End of training
Browse files- README.md +17 -4
- all_results.json +9 -9
- eval_results.json +5 -5
- train_results.json +4 -4
- trainer_state.json +824 -824
README.md
CHANGED
|
@@ -3,20 +3,33 @@ library_name: transformers
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: openai/whisper-tiny
|
| 5 |
tags:
|
|
|
|
| 6 |
- generated_from_trainer
|
|
|
|
|
|
|
| 7 |
metrics:
|
| 8 |
- wer
|
| 9 |
model-index:
|
| 10 |
-
- name:
|
| 11 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
|
| 14 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
|
| 17 |
-
#
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.3719
|
| 22 |
- Wer: 14.8495
|
|
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: openai/whisper-tiny
|
| 5 |
tags:
|
| 6 |
+
- whisper-event
|
| 7 |
- generated_from_trainer
|
| 8 |
+
datasets:
|
| 9 |
+
- asierhv/composite_corpus_eu_v2.1
|
| 10 |
metrics:
|
| 11 |
- wer
|
| 12 |
model-index:
|
| 13 |
+
- name: Whisper Tiny Basque
|
| 14 |
+
results:
|
| 15 |
+
- task:
|
| 16 |
+
name: Automatic Speech Recognition
|
| 17 |
+
type: automatic-speech-recognition
|
| 18 |
+
dataset:
|
| 19 |
+
name: asierhv/composite_corpus_eu_v2.1
|
| 20 |
+
type: asierhv/composite_corpus_eu_v2.1
|
| 21 |
+
metrics:
|
| 22 |
+
- name: Wer
|
| 23 |
+
type: wer
|
| 24 |
+
value: 14.849506681653555
|
| 25 |
---
|
| 26 |
|
| 27 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 28 |
should probably proofread and complete it, then remove this comment. -->
|
| 29 |
|
| 30 |
+
# Whisper Tiny Basque
|
| 31 |
|
| 32 |
+
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the asierhv/composite_corpus_eu_v2.1 dataset.
|
| 33 |
It achieves the following results on the evaluation set:
|
| 34 |
- Loss: 0.3719
|
| 35 |
- Wer: 14.8495
|
all_results.json
CHANGED
|
@@ -1,13 +1,13 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second": 2.
|
| 7 |
-
"eval_wer": 14.
|
| 8 |
"total_flos": 7.8780432384e+18,
|
| 9 |
-
"train_loss": 0.
|
| 10 |
-
"train_runtime":
|
| 11 |
-
"train_samples_per_second":
|
| 12 |
-
"train_steps_per_second": 2.
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"eval_loss": 0.37193799018859863,
|
| 4 |
+
"eval_runtime": 37.8012,
|
| 5 |
+
"eval_samples_per_second": 44.734,
|
| 6 |
+
"eval_steps_per_second": 2.804,
|
| 7 |
+
"eval_wer": 14.849506681653555,
|
| 8 |
"total_flos": 7.8780432384e+18,
|
| 9 |
+
"train_loss": 0.2454334835767746,
|
| 10 |
+
"train_runtime": 3974.1996,
|
| 11 |
+
"train_samples_per_second": 80.519,
|
| 12 |
+
"train_steps_per_second": 2.516
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second": 2.
|
| 7 |
-
"eval_wer": 14.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"eval_loss": 0.37193799018859863,
|
| 4 |
+
"eval_runtime": 37.8012,
|
| 5 |
+
"eval_samples_per_second": 44.734,
|
| 6 |
+
"eval_steps_per_second": 2.804,
|
| 7 |
+
"eval_wer": 14.849506681653555
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 7.8780432384e+18,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 2.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 7.8780432384e+18,
|
| 4 |
+
"train_loss": 0.2454334835767746,
|
| 5 |
+
"train_runtime": 3974.1996,
|
| 6 |
+
"train_samples_per_second": 80.519,
|
| 7 |
+
"train_steps_per_second": 2.516
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 14.
|
| 3 |
-
"best_model_checkpoint": "./checkpoint-
|
| 4 |
"epoch": 1.0,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
"global_step": 10000,
|
|
@@ -10,2902 +10,2902 @@
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0025,
|
| 13 |
-
"grad_norm": 32.
|
| 14 |
"learning_rate": 7.875e-07,
|
| 15 |
-
"loss": 3.
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.005,
|
| 20 |
-
"grad_norm": 17.
|
| 21 |
"learning_rate": 1.7249999999999998e-06,
|
| 22 |
-
"loss": 2.
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.0075,
|
| 27 |
-
"grad_norm": 13.
|
| 28 |
"learning_rate": 2.6624999999999995e-06,
|
| 29 |
"loss": 1.9843,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.01,
|
| 34 |
-
"grad_norm": 10.
|
| 35 |
"learning_rate": 3.6e-06,
|
| 36 |
-
"loss": 1.
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.0125,
|
| 41 |
-
"grad_norm": 9.
|
| 42 |
"learning_rate": 4.537499999999999e-06,
|
| 43 |
"loss": 1.3704,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.015,
|
| 48 |
-
"grad_norm": 8.
|
| 49 |
"learning_rate": 5.474999999999999e-06,
|
| 50 |
-
"loss": 1.
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.0175,
|
| 55 |
-
"grad_norm": 9.
|
| 56 |
"learning_rate": 6.4125e-06,
|
| 57 |
"loss": 1.0771,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.02,
|
| 62 |
-
"grad_norm": 9.
|
| 63 |
"learning_rate": 7.35e-06,
|
| 64 |
"loss": 1.0148,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.0225,
|
| 69 |
-
"grad_norm": 7.
|
| 70 |
"learning_rate": 8.2875e-06,
|
| 71 |
-
"loss": 0.
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.025,
|
| 76 |
-
"grad_norm": 10.
|
| 77 |
"learning_rate": 9.224999999999999e-06,
|
| 78 |
-
"loss": 0.
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.0275,
|
| 83 |
-
"grad_norm": 8.
|
| 84 |
"learning_rate": 1.01625e-05,
|
| 85 |
"loss": 0.8811,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.03,
|
| 90 |
-
"grad_norm": 9.
|
| 91 |
"learning_rate": 1.1099999999999999e-05,
|
| 92 |
"loss": 0.8374,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.0325,
|
| 97 |
-
"grad_norm": 7.
|
| 98 |
"learning_rate": 1.20375e-05,
|
| 99 |
"loss": 0.7848,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.035,
|
| 104 |
-
"grad_norm": 7.
|
| 105 |
"learning_rate": 1.2974999999999999e-05,
|
| 106 |
"loss": 0.7056,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.0375,
|
| 111 |
-
"grad_norm": 8.
|
| 112 |
"learning_rate": 1.39125e-05,
|
| 113 |
-
"loss": 0.
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.04,
|
| 118 |
-
"grad_norm": 6.
|
| 119 |
"learning_rate": 1.485e-05,
|
| 120 |
-
"loss": 0.
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"epoch": 0.0425,
|
| 125 |
-
"grad_norm": 7.
|
| 126 |
"learning_rate": 1.5787499999999997e-05,
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.045,
|
| 132 |
-
"grad_norm": 6.
|
| 133 |
"learning_rate": 1.6725e-05,
|
| 134 |
"loss": 0.5099,
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.0475,
|
| 139 |
-
"grad_norm": 6.
|
| 140 |
"learning_rate": 1.76625e-05,
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 0.05,
|
| 146 |
-
"grad_norm": 6.
|
| 147 |
"learning_rate": 1.8599999999999998e-05,
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.0525,
|
| 153 |
-
"grad_norm": 7.
|
| 154 |
"learning_rate": 1.95375e-05,
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.055,
|
| 160 |
-
"grad_norm": 5.
|
| 161 |
"learning_rate": 2.0475e-05,
|
| 162 |
"loss": 0.4271,
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.0575,
|
| 167 |
-
"grad_norm": 5.
|
| 168 |
"learning_rate": 2.1412499999999995e-05,
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.06,
|
| 174 |
-
"grad_norm": 8.
|
| 175 |
"learning_rate": 2.2349999999999998e-05,
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.0625,
|
| 181 |
-
"grad_norm": 6.
|
| 182 |
"learning_rate": 2.3287499999999997e-05,
|
| 183 |
-
"loss": 0.
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.065,
|
| 188 |
-
"grad_norm": 5.
|
| 189 |
"learning_rate": 2.4225e-05,
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 0.0675,
|
| 195 |
-
"grad_norm": 5.
|
| 196 |
"learning_rate": 2.51625e-05,
|
| 197 |
-
"loss": 0.
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
"epoch": 0.07,
|
| 202 |
-
"grad_norm": 6.
|
| 203 |
"learning_rate": 2.6099999999999997e-05,
|
| 204 |
"loss": 0.3597,
|
| 205 |
"step": 700
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.0725,
|
| 209 |
-
"grad_norm": 8.
|
| 210 |
"learning_rate": 2.7037499999999997e-05,
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 725
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.075,
|
| 216 |
-
"grad_norm": 6.
|
| 217 |
"learning_rate": 2.7975e-05,
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 750
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.0775,
|
| 223 |
-
"grad_norm": 6.
|
| 224 |
"learning_rate": 2.8912499999999998e-05,
|
| 225 |
-
"loss": 0.
|
| 226 |
"step": 775
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.08,
|
| 230 |
-
"grad_norm": 7.
|
| 231 |
"learning_rate": 2.985e-05,
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 800
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 0.0825,
|
| 237 |
-
"grad_norm": 5.
|
| 238 |
"learning_rate": 3.0787499999999996e-05,
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 825
|
| 241 |
},
|
| 242 |
{
|
| 243 |
"epoch": 0.085,
|
| 244 |
-
"grad_norm": 4.
|
| 245 |
"learning_rate": 3.1725e-05,
|
| 246 |
-
"loss": 0.
|
| 247 |
"step": 850
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"epoch": 0.0875,
|
| 251 |
-
"grad_norm": 4.
|
| 252 |
"learning_rate": 3.2662499999999994e-05,
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 875
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.09,
|
| 258 |
-
"grad_norm": 6.
|
| 259 |
"learning_rate": 3.36e-05,
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 900
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.0925,
|
| 265 |
-
"grad_norm": 7.
|
| 266 |
"learning_rate": 3.45375e-05,
|
| 267 |
"loss": 0.4779,
|
| 268 |
"step": 925
|
| 269 |
},
|
| 270 |
{
|
| 271 |
"epoch": 0.095,
|
| 272 |
-
"grad_norm": 6.
|
| 273 |
"learning_rate": 3.5474999999999995e-05,
|
| 274 |
-
"loss": 0.
|
| 275 |
"step": 950
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 0.0975,
|
| 279 |
-
"grad_norm": 7.
|
| 280 |
"learning_rate": 3.64125e-05,
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 975
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 0.1,
|
| 286 |
-
"grad_norm": 7.
|
| 287 |
"learning_rate": 3.735e-05,
|
| 288 |
-
"loss": 0.
|
| 289 |
"step": 1000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.1,
|
| 293 |
-
"eval_loss": 0.
|
| 294 |
-
"eval_runtime":
|
| 295 |
-
"eval_samples_per_second":
|
| 296 |
-
"eval_steps_per_second": 2.
|
| 297 |
-
"eval_wer":
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.1025,
|
| 302 |
-
"grad_norm": 6.
|
| 303 |
"learning_rate": 3.74125e-05,
|
| 304 |
"loss": 0.5033,
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.105,
|
| 309 |
-
"grad_norm": 5.
|
| 310 |
"learning_rate": 3.730833333333333e-05,
|
| 311 |
-
"loss": 0.
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 0.1075,
|
| 316 |
-
"grad_norm": 4.
|
| 317 |
"learning_rate": 3.7204166666666665e-05,
|
| 318 |
-
"loss": 0.
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.11,
|
| 323 |
-
"grad_norm": 4.
|
| 324 |
"learning_rate": 3.7099999999999994e-05,
|
| 325 |
-
"loss": 0.
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.1125,
|
| 330 |
-
"grad_norm": 4.
|
| 331 |
"learning_rate": 3.699583333333333e-05,
|
| 332 |
"loss": 0.2439,
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.115,
|
| 337 |
-
"grad_norm": 4.
|
| 338 |
"learning_rate": 3.6891666666666664e-05,
|
| 339 |
"loss": 0.2602,
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.1175,
|
| 344 |
-
"grad_norm":
|
| 345 |
"learning_rate": 3.678749999999999e-05,
|
| 346 |
-
"loss": 0.
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 0.12,
|
| 351 |
-
"grad_norm": 4.
|
| 352 |
"learning_rate": 3.668333333333333e-05,
|
| 353 |
-
"loss": 0.
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.1225,
|
| 358 |
-
"grad_norm":
|
| 359 |
"learning_rate": 3.6579166666666664e-05,
|
| 360 |
-
"loss": 0.
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
"epoch": 0.125,
|
| 365 |
-
"grad_norm": 6.
|
| 366 |
"learning_rate": 3.6475e-05,
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.1275,
|
| 372 |
-
"grad_norm": 6.
|
| 373 |
"learning_rate": 3.6370833333333334e-05,
|
| 374 |
-
"loss": 0.
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.13,
|
| 379 |
-
"grad_norm":
|
| 380 |
"learning_rate": 3.626666666666666e-05,
|
| 381 |
-
"loss": 0.
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
"epoch": 0.1325,
|
| 386 |
-
"grad_norm": 6.
|
| 387 |
"learning_rate": 3.61625e-05,
|
| 388 |
-
"loss": 0.
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.135,
|
| 393 |
-
"grad_norm": 6.
|
| 394 |
"learning_rate": 3.6058333333333333e-05,
|
| 395 |
-
"loss": 0.
