 7.7142857142857135,
|
| 876 |
+
"eval_loss": 0.11533673852682114,
|
| 877 |
+
"eval_model_preparation_time": 0.0121,
|
| 878 |
+
"eval_runtime": 120.7728,
|
| 879 |
+
"eval_samples_per_second": 2.484,
|
| 880 |
+
"eval_steps_per_second": 1.242,
|
| 881 |
+
"eval_wer": 7.5,
|
| 882 |
"step": 3000
|
| 883 |
},
|
| 884 |
{
|
| 885 |
+
"epoch": 1.2438,
|
| 886 |
+
"grad_norm": 20.023395538330078,
|
| 887 |
+
"learning_rate": 4.402222222222223e-06,
|
| 888 |
+
"loss": 0.3144,
|
| 889 |
"step": 3025
|
| 890 |
},
|
| 891 |
{
|
| 892 |
+
"epoch": 1.2488,
|
| 893 |
+
"grad_norm": 32.98371505737305,
|
| 894 |
+
"learning_rate": 4.346666666666667e-06,
|
| 895 |
+
"loss": 0.2618,
|
| 896 |
"step": 3050
|
| 897 |
},
|
| 898 |
{
|
| 899 |
+
"epoch": 1.2538,
|
| 900 |
+
"grad_norm": 42.568119049072266,
|
| 901 |
+
"learning_rate": 4.291111111111112e-06,
|
| 902 |
+
"loss": 0.331,
|
| 903 |
"step": 3075
|
| 904 |
},
|
| 905 |
{
|
| 906 |
+
"epoch": 1.2588,
|
| 907 |
+
"grad_norm": 0.5769469738006592,
|
| 908 |
+
"learning_rate": 4.235555555555556e-06,
|
| 909 |
+
"loss": 0.3391,
|
| 910 |
"step": 3100
|
| 911 |
},
|
| 912 |
{
|
| 913 |
+
"epoch": 1.2638,
|
| 914 |
+
"grad_norm": 20.518579483032227,
|
| 915 |
+
"learning_rate": 4.18e-06,
|
| 916 |
+
"loss": 0.2234,
|
| 917 |
"step": 3125
|
| 918 |
},
|
| 919 |
{
|
| 920 |
+
"epoch": 1.2688,
|
| 921 |
+
"grad_norm": 27.69402313232422,
|
| 922 |
+
"learning_rate": 4.124444444444445e-06,
|
| 923 |
+
"loss": 0.295,
|
| 924 |
"step": 3150
|
| 925 |
},
|
| 926 |
{
|
| 927 |
+
"epoch": 1.2738,
|
| 928 |
+
"grad_norm": 31.121999740600586,
|
| 929 |
+
"learning_rate": 4.0688888888888896e-06,
|
| 930 |
+
"loss": 0.3753,
|
| 931 |
"step": 3175
|
| 932 |
},
|
| 933 |
{
|
| 934 |
+
"epoch": 1.2788,
|
| 935 |
+
"grad_norm": 14.77844524383545,
|
| 936 |
+
"learning_rate": 4.013333333333334e-06,
|
| 937 |
+
"loss": 0.3706,
|
| 938 |
"step": 3200
|
| 939 |
},
|
| 940 |
{
|
| 941 |
+
"epoch": 1.2838,
|
| 942 |
+
"grad_norm": 27.002138137817383,
|
| 943 |
+
"learning_rate": 3.9577777777777785e-06,
|
| 944 |
+
"loss": 0.2957,
|
| 945 |
"step": 3225
|
| 946 |
},
|
| 947 |
{
|
| 948 |
+
"epoch": 1.2888,
|
| 949 |
+
"grad_norm": 13.426039695739746,
|
| 950 |
+
"learning_rate": 3.9022222222222225e-06,
|
| 951 |
+
"loss": 0.2902,
|
| 952 |
"step": 3250
|
| 953 |
},
|
| 954 |
{
|
| 955 |
+
"epoch": 1.2938,
|
| 956 |
+
"grad_norm": 41.75296401977539,
|
| 957 |
+
"learning_rate": 3.8466666666666665e-06,
|
| 958 |
+
"loss": 0.2924,
|
| 959 |
"step": 3275
|
| 960 |
},
|
| 961 |
{
|
| 962 |
+
"epoch": 1.2988,
|
| 963 |
+
"grad_norm": 52.886409759521484,
|
| 964 |
+
"learning_rate": 3.7911111111111114e-06,
|
| 965 |
+
"loss": 0.2519,
|
| 966 |
"step": 3300
|
| 967 |
},
|
| 968 |
{
|
| 969 |
+
"epoch": 1.3038,
|
| 970 |
+
"grad_norm": 33.73487854003906,
|
| 971 |
+
"learning_rate": 3.7355555555555555e-06,
|
| 972 |
+
"loss": 0.2966,
|
| 973 |
"step": 3325
|
| 974 |
},
|
| 975 |
{
|
| 976 |
+
"epoch": 1.3088,
|
| 977 |
+
"grad_norm": 27.994157791137695,
|
| 978 |
+
"learning_rate": 3.6800000000000003e-06,
|
| 979 |
+
"loss": 0.2887,
|
| 980 |
"step": 3350
|
| 981 |
},
|
| 982 |
{
|
| 983 |
+
"epoch": 1.3138,
|
| 984 |
+
"grad_norm": 32.397579193115234,
|
| 985 |
+
"learning_rate": 3.624444444444445e-06,
|
| 986 |
+
"loss": 0.2368,
|
| 987 |
"step": 3375
|
| 988 |
},
|
| 989 |
{
|
| 990 |
+
"epoch": 1.3188,
|
| 991 |
+
"grad_norm": 25.96181869506836,
|
| 992 |
+
"learning_rate": 3.568888888888889e-06,
|
| 993 |
+
"loss": 0.2911,
|
| 994 |
"step": 3400
|
| 995 |
},
|
| 996 |
{
|
| 997 |
+
"epoch": 1.3237999999999999,
|
| 998 |
+
"grad_norm": 7.705018520355225,
|
| 999 |
+
"learning_rate": 3.5133333333333337e-06,
|
| 1000 |
+
"loss": 0.2647,
|
| 1001 |
"step": 3425
|
| 1002 |
},
|
| 1003 |
{
|
| 1004 |
+
"epoch": 1.3288,
|
| 1005 |
+
"grad_norm": 28.499221801757812,
|
| 1006 |
+
"learning_rate": 3.457777777777778e-06,
|
| 1007 |
+
"loss": 0.4149,
|
| 1008 |
"step": 3450
|
| 1009 |
},
|
| 1010 |
{
|
| 1011 |
+
"epoch": 1.3338,
|
| 1012 |
+
"grad_norm": 40.06334686279297,
|
| 1013 |
+
"learning_rate": 3.4022222222222222e-06,
|
| 1014 |
+
"loss": 0.2952,
|
| 1015 |
"step": 3475
|
| 1016 |
},
|
| 1017 |
{
|
| 1018 |
+
"epoch": 1.3388,
|
| 1019 |
+
"grad_norm": 1.474075198173523,
|
| 1020 |
+
"learning_rate": 3.346666666666667e-06,
|
| 1021 |
+
"loss": 0.207,
|
| 1022 |
"step": 3500
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
+
"epoch": 1.3437999999999999,
|
| 1026 |
+
"grad_norm": 27.121145248413086,
|
| 1027 |
+
"learning_rate": 3.2911111111111116e-06,
|
| 1028 |
+
"loss": 0.3432,
|
| 1029 |
"step": 3525
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
+
"epoch": 1.3488,
|
| 1033 |
+
"grad_norm": 22.38213539123535,
|
| 1034 |
+