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 0.1375,
|
| 400 |
-
"grad_norm": 6.
|
| 401 |
"learning_rate": 3.595416666666666e-05,
|
| 402 |
"loss": 0.3569,
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
"epoch": 0.14,
|
| 407 |
-
"grad_norm": 6.
|
| 408 |
"learning_rate": 3.585e-05,
|
| 409 |
-
"loss": 0.
|
| 410 |
"step": 1400
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 0.1425,
|
| 414 |
-
"grad_norm": 6.
|
| 415 |
"learning_rate": 3.5745833333333326e-05,
|
| 416 |
-
"loss": 0.
|
| 417 |
"step": 1425
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.145,
|
| 421 |
-
"grad_norm": 4.
|
| 422 |
"learning_rate": 3.564166666666666e-05,
|
| 423 |
-
"loss": 0.
|
| 424 |
"step": 1450
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 0.1475,
|
| 428 |
-
"grad_norm":
|
| 429 |
"learning_rate": 3.5537499999999996e-05,
|
| 430 |
-
"loss": 0.
|
| 431 |
"step": 1475
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.15,
|
| 435 |
-
"grad_norm": 5.
|
| 436 |
"learning_rate": 3.543333333333333e-05,
|
| 437 |
-
"loss": 0.
|
| 438 |
"step": 1500
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 0.1525,
|
| 442 |
-
"grad_norm": 3.
|
| 443 |
"learning_rate": 3.532916666666667e-05,
|
| 444 |
-
"loss": 0.
|
| 445 |
"step": 1525
|
| 446 |
},
|
| 447 |
{
|
| 448 |
"epoch": 0.155,
|
| 449 |
-
"grad_norm":
|
| 450 |
"learning_rate": 3.5224999999999996e-05,
|
| 451 |
-
"loss": 0.
|
| 452 |
"step": 1550
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 0.1575,
|
| 456 |
-
"grad_norm": 4.
|
| 457 |
"learning_rate": 3.512083333333333e-05,
|
| 458 |
-
"loss": 0.
|
| 459 |
"step": 1575
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.16,
|
| 463 |
-
"grad_norm": 4.
|
| 464 |
"learning_rate": 3.5016666666666666e-05,
|
| 465 |
-
"loss": 0.
|
| 466 |
"step": 1600
|
| 467 |
},
|
| 468 |
{
|
| 469 |
"epoch": 0.1625,
|
| 470 |
-
"grad_norm":
|
| 471 |
"learning_rate": 3.49125e-05,
|
| 472 |
-
"loss": 0.
|
| 473 |
"step": 1625
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 0.165,
|
| 477 |
-
"grad_norm": 5.
|
| 478 |
"learning_rate": 3.480833333333333e-05,
|
| 479 |
-
"loss": 0.
|
| 480 |
"step": 1650
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 0.1675,
|
| 484 |
-
"grad_norm":
|
| 485 |
"learning_rate": 3.4704166666666665e-05,
|
| 486 |
-
"loss": 0.
|
| 487 |
"step": 1675
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 0.17,
|
| 491 |
-
"grad_norm": 4.
|
| 492 |
"learning_rate": 3.4599999999999994e-05,
|
| 493 |
-
"loss": 0.
|
| 494 |
"step": 1700
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 0.1725,
|
| 498 |
-
"grad_norm": 4.
|
| 499 |
"learning_rate": 3.449583333333333e-05,
|
| 500 |
-
"loss": 0.
|
| 501 |
"step": 1725
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.175,
|
| 505 |
-
"grad_norm": 4.
|
| 506 |
"learning_rate": 3.4391666666666665e-05,
|
| 507 |
-
"loss": 0.
|
| 508 |
"step": 1750
|
| 509 |
},
|
| 510 |
{
|
| 511 |
"epoch": 0.1775,
|
| 512 |
-
"grad_norm": 4.
|
| 513 |
"learning_rate": 3.42875e-05,
|
| 514 |
-
"loss": 0.
|
| 515 |
"step": 1775
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 0.18,
|
| 519 |
-
"grad_norm": 4.
|
| 520 |
"learning_rate": 3.418333333333333e-05,
|
| 521 |
-
"loss": 0.
|
| 522 |
"step": 1800
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 0.1825,
|
| 526 |
-
"grad_norm": 3.
|
| 527 |
"learning_rate": 3.4079166666666664e-05,
|
| 528 |
-
"loss": 0.
|
| 529 |
"step": 1825
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 0.185,
|
| 533 |
-
"grad_norm": 3.
|
| 534 |
"learning_rate": 3.3975e-05,
|
| 535 |
-
"loss": 0.
|
| 536 |
"step": 1850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 0.1875,
|
| 540 |
-
"grad_norm":
|
| 541 |
"learning_rate": 3.3870833333333334e-05,
|
| 542 |
-
"loss": 0.
|
| 543 |
"step": 1875
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.19,
|
| 547 |
-
"grad_norm":
|
| 548 |
"learning_rate": 3.376666666666666e-05,
|
| 549 |
-
"loss": 0.
|
| 550 |
"step": 1900
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 0.1925,
|
| 554 |
-
"grad_norm": 3.
|
| 555 |
"learning_rate": 3.36625e-05,
|
| 556 |
-
"loss": 0.
|
| 557 |
"step": 1925
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 0.195,
|
| 561 |
-
"grad_norm": 4.
|
| 562 |
"learning_rate": 3.355833333333333e-05,
|
| 563 |
-
"loss": 0.
|
| 564 |
"step": 1950
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.1975,
|
| 568 |
-
"grad_norm": 5.
|
| 569 |
"learning_rate": 3.345416666666666e-05,
|
| 570 |
-
"loss": 0.
|
| 571 |
"step": 1975
|
| 572 |
},
|
| 573 |
{
|
| 574 |
"epoch": 0.2,
|
| 575 |
-
"grad_norm": 5.
|
| 576 |
"learning_rate": 3.335e-05,
|
| 577 |
-
"loss": 0.
|
| 578 |
"step": 2000
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 0.2,
|
| 582 |
-
"eval_loss": 0.
|
| 583 |
-
"eval_runtime":
|
| 584 |
-
"eval_samples_per_second":
|
| 585 |
-
"eval_steps_per_second": 2.
|
| 586 |
-
"eval_wer":
|
| 587 |
"step": 2000
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 0.2025,
|
| 591 |
-
"grad_norm":
|
| 592 |
"learning_rate": 3.324583333333333e-05,
|
| 593 |
-
"loss": 0.
|
| 594 |
"step": 2025
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.205,
|
| 598 |
-
"grad_norm":
|
| 599 |
"learning_rate": 3.314166666666666e-05,
|
| 600 |
-
"loss": 0.
|
| 601 |
"step": 2050
|
| 602 |
},
|
| 603 |
{
|
| 604 |
"epoch": 0.2075,
|
| 605 |
-
"grad_norm":
|
| 606 |
"learning_rate": 3.30375e-05,
|
| 607 |
-
"loss": 0.
|
| 608 |
"step": 2075
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 0.21,
|
| 612 |
-
"grad_norm": 3.
|
| 613 |
"learning_rate": 3.293333333333333e-05,
|
| 614 |
-
"loss": 0.
|
| 615 |
"step": 2100
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.2125,
|
| 619 |
-
"grad_norm": 3.
|
| 620 |
"learning_rate": 3.282916666666667e-05,
|
| 621 |
-
"loss": 0.
|
| 622 |
"step": 2125
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 0.215,
|
| 626 |
-
"grad_norm": 3.
|
| 627 |
"learning_rate": 3.2724999999999996e-05,
|
| 628 |
-
"loss": 0.
|
| 629 |
"step": 2150
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 0.2175,
|
| 633 |
-
"grad_norm": 4.
|
| 634 |
"learning_rate": 3.262083333333333e-05,
|
| 635 |
-
"loss": 0.
|
| 636 |
"step": 2175
|
| 637 |
},
|
| 638 |
{
|
| 639 |
"epoch": 0.22,
|
| 640 |
-
"grad_norm":
|
| 641 |
"learning_rate": 3.2516666666666666e-05,
|
| 642 |
-
"loss": 0.
|
| 643 |
"step": 2200
|
| 644 |
},
|
| 645 |
{
|
| 646 |
"epoch": 0.2225,
|
| 647 |
-
"grad_norm": 5.
|
| 648 |
"learning_rate": 3.2412499999999995e-05,
|
| 649 |
-
"loss": 0.
|
| 650 |
"step": 2225
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 0.225,
|
| 654 |
-
"grad_norm": 4.
|
| 655 |
"learning_rate": 3.230833333333333e-05,
|
| 656 |
-
"loss": 0.
|
| 657 |
"step": 2250
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.2275,
|
| 661 |
-
"grad_norm": 4.
|
| 662 |
"learning_rate": 3.2204166666666666e-05,
|
| 663 |
-
"loss": 0.
|
| 664 |
"step": 2275
|
| 665 |
},
|
| 666 |
{
|
| 667 |
"epoch": 0.23,
|
| 668 |
-
"grad_norm": 3.
|
| 669 |
"learning_rate": 3.2099999999999994e-05,
|
| 670 |
-
"loss": 0.
|
| 671 |
"step": 2300
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 0.2325,
|
| 675 |
-
"grad_norm":
|
| 676 |
"learning_rate": 3.199583333333333e-05,
|
| 677 |
-
"loss": 0.
|
| 678 |
"step": 2325
|
| 679 |
},
|
| 680 |
{
|
| 681 |
"epoch": 0.235,
|
| 682 |
-
"grad_norm":
|
| 683 |
"learning_rate": 3.1891666666666665e-05,
|
| 684 |
-
"loss": 0.
|
| 685 |
"step": 2350
|
| 686 |
},
|
| 687 |
{
|
| 688 |
"epoch": 0.2375,
|
| 689 |
-
"grad_norm": 2.
|
| 690 |
"learning_rate": 3.17875e-05,
|
| 691 |
-
"loss": 0.
|
| 692 |
"step": 2375
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 0.24,
|
| 696 |
-
"grad_norm": 3.
|
| 697 |
"learning_rate": 3.168333333333333e-05,
|
| 698 |
-
"loss": 0.
|
| 699 |
"step": 2400
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.2425,
|
| 703 |
-
"grad_norm":
|
| 704 |
"learning_rate": 3.1579166666666664e-05,
|
| 705 |
-
"loss": 0.
|
| 706 |
"step": 2425
|
| 707 |
},
|
| 708 |
{
|
| 709 |
"epoch": 0.245,
|
| 710 |
-
"grad_norm": 2.
|
| 711 |
"learning_rate": 3.1475e-05,
|
| 712 |
-
"loss": 0.
|
| 713 |
"step": 2450
|
| 714 |
},
|
| 715 |
{
|
| 716 |
"epoch": 0.2475,
|
| 717 |
-
"grad_norm": 3.
|
| 718 |
"learning_rate": 3.1370833333333335e-05,
|
| 719 |
-
"loss": 0.
|
| 720 |
"step": 2475
|
| 721 |
},
|
| 722 |
{
|
| 723 |
"epoch": 0.25,
|
| 724 |
-
"grad_norm": 3.
|
| 725 |
"learning_rate": 3.126666666666666e-05,
|
| 726 |
-
"loss": 0.
|
| 727 |
"step": 2500
|
| 728 |
},
|
| 729 |
{
|
| 730 |
"epoch": 0.2525,
|
| 731 |
-
"grad_norm":
|
| 732 |
"learning_rate": 3.11625e-05,
|
| 733 |
-
"loss": 0.
|
| 734 |
"step": 2525
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 0.255,
|
| 738 |
-
"grad_norm": 4.
|
| 739 |
"learning_rate": 3.105833333333333e-05,
|
| 740 |
-
"loss": 0.
|
| 741 |
"step": 2550
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 0.2575,
|
| 745 |
-
"grad_norm": 4.
|
| 746 |
"learning_rate": 3.095416666666666e-05,
|
| 747 |
-
"loss": 0.
|
| 748 |
"step": 2575
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 0.26,
|
| 752 |
-
"grad_norm": 3.
|
| 753 |
"learning_rate": 3.085e-05,
|
| 754 |
-
"loss": 0.
|
| 755 |
"step": 2600
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 0.2625,
|
| 759 |
-
"grad_norm":
|
| 760 |
"learning_rate": 3.074583333333333e-05,
|
| 761 |
"loss": 0.1741,
|
| 762 |
"step": 2625
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 0.265,
|
| 766 |
-
"grad_norm": 3.
|
| 767 |
"learning_rate": 3.064166666666666e-05,
|
| 768 |
-
"loss": 0.
|
| 769 |
"step": 2650
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 0.2675,
|
| 773 |
-
"grad_norm":
|
| 774 |
"learning_rate": 3.05375e-05,
|
| 775 |
-
"loss": 0.
|
| 776 |
"step": 2675
|
| 777 |
},
|
| 778 |
{
|
| 779 |
"epoch": 0.27,
|
| 780 |
-
"grad_norm":
|
| 781 |
"learning_rate": 3.0433333333333332e-05,
|
| 782 |
"loss": 0.1584,
|
| 783 |
"step": 2700
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 0.2725,
|
| 787 |
-
"grad_norm": 3.
|
| 788 |
"learning_rate": 3.0329166666666664e-05,
|
| 789 |
-
"loss": 0.
|
| 790 |
"step": 2725
|
| 791 |
},
|
| 792 |
{
|
| 793 |
"epoch": 0.275,
|
| 794 |
-
"grad_norm": 4.
|
| 795 |
"learning_rate": 3.0225e-05,
|
| 796 |
-
"loss": 0.
|
| 797 |
"step": 2750
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 0.2775,
|
| 801 |
-
"grad_norm": 5.
|
| 802 |
"learning_rate": 3.0120833333333328e-05,
|
| 803 |
-
"loss": 0.
|
| 804 |
"step": 2775
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 0.28,
|
| 808 |
-
"grad_norm": 4.
|
| 809 |
"learning_rate": 3.0016666666666663e-05,
|
| 810 |
-
"loss": 0.
|
| 811 |
"step": 2800
|
| 812 |
},
|
| 813 |
{
|
| 814 |
"epoch": 0.2825,
|
| 815 |
-
"grad_norm": 4.
|
| 816 |
"learning_rate": 2.9912499999999995e-05,
|
| 817 |
-
"loss": 0.
|
| 818 |
"step": 2825
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 0.285,
|
| 822 |
-
"grad_norm": 4.
|
| 823 |
"learning_rate": 2.980833333333333e-05,
|
| 824 |
-
"loss": 0.
|
| 825 |
"step": 2850
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 0.2875,
|
| 829 |
-
"grad_norm":
|
| 830 |
"learning_rate": 2.9704166666666662e-05,
|
| 831 |
"loss": 0.2561,
|
| 832 |
"step": 2875
|
| 833 |
},
|
| 834 |
{
|
| 835 |
"epoch": 0.29,
|
| 836 |
-
"grad_norm":
|
| 837 |
"learning_rate": 2.9599999999999998e-05,
|
| 838 |
-
"loss": 0.
|
| 839 |
"step": 2900
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 0.2925,
|
| 843 |
-
"grad_norm": 3.
|
| 844 |
"learning_rate": 2.949583333333333e-05,
|
| 845 |
-
"loss": 0.
|
| 846 |
"step": 2925
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 0.295,
|
| 850 |
-
"grad_norm": 3.
|
| 851 |
"learning_rate": 2.9391666666666665e-05,
|
| 852 |
-
"loss": 0.
|
| 853 |
"step": 2950
|
| 854 |
},
|
| 855 |
{
|
| 856 |
"epoch": 0.2975,
|
| 857 |
-
"grad_norm":
|
| 858 |
"learning_rate": 2.9287499999999997e-05,
|
| 859 |
-
"loss": 0.
|
| 860 |
"step": 2975
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 0.3,
|
| 864 |
-
"grad_norm":
|
| 865 |
"learning_rate": 2.9183333333333332e-05,
|
| 866 |
-
"loss": 0.
|
| 867 |
"step": 3000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 0.3,
|
| 871 |
-
"eval_loss": 0.
|
| 872 |
-
"eval_runtime":
|
| 873 |
-
"eval_samples_per_second":
|
| 874 |
-
"eval_steps_per_second": 2.