"learning_rate": 3.2355555555555556e-06,
|
| 1035 |
+
"loss": 0.3522,
|
| 1036 |
"step": 3550
|
| 1037 |
},
|
| 1038 |
{
|
| 1039 |
+
"epoch": 1.3538000000000001,
|
| 1040 |
+
"grad_norm": 42.13764190673828,
|
| 1041 |
+
"learning_rate": 3.1800000000000005e-06,
|
| 1042 |
+
"loss": 0.3729,
|
| 1043 |
"step": 3575
|
| 1044 |
},
|
| 1045 |
{
|
| 1046 |
+
"epoch": 1.3588,
|
| 1047 |
+
"grad_norm": 27.436649322509766,
|
| 1048 |
+
"learning_rate": 3.124444444444445e-06,
|
| 1049 |
+
"loss": 0.1482,
|
| 1050 |
"step": 3600
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
+
"epoch": 2.0026,
|
| 1054 |
+
"grad_norm": 16.43301010131836,
|
| 1055 |
+
"learning_rate": 3.068888888888889e-06,
|
| 1056 |
+
"loss": 0.1802,
|
| 1057 |
"step": 3625
|
| 1058 |
},
|
| 1059 |
{
|
| 1060 |
+
"epoch": 2.0076,
|
| 1061 |
+
"grad_norm": 11.994711875915527,
|
| 1062 |
+
"learning_rate": 3.013333333333334e-06,
|
| 1063 |
+
"loss": 0.0922,
|
| 1064 |
"step": 3650
|
| 1065 |
},
|
| 1066 |
{
|
| 1067 |
+
"epoch": 2.0126,
|
| 1068 |
+
"grad_norm": 7.826560020446777,
|
| 1069 |
+
"learning_rate": 2.957777777777778e-06,
|
| 1070 |
+
"loss": 0.0312,
|
| 1071 |
"step": 3675
|
| 1072 |
},
|
| 1073 |
{
|
| 1074 |
+
"epoch": 2.0176,
|
| 1075 |
+
"grad_norm": 6.457350730895996,
|
| 1076 |
+
"learning_rate": 2.9022222222222223e-06,
|
| 1077 |
+
"loss": 0.0343,
|
| 1078 |
"step": 3700
|
| 1079 |
},
|
| 1080 |
{
|
| 1081 |
+
"epoch": 2.0226,
|
| 1082 |
+
"grad_norm": 1.624599575996399,
|
| 1083 |
+
"learning_rate": 2.8466666666666672e-06,
|
| 1084 |
+
"loss": 0.0751,
|
| 1085 |
"step": 3725
|
| 1086 |
},
|
| 1087 |
{
|
| 1088 |
+
"epoch": 2.0276,
|
| 1089 |
+
"grad_norm": 4.553808212280273,
|
| 1090 |
+
"learning_rate": 2.7911111111111113e-06,
|
| 1091 |
+
"loss": 0.0905,
|
| 1092 |
"step": 3750
|
| 1093 |
},
|
| 1094 |
{
|
| 1095 |
+
"epoch": 2.0326,
|
| 1096 |
+
"grad_norm": 4.33929967880249,
|
| 1097 |
+
"learning_rate": 2.7355555555555557e-06,
|
| 1098 |
+
"loss": 0.079,
|
| 1099 |
"step": 3775
|
| 1100 |
},
|
| 1101 |
{
|
| 1102 |
+
"epoch": 2.0376,
|
| 1103 |
+
"grad_norm": 11.390565872192383,
|
| 1104 |
+
"learning_rate": 2.68e-06,
|
| 1105 |
+
"loss": 0.064,
|
| 1106 |
"step": 3800
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
+
"epoch": 2.0426,
|
| 1110 |
+
"grad_norm": 0.5454270243644714,
|
| 1111 |
+
"learning_rate": 2.6244444444444446e-06,
|
| 1112 |
+
"loss": 0.0351,
|
| 1113 |
"step": 3825
|
| 1114 |
},
|
| 1115 |
{
|
| 1116 |
+
"epoch": 2.0476,
|
| 1117 |
+
"grad_norm": 25.50840950012207,
|
| 1118 |
+
"learning_rate": 2.568888888888889e-06,
|
| 1119 |
+
"loss": 0.0726,
|
| 1120 |
"step": 3850
|
| 1121 |
},
|
| 1122 |
{
|
| 1123 |
+
"epoch": 2.0526,
|
| 1124 |
+
"grad_norm": 5.380075931549072,
|
| 1125 |
+
"learning_rate": 2.5133333333333336e-06,
|
| 1126 |
+
"loss": 0.0405,
|
| 1127 |
"step": 3875
|
| 1128 |
},
|
| 1129 |
{
|
| 1130 |
+
"epoch": 2.0576,
|
| 1131 |
+
"grad_norm": 1.0126858949661255,
|
| 1132 |
+
"learning_rate": 2.457777777777778e-06,
|
| 1133 |
+
"loss": 0.0858,
|
| 1134 |
"step": 3900
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
+
"epoch": 2.0626,
|
| 1138 |
+
"grad_norm": 1.7211323976516724,
|
| 1139 |
+
"learning_rate": 2.4022222222222225e-06,
|
| 1140 |
+
"loss": 0.0431,
|
| 1141 |
"step": 3925
|
| 1142 |
},
|
| 1143 |
{
|
| 1144 |
+
"epoch": 2.0676,
|
| 1145 |
+
"grad_norm": 2.049405574798584,
|
| 1146 |
+
"learning_rate": 2.346666666666667e-06,
|
| 1147 |
+
"loss": 0.0127,
|
| 1148 |
"step": 3950
|
| 1149 |
},
|
| 1150 |
{
|
| 1151 |
+
"epoch": 2.0726,
|
| 1152 |
+
"grad_norm": 1.535748839378357,
|
| 1153 |
+
"learning_rate": 2.2911111111111114e-06,
|
| 1154 |
+
"loss": 0.0559,
|
| 1155 |
"step": 3975
|
| 1156 |
},
|
| 1157 |
{
|
| 1158 |
+
"epoch": 2.0776,
|
| 1159 |
+
"grad_norm": 19.566328048706055,
|
| 1160 |
+
"learning_rate": 2.235555555555556e-06,
|
| 1161 |
+
"loss": 0.0544,
|
| 1162 |
"step": 4000
|
| 1163 |
},
|
| 1164 |
{
|
| 1165 |
+
"epoch": 2.0776,
|
| 1166 |
+
"eval_cer": 2.2857142857142856,
|
| 1167 |
+
"eval_loss": 0.029534637928009033,
|
| 1168 |
+
"eval_model_preparation_time": 0.0121,
|
| 1169 |
+
"eval_runtime": 119.7419,
|
| 1170 |
+
"eval_samples_per_second": 2.505,
|
| 1171 |
+
"eval_steps_per_second": 1.253,
|
| 1172 |
+
"eval_wer": 2.307692307692308,
|
| 1173 |
"step": 4000
|
| 1174 |
},
|
| 1175 |
{
|
| 1176 |
+
"epoch": 2.0826000000000002,
|
| 1177 |
+
"grad_norm": 0.4915294945240021,
|
| 1178 |
+
"learning_rate": 2.1800000000000003e-06,
|
| 1179 |
+
"loss": 0.0556,
|
| 1180 |
"step": 4025
|
| 1181 |
},
|
| 1182 |
{
|
| 1183 |
+
"epoch": 2.0876,
|
| 1184 |
+
"grad_norm": 13.67375659942627,
|
| 1185 |
+
"learning_rate": 2.1244444444444443e-06,
|
| 1186 |
+
"loss": 0.0377,
|
| 1187 |
"step": 4050
|
| 1188 |
},
|
| 1189 |
{
|