|
| 875 |
-
"eval_wer":
|
| 876 |
"step": 3000
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 0.3025,
|
| 880 |
-
"grad_norm": 5.
|
| 881 |
"learning_rate": 2.9079166666666664e-05,
|
| 882 |
-
"loss": 0.
|
| 883 |
"step": 3025
|
| 884 |
},
|
| 885 |
{
|
| 886 |
"epoch": 0.305,
|
| 887 |
-
"grad_norm": 6.
|
| 888 |
"learning_rate": 2.8974999999999996e-05,
|
| 889 |
-
"loss": 0.
|
| 890 |
"step": 3050
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 0.3075,
|
| 894 |
-
"grad_norm": 3.
|
| 895 |
"learning_rate": 2.8870833333333328e-05,
|
| 896 |
-
"loss": 0.
|
| 897 |
"step": 3075
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 0.31,
|
| 901 |
-
"grad_norm": 3.
|
| 902 |
"learning_rate": 2.8766666666666663e-05,
|
| 903 |
-
"loss": 0.
|
| 904 |
"step": 3100
|
| 905 |
},
|
| 906 |
{
|
| 907 |
"epoch": 0.3125,
|
| 908 |
-
"grad_norm": 3.
|
| 909 |
"learning_rate": 2.8662499999999995e-05,
|
| 910 |
-
"loss": 0.
|
| 911 |
"step": 3125
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 0.315,
|
| 915 |
-
"grad_norm": 4.
|
| 916 |
"learning_rate": 2.855833333333333e-05,
|
| 917 |
-
"loss": 0.
|
| 918 |
"step": 3150
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 0.3175,
|
| 922 |
-
"grad_norm": 4.
|
| 923 |
"learning_rate": 2.8454166666666663e-05,
|
| 924 |
-
"loss": 0.
|
| 925 |
"step": 3175
|
| 926 |
},
|
| 927 |
{
|
| 928 |
"epoch": 0.32,
|
| 929 |
-
"grad_norm": 5.
|
| 930 |
"learning_rate": 2.8349999999999998e-05,
|
| 931 |
-
"loss": 0.
|
| 932 |
"step": 3200
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 0.3225,
|
| 936 |
-
"grad_norm": 4.
|
| 937 |
"learning_rate": 2.824583333333333e-05,
|
| 938 |
-
"loss": 0.
|
| 939 |
"step": 3225
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 0.325,
|
| 943 |
-
"grad_norm":
|
| 944 |
"learning_rate": 2.8141666666666665e-05,
|
| 945 |
-
"loss": 0.
|
| 946 |
"step": 3250
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 0.3275,
|
| 950 |
-
"grad_norm": 5.
|
| 951 |
"learning_rate": 2.80375e-05,
|
| 952 |
-
"loss": 0.
|
| 953 |
"step": 3275
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 0.33,
|
| 957 |
-
"grad_norm": 3.
|
| 958 |
"learning_rate": 2.7933333333333332e-05,
|
| 959 |
-
"loss": 0.
|
| 960 |
"step": 3300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 0.3325,
|
| 964 |
-
"grad_norm": 3.
|
| 965 |
"learning_rate": 2.7829166666666668e-05,
|
| 966 |
-
"loss": 0.
|
| 967 |
"step": 3325
|
| 968 |
},
|
| 969 |
{
|
| 970 |
"epoch": 0.335,
|
| 971 |
-
"grad_norm": 3.
|
| 972 |
"learning_rate": 2.7724999999999996e-05,
|
| 973 |
-
"loss": 0.
|
| 974 |
"step": 3350
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 0.3375,
|
| 978 |
-
"grad_norm": 3.
|
| 979 |
"learning_rate": 2.7620833333333328e-05,
|
| 980 |
-
"loss": 0.
|
| 981 |
"step": 3375
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 0.34,
|
| 985 |
-
"grad_norm": 3.
|
| 986 |
"learning_rate": 2.7516666666666664e-05,
|
| 987 |
-
"loss": 0.
|
| 988 |
"step": 3400
|
| 989 |
},
|
| 990 |
{
|
| 991 |
"epoch": 0.3425,
|
| 992 |
-
"grad_norm": 3.
|
| 993 |
"learning_rate": 2.7412499999999995e-05,
|
| 994 |
-
"loss": 0.
|
| 995 |
"step": 3425
|
| 996 |
},
|
| 997 |
{
|
| 998 |
"epoch": 0.345,
|
| 999 |
-
"grad_norm": 2.
|
| 1000 |
"learning_rate": 2.730833333333333e-05,
|
| 1001 |
-
"loss": 0.
|
| 1002 |
"step": 3450
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 0.3475,
|
| 1006 |
-
"grad_norm": 3.
|
| 1007 |
"learning_rate": 2.7204166666666663e-05,
|
| 1008 |
-
"loss": 0.
|
| 1009 |
"step": 3475
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 0.35,
|
| 1013 |
-
"grad_norm": 4.
|
| 1014 |
"learning_rate": 2.7099999999999998e-05,
|
| 1015 |
-
"loss": 0.
|
| 1016 |
"step": 3500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 0.3525,
|
| 1020 |
-
"grad_norm": 3.
|
| 1021 |
"learning_rate": 2.6995833333333333e-05,
|
| 1022 |
-
"loss": 0.
|
| 1023 |
"step": 3525
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 0.355,
|
| 1027 |
-
"grad_norm": 5.
|
| 1028 |
"learning_rate": 2.6891666666666665e-05,
|
| 1029 |
-
"loss": 0.
|
| 1030 |
"step": 3550
|
| 1031 |
},
|
| 1032 |
{
|
| 1033 |
"epoch": 0.3575,
|
| 1034 |
-
"grad_norm": 5.
|
| 1035 |
"learning_rate": 2.67875e-05,
|
| 1036 |
-
"loss": 0.
|
| 1037 |
"step": 3575
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 0.36,
|
| 1041 |
-
"grad_norm": 4.
|
| 1042 |
"learning_rate": 2.6683333333333333e-05,
|
| 1043 |
-
"loss": 0.
|
| 1044 |
"step": 3600
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 0.3625,
|
| 1048 |
-
"grad_norm": 4.
|
| 1049 |
"learning_rate": 2.657916666666666e-05,
|
| 1050 |
-
"loss": 0.
|
| 1051 |
"step": 3625
|
| 1052 |
},
|
| 1053 |
{
|
| 1054 |
"epoch": 0.365,
|
| 1055 |
-
"grad_norm": 2.
|
| 1056 |
"learning_rate": 2.6474999999999996e-05,
|
| 1057 |
-
"loss": 0.
|
| 1058 |
"step": 3650
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 0.3675,
|
| 1062 |
-
"grad_norm": 3.
|
| 1063 |
"learning_rate": 2.637083333333333e-05,
|
| 1064 |
-
"loss": 0.
|
| 1065 |
"step": 3675
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 0.37,
|
| 1069 |
-
"grad_norm":
|
| 1070 |
"learning_rate": 2.6266666666666664e-05,
|
| 1071 |
-
"loss": 0.
|
| 1072 |
"step": 3700
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 0.3725,
|
| 1076 |
-
"grad_norm": 3.
|
| 1077 |
"learning_rate": 2.61625e-05,
|
| 1078 |
-
"loss": 0.
|
| 1079 |
"step": 3725
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 0.375,
|
| 1083 |
-
"grad_norm": 4.
|
| 1084 |
"learning_rate": 2.605833333333333e-05,
|
| 1085 |
-
"loss": 0.
|
| 1086 |
"step": 3750
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 0.3775,
|
| 1090 |
-
"grad_norm": 3.
|
| 1091 |
"learning_rate": 2.5954166666666666e-05,
|
| 1092 |
-
"loss": 0.
|
| 1093 |
"step": 3775
|
| 1094 |
},
|
| 1095 |
{
|
| 1096 |
"epoch": 0.38,
|
| 1097 |
-
"grad_norm":
|
| 1098 |
"learning_rate": 2.5849999999999998e-05,
|
| 1099 |
-
"loss": 0.
|
| 1100 |
"step": 3800
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 0.3825,
|
| 1104 |
-
"grad_norm":
|
| 1105 |
"learning_rate": 2.5745833333333333e-05,
|
| 1106 |
-
"loss": 0.
|
| 1107 |
"step": 3825
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 0.385,
|
| 1111 |
-
"grad_norm": 5.
|
| 1112 |
"learning_rate": 2.5641666666666665e-05,
|
| 1113 |
-
"loss": 0.
|
| 1114 |
"step": 3850
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
"epoch": 0.3875,
|
| 1118 |
-
"grad_norm": 4.
|
| 1119 |
"learning_rate": 2.55375e-05,
|
| 1120 |
-
"loss": 0.
|
| 1121 |
"step": 3875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 0.39,
|
| 1125 |
-
"grad_norm":
|
| 1126 |
"learning_rate": 2.543333333333333e-05,
|
| 1127 |
-
"loss": 0.
|
| 1128 |
"step": 3900
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 0.3925,
|
| 1132 |
-
"grad_norm":
|
| 1133 |
"learning_rate": 2.5329166666666665e-05,
|
| 1134 |
-
"loss": 0.
|
| 1135 |
"step": 3925
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
"epoch": 0.395,
|
| 1139 |
-
"grad_norm": 3.
|
| 1140 |
"learning_rate": 2.5224999999999997e-05,
|
| 1141 |
-
"loss": 0.
|
| 1142 |
"step": 3950
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 0.3975,
|
| 1146 |
-
"grad_norm": 6.
|
| 1147 |
"learning_rate": 2.5120833333333332e-05,
|
| 1148 |
-
"loss": 0.
|
| 1149 |
"step": 3975
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 0.4,
|
| 1153 |
-
"grad_norm": 4.
|
| 1154 |
"learning_rate": 2.5016666666666664e-05,
|
| 1155 |
-
"loss": 0.
|
| 1156 |
"step": 4000
|
| 1157 |
},
|
| 1158 |
{
|
| 1159 |
"epoch": 0.4,
|
| 1160 |
-
"eval_loss": 0.
|
| 1161 |
-
"eval_runtime":
|
| 1162 |
-
"eval_samples_per_second":
|
| 1163 |
-
"eval_steps_per_second": 2.
|
| 1164 |
-
"eval_wer":
|
| 1165 |
"step": 4000
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
"epoch": 0.4025,
|
| 1169 |
-
"grad_norm": 3.
|
| 1170 |
"learning_rate": 2.49125e-05,
|
| 1171 |
-
"loss": 0.
|
| 1172 |
"step": 4025
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 0.405,
|
| 1176 |
-
"grad_norm": 5.
|
| 1177 |
"learning_rate": 2.480833333333333e-05,
|
| 1178 |
-
"loss": 0.
|
| 1179 |
"step": 4050
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 0.4075,
|
| 1183 |
-
"grad_norm": 4.
|
| 1184 |
"learning_rate": 2.4704166666666666e-05,
|
| 1185 |
-
"loss": 0.
|
| 1186 |
"step": 4075
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 0.41,
|
| 1190 |
-
"grad_norm": 3.
|
| 1191 |
"learning_rate": 2.4599999999999998e-05,
|
| 1192 |
-
"loss": 0.
|
| 1193 |
"step": 4100
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
"epoch": 0.4125,
|
| 1197 |
-
"grad_norm": 2.
|
| 1198 |
"learning_rate": 2.4495833333333334e-05,
|
| 1199 |
-
"loss": 0.
|
| 1200 |
"step": 4125
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 0.415,
|
| 1204 |
-
"grad_norm": 3.
|
| 1205 |
"learning_rate": 2.4391666666666666e-05,
|
| 1206 |
-
"loss": 0.
|
| 1207 |
"step": 4150
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
"epoch": 0.4175,
|
| 1211 |
-
"grad_norm": 3.
|
| 1212 |
"learning_rate": 2.4287499999999997e-05,
|
| 1213 |
-
"loss": 0.
|
| 1214 |
"step": 4175
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 0.42,
|
| 1218 |
-
"grad_norm":
|
| 1219 |
"learning_rate": 2.418333333333333e-05,
|
| 1220 |
-
"loss": 0.
|
| 1221 |
"step": 4200
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 0.4225,
|
| 1225 |
-
"grad_norm": 2.
|
| 1226 |
"learning_rate": 2.4079166666666665e-05,
|
| 1227 |
-
"loss": 0.
|
| 1228 |
"step": 4225
|
| 1229 |
},
|
| 1230 |
{
|
| 1231 |
"epoch": 0.425,
|
| 1232 |
-
"grad_norm": 3.
|
| 1233 |
"learning_rate": 2.3974999999999997e-05,
|
| 1234 |
-
"loss": 0.
|
| 1235 |
"step": 4250
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 0.4275,
|
| 1239 |
-
"grad_norm": 3.
|
| 1240 |
"learning_rate": 2.3870833333333332e-05,
|
| 1241 |
-
"loss": 0.
|
| 1242 |
"step": 4275
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 0.43,
|
| 1246 |
-
"grad_norm": 4.
|
| 1247 |
"learning_rate": 2.3766666666666664e-05,
|
| 1248 |
-
"loss": 0.
|
| 1249 |
"step": 4300
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 0.4325,
|
| 1253 |
-
"grad_norm": 4.
|
| 1254 |
"learning_rate": 2.36625e-05,
|
| 1255 |
-
"loss": 0.
|
| 1256 |
"step": 4325
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 0.435,
|
| 1260 |
-
"grad_norm":
|
| 1261 |
"learning_rate": 2.355833333333333e-05,
|
| 1262 |
-
"loss": 0.
|
| 1263 |
"step": 4350
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 0.4375,
|
| 1267 |
-
"grad_norm": 2.
|
| 1268 |
"learning_rate": 2.3454166666666666e-05,
|
| 1269 |
-
"loss": 0.
|
| 1270 |
"step": 4375
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 0.44,
|
| 1274 |
-
"grad_norm":
|
| 1275 |
"learning_rate": 2.335e-05,
|
| 1276 |
-
"loss": 0.
|
| 1277 |
"step": 4400
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 0.4425,
|
| 1281 |
-
"grad_norm": 3.
|
| 1282 |
"learning_rate": 2.3245833333333334e-05,
|
| 1283 |
-
"loss": 0.
|
| 1284 |
"step": 4425
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 0.445,
|
| 1288 |
-
"grad_norm": 3.
|
| 1289 |
"learning_rate": 2.3141666666666666e-05,
|
| 1290 |
"loss": 0.1466,
|
| 1291 |
"step": 4450
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 0.4475,
|
| 1295 |
-
"grad_norm": 3.
|
| 1296 |
"learning_rate": 2.3037499999999998e-05,
|
| 1297 |
-
"loss": 0.
|
| 1298 |
"step": 4475
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 0.45,
|
| 1302 |
-
"grad_norm": 3.
|
| 1303 |
"learning_rate": 2.293333333333333e-05,
|
| 1304 |
-
"loss": 0.
|
| 1305 |
"step": 4500
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 0.4525,
|
| 1309 |
-
"grad_norm": 4.