| 1190 |
+
"epoch": 2.0926,
|
| 1191 |
+
"grad_norm": 0.06471225619316101,
|
| 1192 |
+
"learning_rate": 2.0688888888888892e-06,
|
| 1193 |
+
"loss": 0.0535,
|
| 1194 |
"step": 4075
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
+
"epoch": 2.0976,
|
| 1198 |
+
"grad_norm": 0.1591552495956421,
|
| 1199 |
+
"learning_rate": 2.0133333333333337e-06,
|
| 1200 |
+
"loss": 0.037,
|
| 1201 |
"step": 4100
|
| 1202 |
},
|
| 1203 |
{
|
| 1204 |
+
"epoch": 2.1026,
|
| 1205 |
+
"grad_norm": 1.6957018375396729,
|
| 1206 |
+
"learning_rate": 1.9577777777777777e-06,
|
| 1207 |
+
"loss": 0.0548,
|
| 1208 |
"step": 4125
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
+
"epoch": 2.1076,
|
| 1212 |
+
"grad_norm": 7.906589508056641,
|
| 1213 |
+
"learning_rate": 1.9022222222222222e-06,
|
| 1214 |
+
"loss": 0.066,
|
| 1215 |
"step": 4150
|
| 1216 |
},
|
| 1217 |
{
|
| 1218 |
+
"epoch": 2.1126,
|
| 1219 |
+
"grad_norm": 0.4227987825870514,
|
| 1220 |
+
"learning_rate": 1.8466666666666668e-06,
|
| 1221 |
+
"loss": 0.0687,
|
| 1222 |
"step": 4175
|
| 1223 |
},
|
| 1224 |
{
|
| 1225 |
+
"epoch": 2.1176,
|
| 1226 |
+
"grad_norm": 0.3206275999546051,
|
| 1227 |
+
"learning_rate": 1.7911111111111113e-06,
|
| 1228 |
+
"loss": 0.0263,
|
| 1229 |
"step": 4200
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
+
"epoch": 2.1226,
|
| 1233 |
+
"grad_norm": 0.16232123970985413,
|
| 1234 |
+
"learning_rate": 1.7355555555555555e-06,
|
| 1235 |
+
"loss": 0.0257,
|
| 1236 |
"step": 4225
|
| 1237 |
},
|
| 1238 |
{
|
| 1239 |
+
"epoch": 2.1276,
|
| 1240 |
+
"grad_norm": 21.80668830871582,
|
| 1241 |
+
"learning_rate": 1.6800000000000002e-06,
|
| 1242 |
+
"loss": 0.0346,
|
| 1243 |
"step": 4250
|
| 1244 |
},
|
| 1245 |
{
|
| 1246 |
+
"epoch": 2.1326,
|
| 1247 |
+
"grad_norm": 3.021885871887207,
|
| 1248 |
+
"learning_rate": 1.6244444444444447e-06,
|
| 1249 |
+
"loss": 0.0715,
|
| 1250 |
"step": 4275
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
+
"epoch": 2.1376,
|
| 1254 |
+
"grad_norm": 9.51052188873291,
|
| 1255 |
+
"learning_rate": 1.568888888888889e-06,
|
| 1256 |
+
"loss": 0.0686,
|
| 1257 |
"step": 4300
|
| 1258 |
},
|
| 1259 |
{
|
| 1260 |
+
"epoch": 2.1426,
|
| 1261 |
+
"grad_norm": 7.770538806915283,
|
| 1262 |
+
"learning_rate": 1.5133333333333334e-06,
|
| 1263 |
+
"loss": 0.0681,
|
| 1264 |
"step": 4325
|
| 1265 |
},
|
| 1266 |
{
|
| 1267 |
+
"epoch": 2.1476,
|
| 1268 |
+
"grad_norm": 24.541362762451172,
|
| 1269 |
+
"learning_rate": 1.457777777777778e-06,
|
| 1270 |
+
"loss": 0.0589,
|
| 1271 |
"step": 4350
|
| 1272 |
},
|
| 1273 |
{
|
| 1274 |
+
"epoch": 2.1526,
|
| 1275 |
+
"grad_norm": 0.2108163684606552,
|
| 1276 |
+
"learning_rate": 1.4022222222222223e-06,
|
| 1277 |
+
"loss": 0.0407,
|
| 1278 |
"step": 4375
|
| 1279 |
},
|
| 1280 |
{
|
| 1281 |
+
"epoch": 2.1576,
|
| 1282 |
+
"grad_norm": 39.73221206665039,
|
| 1283 |
+
"learning_rate": 1.3466666666666668e-06,
|
| 1284 |
+
"loss": 0.0895,
|
| 1285 |
"step": 4400
|
| 1286 |
},
|
| 1287 |
{
|
| 1288 |
+
"epoch": 2.1626,
|
| 1289 |
+
"grad_norm": 4.3332438468933105,
|
| 1290 |
+
"learning_rate": 1.2911111111111112e-06,
|
| 1291 |
+
"loss": 0.0645,
|
| 1292 |
"step": 4425
|
| 1293 |
},
|
| 1294 |
{
|
| 1295 |
+
"epoch": 2.1676,
|
| 1296 |
+
"grad_norm": 1.142255425453186,
|
| 1297 |
+
"learning_rate": 1.2355555555555557e-06,
|
| 1298 |
+
"loss": 0.0592,
|
| 1299 |
"step": 4450
|
| 1300 |
},
|
| 1301 |
{
|
| 1302 |
+
"epoch": 2.1726,
|
| 1303 |
+
"grad_norm": 0.046843066811561584,
|
| 1304 |
+
"learning_rate": 1.1800000000000001e-06,
|
| 1305 |
+
"loss": 0.034,
|
| 1306 |
"step": 4475
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
+
"epoch": 2.1776,
|
| 1310 |
+
"grad_norm": 14.69985580444336,
|
| 1311 |
+
"learning_rate": 1.1244444444444446e-06,
|
| 1312 |
+
"loss": 0.0704,
|
| 1313 |
"step": 4500
|
| 1314 |
},
|
| 1315 |
{
|
| 1316 |
+
"epoch": 2.1826,
|
| 1317 |
+
"grad_norm": 11.02341079711914,
|
| 1318 |
+
"learning_rate": 1.068888888888889e-06,
|
| 1319 |
+
"loss": 0.0621,
|
| 1320 |
"step": 4525
|
| 1321 |
},
|
| 1322 |
{
|
| 1323 |
+
"epoch": 2.1875999999999998,
|
| 1324 |
+
"grad_norm": 0.05967080965638161,
|
| 1325 |
+
"learning_rate": 1.0133333333333333e-06,
|
| 1326 |
+
"loss": 0.0331,
|
| 1327 |
"step": 4550
|
| 1328 |
},
|
| 1329 |
{
|
| 1330 |
+
"epoch": 2.1926,
|
| 1331 |
+
"grad_norm": 9.536999702453613,
|
| 1332 |
+
"learning_rate": 9.57777777777778e-07,
|
| 1333 |
+
"loss": 0.0762,
|
| 1334 |
"step": 4575
|
| 1335 |
},
|
| 1336 |
{
|
| 1337 |
+
"epoch": 2.1976,
|
| 1338 |
+
"grad_norm": 3.740366220474243,
|
| 1339 |
+
"learning_rate": 9.022222222222222e-07,
|
| 1340 |
+
"loss": 0.0346,
|
| 1341 |
"step": 4600
|
| 1342 |
},
|
| 1343 |
{
|
| 1344 |
+
"epoch": 2.2026,
|
| 1345 |
+