|
| 1310 |
"learning_rate": 2.2829166666666665e-05,
|
| 1311 |
"loss": 0.1807,
|
| 1312 |
"step": 4525
|
| 1313 |
},
|
| 1314 |
{
|
| 1315 |
"epoch": 0.455,
|
| 1316 |
-
"grad_norm": 4.
|
| 1317 |
"learning_rate": 2.2724999999999997e-05,
|
| 1318 |
-
"loss": 0.
|
| 1319 |
"step": 4550
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 0.4575,
|
| 1323 |
-
"grad_norm": 5.
|
| 1324 |
"learning_rate": 2.2620833333333332e-05,
|
| 1325 |
-
"loss": 0.
|
| 1326 |
"step": 4575
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 0.46,
|
| 1330 |
-
"grad_norm": 4.
|
| 1331 |
"learning_rate": 2.2516666666666664e-05,
|
| 1332 |
-
"loss": 0.
|
| 1333 |
"step": 4600
|
| 1334 |
},
|
| 1335 |
{
|
| 1336 |
"epoch": 0.4625,
|
| 1337 |
-
"grad_norm": 4.
|
| 1338 |
"learning_rate": 2.24125e-05,
|
| 1339 |
-
"loss": 0.
|
| 1340 |
"step": 4625
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 0.465,
|
| 1344 |
-
"grad_norm": 5.
|
| 1345 |
"learning_rate": 2.230833333333333e-05,
|
| 1346 |
-
"loss": 0.
|
| 1347 |
"step": 4650
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 0.4675,
|
| 1351 |
-
"grad_norm": 4.
|
| 1352 |
"learning_rate": 2.2204166666666667e-05,
|
| 1353 |
-
"loss": 0.
|
| 1354 |
"step": 4675
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
"epoch": 0.47,
|
| 1358 |
-
"grad_norm": 3.
|
| 1359 |
"learning_rate": 2.21e-05,
|
| 1360 |
-
"loss": 0.
|
| 1361 |
"step": 4700
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 0.4725,
|
| 1365 |
-
"grad_norm": 4.
|
| 1366 |
"learning_rate": 2.1995833333333334e-05,
|
| 1367 |
"loss": 0.1953,
|
| 1368 |
"step": 4725
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 0.475,
|
| 1372 |
-
"grad_norm": 3.
|
| 1373 |
"learning_rate": 2.1891666666666662e-05,
|
| 1374 |
-
"loss": 0.
|
| 1375 |
"step": 4750
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 0.4775,
|
| 1379 |
-
"grad_norm": 3.
|
| 1380 |
"learning_rate": 2.1787499999999998e-05,
|
| 1381 |
-
"loss": 0.
|
| 1382 |
"step": 4775
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 0.48,
|
| 1386 |
-
"grad_norm":
|
| 1387 |
"learning_rate": 2.168333333333333e-05,
|
| 1388 |
-
"loss": 0.
|
| 1389 |
"step": 4800
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 0.4825,
|
| 1393 |
-
"grad_norm": 3.
|
| 1394 |
"learning_rate": 2.1579166666666665e-05,
|
| 1395 |
-
"loss": 0.
|
| 1396 |
"step": 4825
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 0.485,
|
| 1400 |
-
"grad_norm": 3.
|
| 1401 |
"learning_rate": 2.1474999999999997e-05,
|
| 1402 |
-
"loss": 0.
|
| 1403 |
"step": 4850
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 0.4875,
|
| 1407 |
-
"grad_norm":
|
| 1408 |
"learning_rate": 2.1370833333333332e-05,
|
| 1409 |
-
"loss": 0.
|
| 1410 |
"step": 4875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 0.49,
|
| 1414 |
-
"grad_norm": 4.
|
| 1415 |
"learning_rate": 2.1266666666666664e-05,
|
| 1416 |
"loss": 0.1834,
|
| 1417 |
"step": 4900
|
| 1418 |
},
|
| 1419 |
{
|
| 1420 |
"epoch": 0.4925,
|
| 1421 |
-
"grad_norm":
|
| 1422 |
"learning_rate": 2.11625e-05,
|
| 1423 |
-
"loss": 0.
|
| 1424 |
"step": 4925
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 0.495,
|
| 1428 |
-
"grad_norm": 3.
|
| 1429 |
"learning_rate": 2.105833333333333e-05,
|
| 1430 |
-
"loss": 0.
|
| 1431 |
"step": 4950
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 0.4975,
|
| 1435 |
-
"grad_norm": 3.
|
| 1436 |
"learning_rate": 2.0954166666666667e-05,
|
| 1437 |
-
"loss": 0.
|
| 1438 |
"step": 4975
|
| 1439 |
},
|
| 1440 |
{
|
| 1441 |
"epoch": 0.5,
|
| 1442 |
-
"grad_norm": 3.
|
| 1443 |
"learning_rate": 2.085e-05,
|
| 1444 |
-
"loss": 0.
|
| 1445 |
"step": 5000
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 0.5,
|
| 1449 |
-
"eval_loss": 0.
|
| 1450 |
-
"eval_runtime":
|
| 1451 |
-
"eval_samples_per_second":
|
| 1452 |
-
"eval_steps_per_second": 2.
|
| 1453 |
-
"eval_wer":
|
| 1454 |
"step": 5000
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
"epoch": 0.5025,
|
| 1458 |
-
"grad_norm": 3.
|
| 1459 |
"learning_rate": 2.074583333333333e-05,
|
| 1460 |
-
"loss": 0.
|
| 1461 |
"step": 5025
|
| 1462 |
},
|
| 1463 |
{
|
| 1464 |
"epoch": 0.505,
|
| 1465 |
-
"grad_norm": 3.
|
| 1466 |
"learning_rate": 2.0641666666666662e-05,
|
| 1467 |
-
"loss": 0.
|
| 1468 |
"step": 5050
|
| 1469 |
},
|
| 1470 |
{
|
| 1471 |
"epoch": 0.5075,
|
| 1472 |
-
"grad_norm": 4.
|
| 1473 |
"learning_rate": 2.0537499999999998e-05,
|
| 1474 |
-
"loss": 0.
|
| 1475 |
"step": 5075
|
| 1476 |
},
|
| 1477 |
{
|
| 1478 |
"epoch": 0.51,
|
| 1479 |
-
"grad_norm": 4.
|
| 1480 |
"learning_rate": 2.043333333333333e-05,
|
| 1481 |
-
"loss": 0.
|
| 1482 |
"step": 5100
|
| 1483 |
},
|
| 1484 |
{
|
| 1485 |
"epoch": 0.5125,
|
| 1486 |
-
"grad_norm": 4.
|
| 1487 |
"learning_rate": 2.0329166666666665e-05,
|
| 1488 |
-
"loss": 0.
|
| 1489 |
"step": 5125
|
| 1490 |
},
|
| 1491 |
{
|
| 1492 |
"epoch": 0.515,
|
| 1493 |
-
"grad_norm": 5.
|
| 1494 |
"learning_rate": 2.0224999999999997e-05,
|
| 1495 |
-
"loss": 0.
|
| 1496 |
"step": 5150
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
"epoch": 0.5175,
|
| 1500 |
-
"grad_norm": 2.
|
| 1501 |
"learning_rate": 2.0120833333333332e-05,
|
| 1502 |
-
"loss": 0.
|
| 1503 |
"step": 5175
|
| 1504 |
},
|
| 1505 |
{
|
| 1506 |
"epoch": 0.52,
|
| 1507 |
-
"grad_norm": 2.
|
| 1508 |
"learning_rate": 2.0016666666666664e-05,
|
| 1509 |
-
"loss": 0.
|
| 1510 |
"step": 5200
|
| 1511 |
},
|
| 1512 |
{
|
| 1513 |
"epoch": 0.5225,
|
| 1514 |
-
"grad_norm": 2.
|
| 1515 |
"learning_rate": 1.99125e-05,
|
| 1516 |
-
"loss": 0.
|
| 1517 |
"step": 5225
|
| 1518 |
},
|
| 1519 |
{
|
| 1520 |
"epoch": 0.525,
|
| 1521 |
-
"grad_norm": 2.
|
| 1522 |
"learning_rate": 1.980833333333333e-05,
|
| 1523 |
-
"loss": 0.
|
| 1524 |
"step": 5250
|
| 1525 |
},
|
| 1526 |
{
|
| 1527 |
"epoch": 0.5275,
|
| 1528 |
-
"grad_norm": 2.
|
| 1529 |
"learning_rate": 1.9704166666666667e-05,
|
| 1530 |
-
"loss": 0.
|
| 1531 |
"step": 5275
|
| 1532 |
},
|
| 1533 |
{
|
| 1534 |
"epoch": 0.53,
|
| 1535 |
-
"grad_norm": 3.
|
| 1536 |
"learning_rate": 1.9599999999999995e-05,
|
| 1537 |
"loss": 0.1287,
|
| 1538 |
"step": 5300
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
"epoch": 0.5325,
|
| 1542 |
-
"grad_norm": 3.
|
| 1543 |
"learning_rate": 1.949583333333333e-05,
|
| 1544 |
-
"loss": 0.
|
| 1545 |
"step": 5325
|
| 1546 |
},
|
| 1547 |
{
|
| 1548 |
"epoch": 0.535,
|
| 1549 |
-
"grad_norm": 4.
|
| 1550 |
"learning_rate": 1.9391666666666663e-05,
|
| 1551 |
-
"loss": 0.
|
| 1552 |
"step": 5350
|
| 1553 |
},
|
| 1554 |
{
|
| 1555 |
"epoch": 0.5375,
|
| 1556 |
-
"grad_norm": 4.
|
| 1557 |
"learning_rate": 1.9287499999999998e-05,
|
| 1558 |
-
"loss": 0.
|
| 1559 |
"step": 5375
|
| 1560 |
},
|
| 1561 |
{
|
| 1562 |
"epoch": 0.54,
|
| 1563 |
-
"grad_norm": 3.
|
| 1564 |
"learning_rate": 1.918333333333333e-05,
|
| 1565 |
-
"loss": 0.
|
| 1566 |
"step": 5400
|
| 1567 |
},
|
| 1568 |
{
|
| 1569 |
"epoch": 0.5425,
|
| 1570 |
-
"grad_norm": 4.
|
| 1571 |
"learning_rate": 1.9079166666666665e-05,
|
| 1572 |
-
"loss": 0.
|
| 1573 |
"step": 5425
|
| 1574 |
},
|
| 1575 |
{
|
| 1576 |
"epoch": 0.545,
|
| 1577 |
-
"grad_norm": 3.
|
| 1578 |
"learning_rate": 1.8974999999999997e-05,
|
| 1579 |
-
"loss": 0.
|
| 1580 |
"step": 5450
|
| 1581 |
},
|
| 1582 |
{
|
| 1583 |
"epoch": 0.5475,
|
| 1584 |
-
"grad_norm": 3.
|
| 1585 |
"learning_rate": 1.8870833333333332e-05,
|
| 1586 |
-
"loss": 0.
|
| 1587 |
"step": 5475
|
| 1588 |
},
|
| 1589 |
{
|
| 1590 |
"epoch": 0.55,
|
| 1591 |
-
"grad_norm": 3.
|
| 1592 |
"learning_rate": 1.8766666666666664e-05,
|
| 1593 |
-
"loss": 0.
|
| 1594 |
"step": 5500
|
| 1595 |
},
|
| 1596 |
{
|
| 1597 |
"epoch": 0.5525,
|
| 1598 |
-
"grad_norm": 5.
|
| 1599 |
"learning_rate": 1.8662499999999996e-05,
|
| 1600 |
-
"loss": 0.
|
| 1601 |
"step": 5525
|
| 1602 |
},
|
| 1603 |
{
|
| 1604 |
"epoch": 0.555,
|
| 1605 |
-
"grad_norm": 4.
|
| 1606 |
"learning_rate": 1.855833333333333e-05,
|
| 1607 |
-
"loss": 0.
|
| 1608 |
"step": 5550
|
| 1609 |
},
|
| 1610 |
{
|
| 1611 |
"epoch": 0.5575,
|
| 1612 |
-
"grad_norm": 4.
|
| 1613 |
"learning_rate": 1.8454166666666663e-05,
|
| 1614 |
-
"loss": 0.
|
| 1615 |
"step": 5575
|
| 1616 |
},
|
| 1617 |
{
|
| 1618 |
"epoch": 0.56,
|
| 1619 |
-
"grad_norm": 2.
|
| 1620 |
"learning_rate": 1.835e-05,
|
| 1621 |
-
"loss": 0.
|
| 1622 |
"step": 5600
|
| 1623 |
},
|
| 1624 |
{
|
| 1625 |
"epoch": 0.5625,
|
| 1626 |
-
"grad_norm": 3.
|
| 1627 |
"learning_rate": 1.8245833333333334e-05,
|
| 1628 |
-
"loss": 0.
|
| 1629 |
"step": 5625
|
| 1630 |
},
|
| 1631 |
{
|
| 1632 |
"epoch": 0.565,
|
| 1633 |
-
"grad_norm": 2.
|
| 1634 |
"learning_rate": 1.8141666666666663e-05,
|
| 1635 |
-
"loss": 0.
|
| 1636 |
"step": 5650
|
| 1637 |
},
|
| 1638 |
{
|
| 1639 |
"epoch": 0.5675,
|
| 1640 |
-
"grad_norm": 2.
|
| 1641 |
"learning_rate": 1.8037499999999998e-05,
|
| 1642 |
-
"loss": 0.
|
| 1643 |
"step": 5675
|
| 1644 |
},
|
| 1645 |
{
|
| 1646 |
"epoch": 0.57,
|
| 1647 |
-
"grad_norm": 3.
|
| 1648 |
"learning_rate": 1.793333333333333e-05,
|
| 1649 |
-
"loss": 0.
|
| 1650 |
"step": 5700
|
| 1651 |
},
|
| 1652 |
{
|
| 1653 |
"epoch": 0.5725,
|
| 1654 |
-
"grad_norm": 2.
|
| 1655 |
"learning_rate": 1.7829166666666665e-05,
|
| 1656 |
-
"loss": 0.
|
| 1657 |
"step": 5725
|
| 1658 |
},
|
| 1659 |
{
|
| 1660 |
"epoch": 0.575,
|
| 1661 |
-
"grad_norm": 2.
|
| 1662 |
"learning_rate": 1.7725e-05,
|
| 1663 |
-
"loss": 0.
|
| 1664 |
"step": 5750
|
| 1665 |
},
|
| 1666 |
{
|
| 1667 |
"epoch": 0.5775,
|
| 1668 |
-
"grad_norm": 2.
|
| 1669 |
"learning_rate": 1.7620833333333332e-05,
|
| 1670 |
-
"loss": 0.
|
| 1671 |
"step": 5775
|
| 1672 |
},
|
| 1673 |
{
|
| 1674 |
"epoch": 0.58,
|
| 1675 |
-
"grad_norm": 2.
|
| 1676 |
"learning_rate": 1.7516666666666664e-05,
|
| 1677 |
-
"loss": 0.
|
| 1678 |
"step": 5800
|
| 1679 |
},
|
| 1680 |
{
|
| 1681 |
"epoch": 0.5825,
|
| 1682 |
-
"grad_norm": 2.
|
| 1683 |
"learning_rate": 1.7412499999999996e-05,
|
| 1684 |
-
"loss": 0.
|
| 1685 |
"step": 5825
|
| 1686 |
},
|
| 1687 |
{
|
| 1688 |
"epoch": 0.585,
|
| 1689 |
-
"grad_norm": 3.
|
| 1690 |
"learning_rate": 1.730833333333333e-05,
|
| 1691 |
-
"loss": 0.
|
| 1692 |
"step": 5850
|
| 1693 |
},
|
| 1694 |
{
|
| 1695 |
"epoch": 0.5875,
|
| 1696 |
-
"grad_norm": 3.
|
| 1697 |
"learning_rate": 1.7204166666666667e-05,
|
| 1698 |
-
"loss": 0.
|
| 1699 |
"step": 5875
|
| 1700 |
},
|
| 1701 |
{
|
| 1702 |
"epoch": 0.59,
|
| 1703 |
-
"grad_norm": 3.
|
| 1704 |
"learning_rate": 1.71e-05,
|
| 1705 |
-
"loss": 0.
|
| 1706 |
"step": 5900
|
| 1707 |
},
|
| 1708 |
{
|
| 1709 |
"epoch": 0.5925,
|
| 1710 |
-
"grad_norm":
|
| 1711 |
"learning_rate": 1.6995833333333334e-05,
|
| 1712 |
"loss": 0.2024,
|
| 1713 |
"step": 5925
|
| 1714 |
},
|
| 1715 |
{
|
| 1716 |
"epoch": 0.595,
|
| 1717 |
-
"grad_norm": 3.
|
| 1718 |
"learning_rate": 1.6891666666666663e-05,
|
| 1719 |
-
"loss": 0.
|
| 1720 |
"step": 5950
|
| 1721 |
},
|
| 1722 |
{
|
| 1723 |
"epoch": 0.5975,
|
| 1724 |
-
"grad_norm": 4.
|
| 1725 |
"learning_rate": 1.6787499999999998e-05,
|
| 1726 |
-
"loss": 0.
|
| 1727 |
"step": 5975
|
| 1728 |
},
|
| 1729 |
{
|
| 1730 |
"epoch": 0.6,
|
| 1731 |
-
"grad_norm": 3.
|
| 1732 |
"learning_rate": 1.6683333333333333e-05,
|
| 1733 |
-
"loss": 0.
|
| 1734 |
"step": 6000
|
| 1735 |
},
|
| 1736 |
{
|
| 1737 |
"epoch": 0.6,
|
| 1738 |
-
"eval_loss": 0.
|
| 1739 |
-
"eval_runtime":
|
| 1740 |
-
"eval_samples_per_second":
|
| 1741 |
-
"eval_steps_per_second": 2.
|
| 1742 |
-
"eval_wer": 17.
|
| 1743 |
"step": 6000
|
| 1744 |
},
|
| 1745 |
{
|
| 1746 |
"epoch": 0.6025,
|
| 1747 |
-
"grad_norm": 3.
|
| 1748 |
"learning_rate": 1.6579166666666665e-05,
|
| 1749 |
-
"loss": 0.
|
| 1750 |
"step": 6025
|
| 1751 |
},
|
| 1752 |
{
|
| 1753 |
"epoch": 0.605,
|
| 1754 |
-
"grad_norm": 3.
|
| 1755 |
"learning_rate": 1.6475e-05,
|
| 1756 |
-
"loss": 0.
|
| 1757 |
"step": 6050
|
| 1758 |
},
|
| 1759 |
{
|
| 1760 |
"epoch": 0.6075,
|
| 1761 |
-
"grad_norm": 2.
|
| 1762 |
"learning_rate": 1.6370833333333333e-05,
|
| 1763 |
-
"loss": 0.
|
| 1764 |
"step": 6075
|
| 1765 |
},
|
| 1766 |
{
|
| 1767 |
"epoch": 0.61,
|
| 1768 |
-
"grad_norm": 2.
|
| 1769 |
"learning_rate": 1.6266666666666665e-05,
|
| 1770 |
-
"loss": 0.
|
| 1771 |
"step": 6100
|
| 1772 |
},
|
| 1773 |
{
|
| 1774 |
"epoch": 0.6125,
|
| 1775 |
-
"grad_norm": 3.
|
| 1776 |
"learning_rate": 1.61625e-05,
|
| 1777 |
-
"loss": 0.
|
| 1778 |
"step": 6125
|
| 1779 |
},
|
| 1780 |
{
|
| 1781 |
"epoch": 0.615,
|
| 1782 |
-
"grad_norm": 3.
|
| 1783 |
"learning_rate": 1.6058333333333332e-05,
|
| 1784 |
-
"loss": 0.
|
| 1785 |
"step": 6150
|
| 1786 |
},
|
| 1787 |
{
|
| 1788 |
"epoch": 0.6175,
|
| 1789 |
-
"grad_norm": 3.
|
| 1790 |
"learning_rate": 1.5954166666666667e-05,
|
| 1791 |
-
"loss": 0.
|
| 1792 |
"step": 6175
|
| 1793 |
},
|
| 1794 |
{
|
| 1795 |
"epoch": 0.62,
|
| 1796 |
-
"grad_norm": 4.
|
| 1797 |
"learning_rate": 1.585e-05,
|
| 1798 |
-
"loss": 0.
|
| 1799 |
"step": 6200
|
| 1800 |
},
|
| 1801 |
{
|
| 1802 |
"epoch": 0.6225,
|
| 1803 |
-
"grad_norm": 3.
|
| 1804 |
"learning_rate": 1.574583333333333e-05,
|
| 1805 |
-
"loss": 0.
|
| 1806 |
"step": 6225
|
| 1807 |
},
|
| 1808 |
{
|
| 1809 |
"epoch": 0.625,
|
| 1810 |
-
"grad_norm": 3.
|
| 1811 |
"learning_rate": 1.5641666666666666e-05,
|
| 1812 |
-
"loss": 0.
|
| 1813 |
"step": 6250
|
| 1814 |
},
|
| 1815 |
{
|
| 1816 |
"epoch": 0.6275,
|
| 1817 |
-
"grad_norm": 3.
|
| 1818 |
"learning_rate": 1.5537499999999998e-05,
|
| 1819 |
-
"loss": 0.
|
| 1820 |
"step": 6275
|
| 1821 |
},
|
| 1822 |
{
|
| 1823 |
"epoch": 0.63,
|
| 1824 |
-
"grad_norm":
|
| 1825 |
"learning_rate": 1.5433333333333334e-05,
|
| 1826 |
-
"loss": 0.
|
| 1827 |
"step": 6300
|
| 1828 |
},
|
| 1829 |
{
|
| 1830 |
"epoch": 0.6325,
|
| 1831 |
-
"grad_norm": 3.
|
| 1832 |
"learning_rate": 1.5329166666666665e-05,
|
| 1833 |
-
"loss": 0.
|
| 1834 |
"step": 6325
|
| 1835 |
},
|
| 1836 |
{
|
| 1837 |
"epoch": 0.635,
|
| 1838 |
-
"grad_norm": 3.
|
| 1839 |
"learning_rate": 1.5224999999999999e-05,
|
| 1840 |
-
"loss": 0.
|
| 1841 |
"step": 6350
|
| 1842 |
},
|
| 1843 |
{
|
| 1844 |
"epoch": 0.6375,
|
| 1845 |
-
"grad_norm": 3.
|
| 1846 |
"learning_rate": 1.5120833333333331e-05,
|
| 1847 |
-
"loss": 0.
|
| 1848 |
"step": 6375
|
| 1849 |
},
|
| 1850 |
{
|
| 1851 |
"epoch": 0.64,
|
| 1852 |
-
"grad_norm": 2.
|
| 1853 |
"learning_rate": 1.5016666666666665e-05,
|
| 1854 |
-
"loss": 0.
|
| 1855 |
"step": 6400
|
| 1856 |
},
|
| 1857 |
{
|
| 1858 |
"epoch": 0.6425,
|
| 1859 |
-
"grad_norm": 2.
|
| 1860 |
"learning_rate": 1.4912499999999998e-05,
|
| 1861 |
-
"loss": 0.
|
| 1862 |
"step": 6425
|
| 1863 |
},
|
| 1864 |
{
|
| 1865 |
"epoch": 0.645,
|
| 1866 |
-
"grad_norm":
|
| 1867 |
"learning_rate": 1.4808333333333332e-05,
|
| 1868 |
-
"loss": 0.
|
| 1869 |
"step": 6450
|
| 1870 |
},
|
| 1871 |
{
|
| 1872 |
"epoch": 0.6475,
|
| 1873 |
-
"grad_norm": 3.
|
| 1874 |
"learning_rate": 1.4704166666666666e-05,
|
| 1875 |
-
"loss": 0.
|
| 1876 |
"step": 6475
|
| 1877 |
},
|
| 1878 |
{
|
| 1879 |
"epoch": 0.65,
|
| 1880 |
-
"grad_norm": 3.
|
| 1881 |
"learning_rate": 1.4599999999999997e-05,
|
| 1882 |
-
"loss": 0.
|
| 1883 |
"step": 6500
|
| 1884 |
},
|
| 1885 |
{
|
| 1886 |
"epoch": 0.6525,
|
| 1887 |
-
"grad_norm": 5.
|
| 1888 |
"learning_rate": 1.4495833333333331e-05,
|
| 1889 |
-
"loss": 0.
|
| 1890 |
"step": 6525
|
| 1891 |
},
|
| 1892 |
{
|
| 1893 |
"epoch": 0.655,
|
| 1894 |
-
"grad_norm": 3.
|
| 1895 |
"learning_rate": 1.4391666666666665e-05,
|
| 1896 |
-
"loss": 0.
|
| 1897 |
"step": 6550
|
| 1898 |
},
|
| 1899 |
{
|
| 1900 |
"epoch": 0.6575,
|
| 1901 |
-
"grad_norm": 3.
|
| 1902 |
"learning_rate": 1.4287499999999998e-05,
|
| 1903 |
-
"loss": 0.
|
| 1904 |
"step": 6575
|
| 1905 |
},
|
| 1906 |
{
|
| 1907 |
"epoch": 0.66,
|
| 1908 |
-
"grad_norm": 3.
|
| 1909 |
"learning_rate": 1.4183333333333332e-05,
|
| 1910 |
-
"loss": 0.
|
| 1911 |
"step": 6600
|
| 1912 |
},
|
| 1913 |
{
|
| 1914 |
"epoch": 0.6625,
|
| 1915 |
-
"grad_norm":
|
| 1916 |
"learning_rate": 1.4079166666666666e-05,
|
| 1917 |
-
"loss": 0.
|
| 1918 |
"step": 6625
|
| 1919 |
},
|
| 1920 |
{
|
| 1921 |
"epoch": 0.665,
|
| 1922 |
-
"grad_norm": 3.
|
| 1923 |
"learning_rate": 1.3974999999999997e-05,
|
| 1924 |
-
"loss": 0.
|
| 1925 |
"step": 6650
|
| 1926 |
},
|
| 1927 |
{
|
| 1928 |
"epoch": 0.6675,
|
| 1929 |
-
"grad_norm": 3.
|
| 1930 |
"learning_rate": 1.3870833333333331e-05,
|
| 1931 |
-
"loss": 0.
|
| 1932 |
"step": 6675
|
| 1933 |
},
|
| 1934 |
{
|
| 1935 |
"epoch": 0.67,
|
| 1936 |
-
"grad_norm": 2.
|
| 1937 |
"learning_rate": 1.3766666666666665e-05,
|
| 1938 |
-
"loss": 0.
|
| 1939 |
"step": 6700
|
| 1940 |
},
|
| 1941 |
{
|
| 1942 |
"epoch": 0.6725,
|
| 1943 |
-
"grad_norm": 3.
|
| 1944 |
"learning_rate": 1.3662499999999998e-05,
|
| 1945 |
-
"loss": 0.
|
| 1946 |
"step": 6725
|
| 1947 |
},
|
| 1948 |
{
|
| 1949 |
"epoch": 0.675,
|
| 1950 |
-
"grad_norm":
|
| 1951 |
"learning_rate": 1.3558333333333334e-05,
|
| 1952 |
"loss": 0.105,
|
| 1953 |
"step": 6750
|
| 1954 |
},
|
| 1955 |
{
|
| 1956 |
"epoch": 0.6775,
|
| 1957 |
-
"grad_norm": 3.
|
| 1958 |
"learning_rate": 1.3454166666666664e-05,
|
| 1959 |
-
"loss": 0.
|
| 1960 |
"step": 6775
|
| 1961 |
},
|
| 1962 |
{
|
| 1963 |
"epoch": 0.68,
|
| 1964 |
-
"grad_norm": 4.
|
| 1965 |
"learning_rate": 1.3349999999999998e-05,
|
| 1966 |
-
"loss": 0.
|
| 1967 |
"step": 6800
|
| 1968 |
},
|
| 1969 |
{
|
| 1970 |
"epoch": 0.6825,
|
| 1971 |
-
"grad_norm": 6.
|
| 1972 |
"learning_rate": 1.3245833333333331e-05,
|
| 1973 |
-
"loss": 0.
|
| 1974 |
"step": 6825
|
| 1975 |
},
|
| 1976 |
{
|
| 1977 |
"epoch": 0.685,
|
| 1978 |
-
"grad_norm": 5.
|
| 1979 |
"learning_rate": 1.3141666666666665e-05,
|
| 1980 |
-
"loss": 0.
|
| 1981 |
"step": 6850
|
| 1982 |
},
|
| 1983 |
{
|
| 1984 |
"epoch": 0.6875,
|
| 1985 |
-
"grad_norm": 4.
|
| 1986 |
"learning_rate": 1.30375e-05,
|
| 1987 |
-
"loss": 0.
|
| 1988 |
"step": 6875
|
| 1989 |
},
|
| 1990 |
{
|
| 1991 |
"epoch": 0.69,
|
| 1992 |
-
"grad_norm": 4.
|
| 1993 |
"learning_rate": 1.2933333333333334e-05,
|
| 1994 |
-
"loss": 0.
|
| 1995 |
"step": 6900
|
| 1996 |
},
|
| 1997 |
{
|
| 1998 |
"epoch": 0.6925,
|
| 1999 |
-
"grad_norm": 3.
|
| 2000 |
"learning_rate": 1.2829166666666664e-05,
|
| 2001 |
-
"loss": 0.
|
| 2002 |
"step": 6925
|
| 2003 |
},
|
| 2004 |
{
|
| 2005 |
"epoch": 0.695,
|
| 2006 |
-
"grad_norm": 4.
|
| 2007 |
"learning_rate": 1.2724999999999998e-05,
|
| 2008 |
-
"loss": 0.