"grad_norm": 1.1851881742477417,
|
| 1346 |
+
"learning_rate": 8.466666666666668e-07,
|
| 1347 |
+
"loss": 0.0502,
|
| 1348 |
"step": 4625
|
| 1349 |
},
|
| 1350 |
{
|
| 1351 |
+
"epoch": 2.2076000000000002,
|
| 1352 |
+
"grad_norm": 23.711048126220703,
|
| 1353 |
+
"learning_rate": 7.911111111111111e-07,
|
| 1354 |
+
"loss": 0.0661,
|
| 1355 |
"step": 4650
|
| 1356 |
},
|
| 1357 |
{
|
| 1358 |
+
"epoch": 2.2126,
|
| 1359 |
+
"grad_norm": 1.0020112991333008,
|
| 1360 |
+
"learning_rate": 7.355555555555556e-07,
|
| 1361 |
+
"loss": 0.0252,
|
| 1362 |
"step": 4675
|
| 1363 |
},
|
| 1364 |
{
|
| 1365 |
+
"epoch": 2.2176,
|
| 1366 |
+
"grad_norm": 21.021774291992188,
|
| 1367 |
+
"learning_rate": 6.800000000000001e-07,
|
| 1368 |
+
"loss": 0.0415,
|
| 1369 |
"step": 4700
|
| 1370 |
},
|
| 1371 |
{
|
| 1372 |
+
"epoch": 2.2226,
|
| 1373 |
+
"grad_norm": 0.8725367784500122,
|
| 1374 |
+
"learning_rate": 6.244444444444445e-07,
|
| 1375 |
+
"loss": 0.0293,
|
| 1376 |
"step": 4725
|
| 1377 |
},
|
| 1378 |
{
|
| 1379 |
+
"epoch": 2.2276,
|
| 1380 |
+
"grad_norm": 0.3742901086807251,
|
| 1381 |
+
"learning_rate": 5.68888888888889e-07,
|
| 1382 |
+
"loss": 0.0146,
|
| 1383 |
"step": 4750
|
| 1384 |
},
|
| 1385 |
{
|
| 1386 |
+
"epoch": 2.2326,
|
| 1387 |
+
"grad_norm": 12.225566864013672,
|
| 1388 |
+
"learning_rate": 5.133333333333334e-07,
|
| 1389 |
+
"loss": 0.0775,
|
| 1390 |
"step": 4775
|
| 1391 |
},
|
| 1392 |
{
|
| 1393 |
+
"epoch": 2.2376,
|
| 1394 |
+
"grad_norm": 0.5857837200164795,
|
| 1395 |
+
"learning_rate": 4.5777777777777784e-07,
|
| 1396 |
+
"loss": 0.0335,
|
| 1397 |
"step": 4800
|
| 1398 |
},
|
| 1399 |
{
|
| 1400 |
+
"epoch": 2.2426,
|
| 1401 |
+
"grad_norm": 2.0213797092437744,
|
| 1402 |
+
"learning_rate": 4.0222222222222224e-07,
|
| 1403 |
+
"loss": 0.06,
|
| 1404 |
"step": 4825
|
| 1405 |
},
|
| 1406 |
{
|
| 1407 |
+
"epoch": 2.2476,
|
| 1408 |
+
"grad_norm": 0.22115977108478546,
|
| 1409 |
+
"learning_rate": 3.466666666666667e-07,
|
| 1410 |
+
"loss": 0.0437,
|
| 1411 |
"step": 4850
|
| 1412 |
},
|
| 1413 |
{
|
| 1414 |
+
"epoch": 2.2526,
|
| 1415 |
+
"grad_norm": 11.783333778381348,
|
| 1416 |
+
"learning_rate": 2.9111111111111116e-07,
|
| 1417 |
+
"loss": 0.0454,
|
| 1418 |
"step": 4875
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
+
"epoch": 2.2576,
|
| 1422 |
+
"grad_norm": 2.67048978805542,
|
| 1423 |
+