|
| 2009 |
"step": 6950
|
| 2010 |
},
|
| 2011 |
{
|
| 2012 |
"epoch": 0.6975,
|
| 2013 |
-
"grad_norm": 4.
|
| 2014 |
"learning_rate": 1.2620833333333333e-05,
|
| 2015 |
-
"loss": 0.
|
| 2016 |
"step": 6975
|
| 2017 |
},
|
| 2018 |
{
|
| 2019 |
"epoch": 0.7,
|
| 2020 |
-
"grad_norm": 4.
|
| 2021 |
"learning_rate": 1.2516666666666667e-05,
|
| 2022 |
-
"loss": 0.
|
| 2023 |
"step": 7000
|
| 2024 |
},
|
| 2025 |
{
|
| 2026 |
"epoch": 0.7,
|
| 2027 |
-
"eval_loss": 0.
|
| 2028 |
-
"eval_runtime":
|
| 2029 |
-
"eval_samples_per_second":
|
| 2030 |
-
"eval_steps_per_second": 2.
|
| 2031 |
-
"eval_wer": 16.
|
| 2032 |
"step": 7000
|
| 2033 |
},
|
| 2034 |
{
|
| 2035 |
"epoch": 0.7025,
|
| 2036 |
-
"grad_norm":
|
| 2037 |
"learning_rate": 1.2416666666666666e-05,
|
| 2038 |
-
"loss": 0.
|
| 2039 |
"step": 7025
|
| 2040 |
},
|
| 2041 |
{
|
| 2042 |
"epoch": 0.705,
|
| 2043 |
-
"grad_norm": 8.
|
| 2044 |
"learning_rate": 1.2312499999999998e-05,
|
| 2045 |
-
"loss": 0.
|
| 2046 |
"step": 7050
|
| 2047 |
},
|
| 2048 |
{
|
| 2049 |
"epoch": 0.7075,
|
| 2050 |
-
"grad_norm": 5.
|
| 2051 |
"learning_rate": 1.2208333333333331e-05,
|
| 2052 |
-
"loss": 0.
|
| 2053 |
"step": 7075
|
| 2054 |
},
|
| 2055 |
{
|
| 2056 |
"epoch": 0.71,
|
| 2057 |
-
"grad_norm": 4.
|
| 2058 |
"learning_rate": 1.2104166666666665e-05,
|
| 2059 |
-
"loss": 0.
|
| 2060 |
"step": 7100
|
| 2061 |
},
|
| 2062 |
{
|
| 2063 |
"epoch": 0.7125,
|
| 2064 |
-
"grad_norm": 4.
|
| 2065 |
"learning_rate": 1.1999999999999999e-05,
|
| 2066 |
-
"loss": 0.
|
| 2067 |
"step": 7125
|
| 2068 |
},
|
| 2069 |
{
|
| 2070 |
"epoch": 0.715,
|
| 2071 |
-
"grad_norm":
|
| 2072 |
"learning_rate": 1.1895833333333332e-05,
|
| 2073 |
-
"loss": 0.
|
| 2074 |
"step": 7150
|
| 2075 |
},
|
| 2076 |
{
|
| 2077 |
"epoch": 0.7175,
|
| 2078 |
-
"grad_norm":
|
| 2079 |
"learning_rate": 1.1791666666666666e-05,
|
| 2080 |
-
"loss": 0.
|
| 2081 |
"step": 7175
|
| 2082 |
},
|
| 2083 |
{
|
| 2084 |
"epoch": 0.72,
|
| 2085 |
-
"grad_norm": 3.
|
| 2086 |
"learning_rate": 1.1687499999999998e-05,
|
| 2087 |
-
"loss": 0.
|
| 2088 |
"step": 7200
|
| 2089 |
},
|
| 2090 |
{
|
| 2091 |
"epoch": 0.7225,
|
| 2092 |
-
"grad_norm": 2.
|
| 2093 |
"learning_rate": 1.1583333333333331e-05,
|
| 2094 |
-
"loss": 0.
|
| 2095 |
"step": 7225
|
| 2096 |
},
|
| 2097 |
{
|
| 2098 |
"epoch": 0.725,
|
| 2099 |
-
"grad_norm":
|
| 2100 |
"learning_rate": 1.1479166666666665e-05,
|
| 2101 |
-
"loss": 0.
|
| 2102 |
"step": 7250
|
| 2103 |
},
|
| 2104 |
{
|
| 2105 |
"epoch": 0.7275,
|
| 2106 |
-
"grad_norm": 2.
|
| 2107 |
"learning_rate": 1.1374999999999999e-05,
|
| 2108 |
-
"loss": 0.
|
| 2109 |
"step": 7275
|
| 2110 |
},
|
| 2111 |
{
|
| 2112 |
"epoch": 0.73,
|
| 2113 |
-
"grad_norm": 2.
|
| 2114 |
"learning_rate": 1.1270833333333332e-05,
|
| 2115 |
-
"loss": 0.
|
| 2116 |
"step": 7300
|
| 2117 |
},
|
| 2118 |
{
|
| 2119 |
"epoch": 0.7325,
|
| 2120 |
-
"grad_norm": 2.
|
| 2121 |
"learning_rate": 1.1166666666666664e-05,
|
| 2122 |
-
"loss": 0.
|
| 2123 |
"step": 7325
|
| 2124 |
},
|
| 2125 |
{
|
| 2126 |
"epoch": 0.735,
|
| 2127 |
-
"grad_norm": 4.
|
| 2128 |
"learning_rate": 1.1062499999999998e-05,
|
| 2129 |
-
"loss": 0.
|
| 2130 |
"step": 7350
|
| 2131 |
},
|
| 2132 |
{
|
| 2133 |
"epoch": 0.7375,
|
| 2134 |
-
"grad_norm": 3.
|
| 2135 |
"learning_rate": 1.0958333333333331e-05,
|
| 2136 |
-
"loss": 0.
|
| 2137 |
"step": 7375
|
| 2138 |
},
|
| 2139 |
{
|
| 2140 |
"epoch": 0.74,
|
| 2141 |
-
"grad_norm": 5.
|
| 2142 |
"learning_rate": 1.0854166666666665e-05,
|
| 2143 |
-
"loss": 0.
|
| 2144 |
"step": 7400
|
| 2145 |
},
|
| 2146 |
{
|
| 2147 |
"epoch": 0.7425,
|
| 2148 |
-
"grad_norm": 2.
|
| 2149 |
"learning_rate": 1.075e-05,
|
| 2150 |
-
"loss": 0.
|
| 2151 |
"step": 7425
|
| 2152 |
},
|
| 2153 |
{
|
| 2154 |
"epoch": 0.745,
|
| 2155 |
-
"grad_norm": 3.
|
| 2156 |
"learning_rate": 1.0645833333333334e-05,
|
| 2157 |
-
"loss": 0.
|
| 2158 |
"step": 7450
|
| 2159 |
},
|
| 2160 |
{
|
| 2161 |
"epoch": 0.7475,
|
| 2162 |
-
"grad_norm": 2.
|
| 2163 |
"learning_rate": 1.0541666666666664e-05,
|
| 2164 |
-
"loss": 0.
|
| 2165 |
"step": 7475
|
| 2166 |
},
|
| 2167 |
{
|
| 2168 |
"epoch": 0.75,
|
| 2169 |
-
"grad_norm":
|
| 2170 |
"learning_rate": 1.0437499999999998e-05,
|
| 2171 |
-
"loss": 0.
|
| 2172 |
"step": 7500
|
| 2173 |
},
|
| 2174 |
{
|
| 2175 |
"epoch": 0.7525,
|
| 2176 |
-
"grad_norm": 4.
|
| 2177 |
"learning_rate": 1.0333333333333332e-05,
|
| 2178 |
-
"loss": 0.
|
| 2179 |
"step": 7525
|
| 2180 |
},
|
| 2181 |
{
|
| 2182 |
"epoch": 0.755,
|
| 2183 |
-
"grad_norm": 5.
|
| 2184 |
"learning_rate": 1.0229166666666667e-05,
|
| 2185 |
-
"loss": 0.
|
| 2186 |
"step": 7550
|
| 2187 |
},
|
| 2188 |
{
|
| 2189 |
"epoch": 0.7575,
|
| 2190 |
-
"grad_norm": 5.
|
| 2191 |
"learning_rate": 1.0125e-05,
|
| 2192 |
-
"loss": 0.
|
| 2193 |
"step": 7575
|
| 2194 |
},
|
| 2195 |
{
|
| 2196 |
"epoch": 0.76,
|
| 2197 |
-
"grad_norm": 2.
|
| 2198 |
"learning_rate": 1.0020833333333334e-05,
|
| 2199 |
-
"loss": 0.
|
| 2200 |
"step": 7600
|
| 2201 |
},
|
| 2202 |
{
|
| 2203 |
"epoch": 0.7625,
|
| 2204 |
-
"grad_norm": 4.
|
| 2205 |
"learning_rate": 9.916666666666664e-06,
|
| 2206 |
-
"loss": 0.
|
| 2207 |
"step": 7625
|
| 2208 |
},
|
| 2209 |
{
|
| 2210 |
"epoch": 0.765,
|
| 2211 |
-
"grad_norm": 2.
|
| 2212 |
"learning_rate": 9.8125e-06,
|
| 2213 |
-
"loss": 0.
|
| 2214 |
"step": 7650
|
| 2215 |
},
|
| 2216 |
{
|
| 2217 |
"epoch": 0.7675,
|
| 2218 |
-
"grad_norm": 4.
|
| 2219 |
"learning_rate": 9.708333333333333e-06,
|
| 2220 |
-
"loss": 0.
|
| 2221 |
"step": 7675
|
| 2222 |
},
|
| 2223 |
{
|
| 2224 |
"epoch": 0.77,
|
| 2225 |
-
"grad_norm": 4.
|
| 2226 |
"learning_rate": 9.604166666666667e-06,
|
| 2227 |
-
"loss": 0.
|
| 2228 |
"step": 7700
|
| 2229 |
},
|
| 2230 |
{
|
| 2231 |
"epoch": 0.7725,
|
| 2232 |
-
"grad_norm":
|
| 2233 |
"learning_rate": 9.5e-06,
|
| 2234 |
-
"loss": 0.
|
| 2235 |
"step": 7725
|
| 2236 |
},
|
| 2237 |
{
|
| 2238 |
"epoch": 0.775,
|
| 2239 |
-
"grad_norm": 3.
|
| 2240 |
"learning_rate": 9.39583333333333e-06,
|
| 2241 |
"loss": 0.1671,
|
| 2242 |
"step": 7750
|
| 2243 |
},
|
| 2244 |
{
|
| 2245 |
"epoch": 0.7775,
|
| 2246 |
-
"grad_norm": 3.
|
| 2247 |
"learning_rate": 9.291666666666666e-06,
|
| 2248 |
-
"loss": 0.
|
| 2249 |
"step": 7775
|
| 2250 |
},
|
| 2251 |
{
|
| 2252 |
"epoch": 0.78,
|
| 2253 |
-
"grad_norm": 4.
|
| 2254 |
"learning_rate": 9.1875e-06,
|
| 2255 |
-
"loss": 0.
|
| 2256 |
"step": 7800
|
| 2257 |
},
|
| 2258 |
{
|
| 2259 |
"epoch": 0.7825,
|
| 2260 |
-
"grad_norm": 3.
|
| 2261 |
"learning_rate": 9.083333333333333e-06,
|
| 2262 |
-
"loss": 0.
|
| 2263 |
"step": 7825
|
| 2264 |
},
|
| 2265 |
{
|
| 2266 |
"epoch": 0.785,
|
| 2267 |
-
"grad_norm": 2.
|
| 2268 |
"learning_rate": 8.979166666666665e-06,
|
| 2269 |
-
"loss": 0.
|
| 2270 |
"step": 7850
|
| 2271 |
},
|
| 2272 |
{
|
| 2273 |
"epoch": 0.7875,
|
| 2274 |
-
"grad_norm": 2.
|
| 2275 |
"learning_rate": 8.874999999999999e-06,
|
| 2276 |
-
"loss": 0.
|
| 2277 |
"step": 7875
|
| 2278 |
},
|
| 2279 |
{
|
| 2280 |
"epoch": 0.79,
|
| 2281 |
-
"grad_norm":
|
| 2282 |
"learning_rate": 8.770833333333333e-06,
|
| 2283 |
-
"loss": 0.
|
| 2284 |
"step": 7900
|
| 2285 |
},
|
| 2286 |
{
|
| 2287 |
"epoch": 0.7925,
|
| 2288 |
-
"grad_norm": 2.
|
| 2289 |
"learning_rate": 8.666666666666666e-06,
|
| 2290 |
-
"loss": 0.
|
| 2291 |
"step": 7925
|
| 2292 |
},
|
| 2293 |
{
|
| 2294 |
"epoch": 0.795,
|
| 2295 |
-
"grad_norm":
|
| 2296 |
"learning_rate": 8.5625e-06,
|
| 2297 |
-
"loss": 0.
|
| 2298 |
"step": 7950
|
| 2299 |
},
|
| 2300 |
{
|
| 2301 |
"epoch": 0.7975,
|
| 2302 |
-
"grad_norm":
|
| 2303 |
"learning_rate": 8.458333333333333e-06,
|
| 2304 |
-
"loss": 0.
|
| 2305 |
"step": 7975
|
| 2306 |
},
|
| 2307 |
{
|
| 2308 |
"epoch": 0.8,
|
| 2309 |
-
"grad_norm": 3.
|
| 2310 |
"learning_rate": 8.354166666666665e-06,
|
| 2311 |
-
"loss": 0.
|
| 2312 |
"step": 8000
|
| 2313 |
},
|
| 2314 |
{
|
| 2315 |
"epoch": 0.8,
|
| 2316 |
-
"eval_loss": 0.
|
| 2317 |
-
"eval_runtime":
|
| 2318 |
-
"eval_samples_per_second":
|
| 2319 |
-
"eval_steps_per_second": 2.
|
| 2320 |
-
"eval_wer":
|
| 2321 |
"step": 8000
|
| 2322 |
},
|
| 2323 |
{
|
| 2324 |
"epoch": 0.8025,
|
| 2325 |
-
"grad_norm": 4.
|
| 2326 |
"learning_rate": 8.249999999999999e-06,
|
| 2327 |
-
"loss": 0.
|
| 2328 |
"step": 8025
|
| 2329 |
},
|
| 2330 |
{
|
| 2331 |
"epoch": 0.805,
|
| 2332 |
-
"grad_norm": 3.
|
| 2333 |
"learning_rate": 8.145833333333333e-06,
|
| 2334 |
-
"loss": 0.
|
| 2335 |
"step": 8050
|
| 2336 |
},
|
| 2337 |
{
|
| 2338 |
"epoch": 0.8075,
|
| 2339 |
-
"grad_norm":
|
| 2340 |
"learning_rate": 8.041666666666666e-06,
|
| 2341 |
-
"loss": 0.