"learning_rate": 2.3555555555555556e-07,
|
| 1424 |
+
"loss": 0.093,
|
| 1425 |
"step": 4900
|
| 1426 |
},
|
| 1427 |
{
|
| 1428 |
+
"epoch": 2.2626,
|
| 1429 |
+
"grad_norm": 0.16820040345191956,
|
| 1430 |
+
"learning_rate": 1.8e-07,
|
| 1431 |
+
"loss": 0.1193,
|
| 1432 |
"step": 4925
|
| 1433 |
},
|
| 1434 |
{
|
| 1435 |
+
"epoch": 2.2676,
|
| 1436 |
+
"grad_norm": 1.518930435180664,
|
| 1437 |
+
"learning_rate": 1.2444444444444446e-07,
|
| 1438 |
+
"loss": 0.0493,
|
| 1439 |
"step": 4950
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
+
"epoch": 2.2726,
|
| 1443 |
+
"grad_norm": 0.10118613392114639,
|
| 1444 |
+
"learning_rate": 6.888888888888889e-08,
|
| 1445 |
+
"loss": 0.0538,
|
| 1446 |
"step": 4975
|
| 1447 |
},
|
| 1448 |
{
|
| 1449 |
+
"epoch": 2.2776,
|
| 1450 |
+
"grad_norm": 0.5119066834449768,
|
| 1451 |
+
"learning_rate": 1.3333333333333334e-08,
|
| 1452 |
+
"loss": 0.0678,
|
| 1453 |
"step": 5000
|
| 1454 |
},
|
| 1455 |
{
|
| 1456 |
+
"epoch": 2.2776,
|
| 1457 |
+
"eval_cer": 0.9523809523809524,
|
| 1458 |
+
"eval_loss": 0.014132725074887276,
|
| 1459 |
+
"eval_model_preparation_time": 0.0121,
|
| 1460 |
+
"eval_runtime": 119.856,
|
| 1461 |
+
"eval_samples_per_second": 2.503,
|
| 1462 |
+
"eval_steps_per_second": 1.252,
|
| 1463 |
+
"eval_wer": 0.9615384615384616,
|
| 1464 |
"step": 5000
|
| 1465 |
},
|
| 1466 |
{
|
| 1467 |
+
"epoch": 2.2776,
|
| 1468 |
"step": 5000,
|
| 1469 |
+
"total_flos": 1.02060490752e+19,
|
| 1470 |
+
"train_loss": 0.5477467903137208,
|
| 1471 |
+
"train_runtime": 5409.143,
|
| 1472 |
+
"train_samples_per_second": 1.849,
|
| 1473 |
+
"train_steps_per_second": 0.924
|
| 1474 |
}
|
| 1475 |
],
|
| 1476 |
+
"logging_steps": 25,
|
| 1477 |
"max_steps": 5000,
|
| 1478 |
+
"num_input_tokens_seen": 0,
|
| 1479 |
"num_train_epochs": 9223372036854775807,
|
| 1480 |
+
"save_steps": 1000,
|
| 1481 |
+
"stateful_callbacks": {
|
| 1482 |
+
"TrainerControl": {
|
| 1483 |
+
"args": {
|
| 1484 |
+
"should_epoch_stop": false,
|
| 1485 |
+
"should_evaluate": false,
|
| 1486 |
+
"should_log": false,
|
| 1487 |
+
"should_save": true,
|
| 1488 |
+
"should_training_stop": true
|
| 1489 |
+
},
|
| 1490 |
+
"attributes": {}
|
| 1491 |
+
}
|
| 1492 |
+
},
|
| 1493 |
+
"total_flos": 1.02060490752e+19,
|
| 1494 |
+
"train_batch_size": 2,
|
| 1495 |
"trial_name": null,
|
| 1496 |
"trial_params": null
|
| 1497 |
}
|