|
| 2342 |
"step": 8075
|
| 2343 |
},
|
| 2344 |
{
|
| 2345 |
"epoch": 0.81,
|
| 2346 |
-
"grad_norm": 2.
|
| 2347 |
"learning_rate": 7.9375e-06,
|
| 2348 |
-
"loss": 0.
|
| 2349 |
"step": 8100
|
| 2350 |
},
|
| 2351 |
{
|
| 2352 |
"epoch": 0.8125,
|
| 2353 |
-
"grad_norm": 2.
|
| 2354 |
"learning_rate": 7.833333333333333e-06,
|
| 2355 |
-
"loss": 0.
|
| 2356 |
"step": 8125
|
| 2357 |
},
|
| 2358 |
{
|
| 2359 |
"epoch": 0.815,
|
| 2360 |
-
"grad_norm": 3.
|
| 2361 |
"learning_rate": 7.729166666666665e-06,
|
| 2362 |
-
"loss": 0.
|
| 2363 |
"step": 8150
|
| 2364 |
},
|
| 2365 |
{
|
| 2366 |
"epoch": 0.8175,
|
| 2367 |
-
"grad_norm": 2.
|
| 2368 |
"learning_rate": 7.625e-06,
|
| 2369 |
-
"loss": 0.
|
| 2370 |
"step": 8175
|
| 2371 |
},
|
| 2372 |
{
|
| 2373 |
"epoch": 0.82,
|
| 2374 |
-
"grad_norm": 3.
|
| 2375 |
"learning_rate": 7.520833333333332e-06,
|
| 2376 |
-
"loss": 0.
|
| 2377 |
"step": 8200
|
| 2378 |
},
|
| 2379 |
{
|
| 2380 |
"epoch": 0.8225,
|
| 2381 |
-
"grad_norm": 2.
|
| 2382 |
"learning_rate": 7.416666666666666e-06,
|
| 2383 |
-
"loss": 0.
|
| 2384 |
"step": 8225
|
| 2385 |
},
|
| 2386 |
{
|
| 2387 |
"epoch": 0.825,
|
| 2388 |
-
"grad_norm": 3.
|
| 2389 |
"learning_rate": 7.3125e-06,
|
| 2390 |
-
"loss": 0.
|
| 2391 |
"step": 8250
|
| 2392 |
},
|
| 2393 |
{
|
| 2394 |
"epoch": 0.8275,
|
| 2395 |
-
"grad_norm": 2.
|
| 2396 |
"learning_rate": 7.208333333333333e-06,
|
| 2397 |
-
"loss": 0.
|
| 2398 |
"step": 8275
|
| 2399 |
},
|
| 2400 |
{
|
| 2401 |
"epoch": 0.83,
|
| 2402 |
-
"grad_norm": 2.
|
| 2403 |
"learning_rate": 7.104166666666666e-06,
|
| 2404 |
-
"loss": 0.
|
| 2405 |
"step": 8300
|
| 2406 |
},
|
| 2407 |
{
|
| 2408 |
"epoch": 0.8325,
|
| 2409 |
-
"grad_norm": 3.
|
| 2410 |
"learning_rate": 7e-06,
|
| 2411 |
-
"loss": 0.
|
| 2412 |
"step": 8325
|
| 2413 |
},
|
| 2414 |
{
|
| 2415 |
"epoch": 0.835,
|
| 2416 |
-
"grad_norm": 4.
|
| 2417 |
"learning_rate": 6.895833333333333e-06,
|
| 2418 |
-
"loss": 0.
|
| 2419 |
"step": 8350
|
| 2420 |
},
|
| 2421 |
{
|
| 2422 |
"epoch": 0.8375,
|
| 2423 |
-
"grad_norm": 4.
|
| 2424 |
"learning_rate": 6.791666666666666e-06,
|
| 2425 |
-
"loss": 0.
|
| 2426 |
"step": 8375
|
| 2427 |
},
|
| 2428 |
{
|
| 2429 |
"epoch": 0.84,
|
| 2430 |
-
"grad_norm": 2.
|
| 2431 |
"learning_rate": 6.6875e-06,
|
| 2432 |
-
"loss": 0.
|
| 2433 |
"step": 8400
|
| 2434 |
},
|
| 2435 |
{
|
| 2436 |
"epoch": 0.8425,
|
| 2437 |
-
"grad_norm":
|
| 2438 |
"learning_rate": 6.583333333333333e-06,
|
| 2439 |
-
"loss": 0.
|
| 2440 |
"step": 8425
|
| 2441 |
},
|
| 2442 |
{
|
| 2443 |
"epoch": 0.845,
|
| 2444 |
-
"grad_norm": 2.
|
| 2445 |
"learning_rate": 6.479166666666666e-06,
|
| 2446 |
"loss": 0.1098,
|
| 2447 |
"step": 8450
|
| 2448 |
},
|
| 2449 |
{
|
| 2450 |
"epoch": 0.8475,
|
| 2451 |
-
"grad_norm": 3.
|
| 2452 |
"learning_rate": 6.375e-06,
|
| 2453 |
-
"loss": 0.
|
| 2454 |
"step": 8475
|
| 2455 |
},
|
| 2456 |
{
|
| 2457 |
"epoch": 0.85,
|
| 2458 |
-
"grad_norm": 3.
|
| 2459 |
"learning_rate": 6.270833333333333e-06,
|
| 2460 |
-
"loss": 0.
|
| 2461 |
"step": 8500
|
| 2462 |
},
|
| 2463 |
{
|
| 2464 |
"epoch": 0.8525,
|
| 2465 |
-
"grad_norm": 3.
|
| 2466 |
"learning_rate": 6.166666666666666e-06,
|
| 2467 |
-
"loss": 0.
|
| 2468 |
"step": 8525
|
| 2469 |
},
|
| 2470 |
{
|
| 2471 |
"epoch": 0.855,
|
| 2472 |
-
"grad_norm": 5.
|
| 2473 |
"learning_rate": 6.062499999999999e-06,
|
| 2474 |
-
"loss": 0.
|
| 2475 |
"step": 8550
|
| 2476 |
},
|
| 2477 |
{
|
| 2478 |
"epoch": 0.8575,
|
| 2479 |
-
"grad_norm": 5.
|
| 2480 |
"learning_rate": 5.958333333333333e-06,
|
| 2481 |
-
"loss": 0.
|
| 2482 |
"step": 8575
|
| 2483 |
},
|
| 2484 |
{
|
| 2485 |
"epoch": 0.86,
|
| 2486 |
-
"grad_norm": 7.
|
| 2487 |
"learning_rate": 5.854166666666666e-06,
|
| 2488 |
-
"loss": 0.
|
| 2489 |
"step": 8600
|
| 2490 |
},
|
| 2491 |
{
|
| 2492 |
"epoch": 0.8625,
|
| 2493 |
-
"grad_norm": 7.
|
| 2494 |
"learning_rate": 5.749999999999999e-06,
|
| 2495 |
-
"loss": 0.
|
| 2496 |
"step": 8625
|
| 2497 |
},
|
| 2498 |
{
|
| 2499 |
"epoch": 0.865,
|
| 2500 |
-
"grad_norm": 3.
|
| 2501 |
"learning_rate": 5.645833333333333e-06,
|
| 2502 |
-
"loss": 0.
|
| 2503 |
"step": 8650
|
| 2504 |
},
|
| 2505 |
{
|
| 2506 |
"epoch": 0.8675,
|
| 2507 |
-
"grad_norm": 4.
|
| 2508 |
"learning_rate": 5.541666666666666e-06,
|
| 2509 |
-
"loss": 0.
|
| 2510 |
"step": 8675
|
| 2511 |
},
|
| 2512 |
{
|
| 2513 |
"epoch": 0.87,
|
| 2514 |
-
"grad_norm": 5.
|
| 2515 |
"learning_rate": 5.437499999999999e-06,
|
| 2516 |
-
"loss": 0.
|
| 2517 |
"step": 8700
|
| 2518 |
},
|
| 2519 |
{
|
| 2520 |
"epoch": 0.8725,
|
| 2521 |
-
"grad_norm": 3.
|
| 2522 |
"learning_rate": 5.333333333333333e-06,
|
| 2523 |
-
"loss": 0.
|
| 2524 |
"step": 8725
|
| 2525 |
},
|
| 2526 |
{
|
| 2527 |
"epoch": 0.875,
|
| 2528 |
-
"grad_norm": 2.
|
| 2529 |
"learning_rate": 5.2291666666666664e-06,
|
| 2530 |
-
"loss": 0.
|
| 2531 |
"step": 8750
|
| 2532 |
},
|
| 2533 |
{
|
| 2534 |
"epoch": 0.8775,
|
| 2535 |
-
"grad_norm": 2.
|
| 2536 |
"learning_rate": 5.124999999999999e-06,
|
| 2537 |
-
"loss": 0.
|
| 2538 |
"step": 8775
|
| 2539 |
},
|
| 2540 |
{
|
| 2541 |
"epoch": 0.88,
|
| 2542 |
-
"grad_norm": 3.
|
| 2543 |
"learning_rate": 5.020833333333333e-06,
|
| 2544 |
-
"loss": 0.
|
| 2545 |
"step": 8800
|
| 2546 |
},
|
| 2547 |
{
|
| 2548 |
"epoch": 0.8825,
|
| 2549 |
-
"grad_norm":
|
| 2550 |
"learning_rate": 4.9166666666666665e-06,
|
| 2551 |
-
"loss": 0.
|
| 2552 |
"step": 8825
|
| 2553 |
},
|
| 2554 |
{
|
| 2555 |
"epoch": 0.885,
|
| 2556 |
-
"grad_norm": 3.
|
| 2557 |
"learning_rate": 4.812499999999999e-06,
|
| 2558 |
-
"loss": 0.
|
| 2559 |
"step": 8850
|
| 2560 |
},
|
| 2561 |
{
|
| 2562 |
"epoch": 0.8875,
|
| 2563 |
-
"grad_norm": 3.
|
| 2564 |
"learning_rate": 4.708333333333333e-06,
|
| 2565 |
-
"loss": 0.
|
| 2566 |
"step": 8875
|
| 2567 |
},
|
| 2568 |
{
|
| 2569 |
"epoch": 0.89,
|
| 2570 |
-
"grad_norm": 3.
|
| 2571 |
"learning_rate": 4.6041666666666665e-06,
|
| 2572 |
-
"loss": 0.
|
| 2573 |
"step": 8900
|
| 2574 |
},
|
| 2575 |
{
|
| 2576 |
"epoch": 0.8925,
|
| 2577 |
-
"grad_norm": 3.
|
| 2578 |
"learning_rate": 4.499999999999999e-06,
|
| 2579 |
-
"loss": 0.
|
| 2580 |
"step": 8925
|
| 2581 |
},
|
| 2582 |
{
|
| 2583 |
"epoch": 0.895,
|
| 2584 |
-
"grad_norm":
|
| 2585 |
"learning_rate": 4.395833333333333e-06,
|
| 2586 |
-
"loss": 0.
|
| 2587 |
"step": 8950
|
| 2588 |
},
|
| 2589 |
{
|
| 2590 |
"epoch": 0.8975,
|
| 2591 |
-
"grad_norm":
|
| 2592 |
"learning_rate": 4.2916666666666665e-06,
|
| 2593 |
-
"loss": 0.
|
| 2594 |
"step": 8975
|
| 2595 |
},
|
| 2596 |
{
|
| 2597 |
"epoch": 0.9,
|
| 2598 |
-
"grad_norm": 2.
|
| 2599 |
"learning_rate": 4.187499999999999e-06,
|
| 2600 |
-
"loss": 0.
|
| 2601 |
"step": 9000
|
| 2602 |
},
|
| 2603 |
{
|
| 2604 |
"epoch": 0.9,
|
| 2605 |
-
"eval_loss": 0.
|
| 2606 |
-
"eval_runtime":
|
| 2607 |
-
"eval_samples_per_second":
|
| 2608 |
-
"eval_steps_per_second": 2.
|
| 2609 |
-
"eval_wer": 15.
|
| 2610 |
"step": 9000
|
| 2611 |
},
|
| 2612 |
{
|
| 2613 |
"epoch": 0.9025,
|
| 2614 |
-
"grad_norm": 3.
|
| 2615 |
"learning_rate": 4.083333333333333e-06,
|
| 2616 |
"loss": 0.1113,
|
| 2617 |
"step": 9025
|
| 2618 |
},
|
| 2619 |
{
|
| 2620 |
"epoch": 0.905,
|
| 2621 |
-
"grad_norm": 3.
|
| 2622 |
"learning_rate": 3.9791666666666665e-06,
|
| 2623 |
-
"loss": 0.
|
| 2624 |
"step": 9050
|
| 2625 |
},
|
| 2626 |
{
|
| 2627 |
"epoch": 0.9075,
|
| 2628 |
-
"grad_norm": 4.
|
| 2629 |
"learning_rate": 3.874999999999999e-06,
|
| 2630 |
"loss": 0.1739,
|
| 2631 |
"step": 9075
|
| 2632 |
},
|
| 2633 |
{
|
| 2634 |
"epoch": 0.91,
|
| 2635 |
-
"grad_norm": 3.
|
| 2636 |
"learning_rate": 3.770833333333333e-06,
|
| 2637 |
-
"loss": 0.
|
| 2638 |
"step": 9100
|
| 2639 |
},
|
| 2640 |
{
|
| 2641 |
"epoch": 0.9125,
|
| 2642 |
-
"grad_norm": 3.
|
| 2643 |
"learning_rate": 3.6666666666666666e-06,
|
| 2644 |
-
"loss": 0.
|
| 2645 |
"step": 9125
|
| 2646 |
},
|
| 2647 |
{
|
| 2648 |
"epoch": 0.915,
|
| 2649 |
-
"grad_norm": 2.
|
| 2650 |
"learning_rate": 3.5624999999999998e-06,
|
| 2651 |
-
"loss": 0.
|
| 2652 |
"step": 9150
|
| 2653 |
},
|
| 2654 |
{
|
| 2655 |
"epoch": 0.9175,
|
| 2656 |
-
"grad_norm": 3.
|
| 2657 |
"learning_rate": 3.458333333333333e-06,
|
| 2658 |
-
"loss": 0.
|
| 2659 |
"step": 9175
|
| 2660 |
},
|
| 2661 |
{
|
| 2662 |
"epoch": 0.92,
|
| 2663 |
-
"grad_norm":
|
| 2664 |
"learning_rate": 3.354166666666666e-06,
|
| 2665 |
-
"loss": 0.
|
| 2666 |
"step": 9200
|
| 2667 |
},
|
| 2668 |
{
|
| 2669 |
"epoch": 0.9225,
|
| 2670 |
-
"grad_norm":
|
| 2671 |
"learning_rate": 3.25e-06,
|
| 2672 |
-
"loss": 0.
|
| 2673 |
"step": 9225
|
| 2674 |
},
|
| 2675 |
{
|
| 2676 |
"epoch": 0.925,
|
| 2677 |
-
"grad_norm": 4.
|
| 2678 |
"learning_rate": 3.145833333333333e-06,
|
| 2679 |
-
"loss": 0.
|
| 2680 |
"step": 9250
|
| 2681 |
},
|
| 2682 |
{
|
| 2683 |
"epoch": 0.9275,
|
| 2684 |
-
"grad_norm": 3.
|
| 2685 |
"learning_rate": 3.041666666666666e-06,
|
| 2686 |
-
"loss": 0.
|
| 2687 |
"step": 9275
|
| 2688 |
},
|
| 2689 |
{
|
| 2690 |
"epoch": 0.93,
|
| 2691 |
-
"grad_norm": 2.
|
| 2692 |
"learning_rate": 2.9375e-06,
|
| 2693 |
-
"loss": 0.
|
| 2694 |
"step": 9300
|
| 2695 |
},
|
| 2696 |
{
|
| 2697 |
"epoch": 0.9325,
|
| 2698 |
-
"grad_norm": 2.
|
| 2699 |
"learning_rate": 2.833333333333333e-06,
|
| 2700 |
-
"loss": 0.
|
| 2701 |
"step": 9325
|
| 2702 |
},
|
| 2703 |
{
|
| 2704 |
"epoch": 0.935,
|
| 2705 |
-
"grad_norm": 2.
|
| 2706 |
"learning_rate": 2.7291666666666662e-06,
|
| 2707 |
-
"loss": 0.
|
| 2708 |
"step": 9350
|
| 2709 |
},
|
| 2710 |
{
|
| 2711 |
"epoch": 0.9375,
|
| 2712 |
-
"grad_norm": 3.
|
| 2713 |
"learning_rate": 2.625e-06,
|
| 2714 |
-
"loss": 0.
|
| 2715 |
"step": 9375
|
| 2716 |
},
|
| 2717 |
{
|
| 2718 |
"epoch": 0.94,
|
| 2719 |
-
"grad_norm": 3.
|
| 2720 |
"learning_rate": 2.520833333333333e-06,
|
| 2721 |
-
"loss": 0.
|
| 2722 |
"step": 9400
|
| 2723 |
},
|
| 2724 |
{
|
| 2725 |
"epoch": 0.9425,
|
| 2726 |
-
"grad_norm": 3.
|
| 2727 |
"learning_rate": 2.4166666666666663e-06,
|
| 2728 |
"loss": 0.169,
|
| 2729 |
"step": 9425
|
| 2730 |
},
|
| 2731 |
{
|
| 2732 |
"epoch": 0.945,
|
| 2733 |
-
"grad_norm": 2.
|
| 2734 |
"learning_rate": 2.3125e-06,
|
| 2735 |
-
"loss": 0.
|
| 2736 |
"step": 9450
|
| 2737 |
},
|
| 2738 |
{
|
| 2739 |
"epoch": 0.9475,
|
| 2740 |
-
"grad_norm": 3.
|
| 2741 |
"learning_rate": 2.208333333333333e-06,
|
| 2742 |
-
"loss": 0.
|
| 2743 |
"step": 9475
|
| 2744 |
},
|
| 2745 |
{
|
| 2746 |
"epoch": 0.95,
|
| 2747 |
-
"grad_norm": 4.
|
| 2748 |
"learning_rate": 2.1041666666666667e-06,
|
| 2749 |
-
"loss": 0.
|
| 2750 |
"step": 9500
|
| 2751 |
},
|
| 2752 |
{
|
| 2753 |
"epoch": 0.9525,
|
| 2754 |
-
"grad_norm": 4.
|
| 2755 |
"learning_rate": 2e-06,
|
| 2756 |
-
"loss": 0.
|
| 2757 |
"step": 9525
|
| 2758 |
},
|
| 2759 |
{
|
| 2760 |
"epoch": 0.955,
|
| 2761 |
-
"grad_norm": 3.
|
| 2762 |
"learning_rate": 1.8958333333333331e-06,
|
| 2763 |
-
"loss": 0.
|
| 2764 |
"step": 9550
|
| 2765 |
},
|
| 2766 |
{
|
| 2767 |
"epoch": 0.9575,
|
| 2768 |
-
"grad_norm": 4.
|
| 2769 |
"learning_rate": 1.7916666666666665e-06,
|
| 2770 |
-
"loss": 0.
|
| 2771 |
"step": 9575
|
| 2772 |
},
|
| 2773 |
{
|
| 2774 |
"epoch": 0.96,
|
| 2775 |
-
"grad_norm": 3.
|
| 2776 |
"learning_rate": 1.6874999999999997e-06,
|
| 2777 |
-
"loss": 0.
|
| 2778 |
"step": 9600
|
| 2779 |
},
|
| 2780 |
{
|
| 2781 |
"epoch": 0.9625,
|
| 2782 |
-
"grad_norm":
|
| 2783 |
"learning_rate": 1.5833333333333331e-06,
|
| 2784 |
-
"loss": 0.
|
| 2785 |
"step": 9625
|
| 2786 |
},
|
| 2787 |
{
|
| 2788 |
"epoch": 0.965,
|
| 2789 |
-
"grad_norm": 2.
|
| 2790 |
"learning_rate": 1.4791666666666663e-06,
|
| 2791 |
-
"loss": 0.
|
| 2792 |
"step": 9650
|
| 2793 |
},
|
| 2794 |
{
|
| 2795 |
"epoch": 0.9675,
|
| 2796 |
-
"grad_norm": 2.
|
| 2797 |
"learning_rate": 1.375e-06,
|
| 2798 |
"loss": 0.0959,
|
| 2799 |
"step": 9675
|
| 2800 |
},
|
| 2801 |
{
|
| 2802 |
"epoch": 0.97,
|
| 2803 |
-
"grad_norm": 2.
|
| 2804 |
"learning_rate": 1.2708333333333334e-06,
|
| 2805 |
-
"loss": 0.
|
| 2806 |
"step": 9700
|
| 2807 |
},
|
| 2808 |
{
|
| 2809 |
"epoch": 0.9725,
|
| 2810 |
-
"grad_norm": 1.
|
| 2811 |
"learning_rate": 1.1666666666666666e-06,
|
| 2812 |
-
"loss": 0.
|
| 2813 |
"step": 9725
|
| 2814 |
},
|
| 2815 |
{
|
| 2816 |
"epoch": 0.975,
|
| 2817 |
-
"grad_norm": 2.
|
| 2818 |
"learning_rate": 1.0624999999999998e-06,
|
| 2819 |
-
"loss": 0.
|
| 2820 |
"step": 9750
|
| 2821 |
},
|
| 2822 |
{
|
| 2823 |
"epoch": 0.9775,
|
| 2824 |
-
"grad_norm":
|
| 2825 |
"learning_rate": 9.583333333333334e-07,
|
| 2826 |
-
"loss": 0.
|
| 2827 |
"step": 9775
|
| 2828 |
},
|
| 2829 |
{
|
| 2830 |
"epoch": 0.98,
|
| 2831 |
-
"grad_norm": 2.
|
| 2832 |
"learning_rate": 8.541666666666666e-07,
|
| 2833 |
-
"loss": 0.
|
| 2834 |
"step": 9800
|
| 2835 |
},
|
| 2836 |
{
|
| 2837 |
"epoch": 0.9825,
|
| 2838 |
-
"grad_norm": 3.
|
| 2839 |
"learning_rate": 7.499999999999999e-07,
|
| 2840 |
-
"loss": 0.
|
| 2841 |
"step": 9825
|
| 2842 |
},
|
| 2843 |
{
|
| 2844 |
"epoch": 0.985,
|
| 2845 |
-
"grad_norm": 3.
|
| 2846 |
"learning_rate": 6.458333333333332e-07,
|
| 2847 |
-
"loss": 0.
|
| 2848 |
"step": 9850
|
| 2849 |
},
|
| 2850 |
{
|
| 2851 |
"epoch": 0.9875,
|
| 2852 |
-
"grad_norm": 2.
|
| 2853 |
"learning_rate": 5.416666666666666e-07,
|
| 2854 |
-
"loss": 0.
|
| 2855 |
"step": 9875
|
| 2856 |
},
|
| 2857 |
{
|
| 2858 |
"epoch": 0.99,
|
| 2859 |
-
"grad_norm":
|
| 2860 |
"learning_rate": 4.375e-07,
|
| 2861 |
-
"loss": 0.
|
| 2862 |
"step": 9900
|
| 2863 |
},
|
| 2864 |
{
|
| 2865 |
"epoch": 0.9925,
|
| 2866 |
-
"grad_norm": 2.
|
| 2867 |
"learning_rate": 3.333333333333333e-07,
|
| 2868 |
-
"loss": 0.
|
| 2869 |
"step": 9925
|
| 2870 |
},
|
| 2871 |
{
|
| 2872 |
"epoch": 0.995,
|
| 2873 |
-
"grad_norm": 3.
|
| 2874 |
"learning_rate": 2.2916666666666666e-07,
|
| 2875 |
-
"loss": 0.
|
| 2876 |
"step": 9950
|
| 2877 |
},
|
| 2878 |
{
|
| 2879 |
"epoch": 0.9975,
|
| 2880 |
-
"grad_norm": 2.
|
| 2881 |
"learning_rate": 1.25e-07,
|
| 2882 |
-
"loss": 0.
|
| 2883 |
"step": 9975
|
| 2884 |
},
|
| 2885 |
{
|
| 2886 |
"epoch": 1.0,
|
| 2887 |
-
"grad_norm": 2.
|
| 2888 |
"learning_rate": 2.083333333333333e-08,
|
| 2889 |
-
"loss": 0.
|
| 2890 |
"step": 10000
|
| 2891 |
},
|
| 2892 |
{
|
| 2893 |
"epoch": 1.0,
|
| 2894 |
-
"eval_loss": 0.
|
| 2895 |
-
"eval_runtime":
|
| 2896 |
-
"eval_samples_per_second":
|
| 2897 |
-
"eval_steps_per_second": 2.
|
| 2898 |
-
"eval_wer":
|
| 2899 |
"step": 10000
|
| 2900 |
},
|
| 2901 |
{
|
| 2902 |
"epoch": 1.0,
|
| 2903 |
"step": 10000,
|
| 2904 |
"total_flos": 7.8780432384e+18,
|
| 2905 |
-
"train_loss": 0.
|
| 2906 |
-
"train_runtime":
|
| 2907 |
-
"train_samples_per_second":
|
| 2908 |
-
"train_steps_per_second": 2.
|
| 2909 |
}
|
| 2910 |
],
|
| 2911 |
"logging_steps": 25,
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 14.849506681653555,
|
| 3 |
+
"best_model_checkpoint": "./checkpoint-10000",
|
| 4 |
"epoch": 1.0,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
"global_step": 10000,
|
|
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0025,
|
| 13 |
+
"grad_norm": 32.17699432373047,
|
| 14 |
"learning_rate": 7.875e-07,
|
| 15 |
+
"loss": 3.5416,
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.005,
|
| 20 |
+
"grad_norm": 17.338768005371094,
|
| 21 |
"learning_rate": 1.7249999999999998e-06,
|
| 22 |
+
"loss": 2.8498,
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.0075,
|
| 27 |
+
"grad_norm": 13.783746719360352,
|
| 28 |
"learning_rate": 2.6624999999999995e-06,
|
| 29 |
"loss": 1.9843,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.01,
|
| 34 |
+
"grad_norm": 10.80038070678711,
|
| 35 |
"learning_rate": 3.6e-06,
|
| 36 |
+
"loss": 1.6113,
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.0125,
|
| 41 |
+
"grad_norm": 9.404985427856445,
|
| 42 |
"learning_rate": 4.537499999999999e-06,
|
| 43 |
"loss": 1.3704,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.015,
|
| 48 |
+
"grad_norm": 8.393142700195312,
|
| 49 |
"learning_rate": 5.474999999999999e-06,
|
| 50 |
+
"loss": 1.1666,
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.0175,
|
| 55 |
+
"grad_norm": 9.127829551696777,
|
| 56 |
"learning_rate": 6.4125e-06,
|
| 57 |
"loss": 1.0771,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.02,
|
| 62 |
+
"grad_norm": 9.412626266479492,
|
| 63 |
"learning_rate": 7.35e-06,
|
| 64 |
"loss": 1.0148,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.0225,
|
| 69 |
+
"grad_norm": 7.45149564743042,
|
| 70 |
"learning_rate": 8.2875e-06,
|
| 71 |
+
"loss": 0.8875,
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.025,
|
| 76 |
+
"grad_norm": 10.142014503479004,
|
| 77 |
"learning_rate": 9.224999999999999e-06,
|
| 78 |
+
"loss": 0.8914,
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.0275,
|
| 83 |
+
"grad_norm": 8.410262107849121,
|
| 84 |
"learning_rate": 1.01625e-05,
|
| 85 |
"loss": 0.8811,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.03,
|
| 90 |
+
"grad_norm": 9.67717170715332,
|
| 91 |
"learning_rate": 1.1099999999999999e-05,
|
| 92 |
"loss": 0.8374,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.0325,
|
| 97 |
+
"grad_norm": 7.850519180297852,
|
| 98 |
"learning_rate": 1.20375e-05,
|
| 99 |
"loss": 0.7848,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.035,
|
| 104 |
+
"grad_norm": 7.787046909332275,
|
| 105 |
"learning_rate": 1.2974999999999999e-05,
|
| 106 |
"loss": 0.7056,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.0375,
|
| 111 |
+
"grad_norm": 8.000933647155762,
|
| 112 |
"learning_rate": 1.39125e-05,
|
| 113 |
+
"loss": 0.6118,
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.04,
|
| 118 |
+
"grad_norm": 6.588298320770264,
|
| 119 |
"learning_rate": 1.485e-05,
|
| 120 |
+
"loss": 0.5839,
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"epoch": 0.0425,
|
| 125 |
+
"grad_norm": 7.084440231323242,
|
| 126 |
"learning_rate": 1.5787499999999997e-05,
|
| 127 |
+
"loss": 0.5352,
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.045,
|
| 132 |
+
"grad_norm": 6.439165115356445,
|
| 133 |
"learning_rate": 1.6725e-05,
|
| 134 |
"loss": 0.5099,
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.0475,
|
| 139 |
+
"grad_norm": 6.848193168640137,
|
| 140 |
"learning_rate": 1.76625e-05,
|
| 141 |
+
"loss": 0.4845,
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 0.05,
|
| 146 |
+
"grad_norm": 6.957454204559326,
|
| 147 |
"learning_rate": 1.8599999999999998e-05,
|
| 148 |
+
"loss": 0.4591,
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.0525,
|
| 153 |
+
"grad_norm": 7.173631191253662,
|
| 154 |
"learning_rate": 1.95375e-05,
|
| 155 |
+
"loss": 0.4265,
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.055,
|
| 160 |
+
"grad_norm": 5.627034664154053,
|
| 161 |
"learning_rate": 2.0475e-05,
|
| 162 |
"loss": 0.4271,
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.0575,
|
| 167 |
+
"grad_norm": 5.964080333709717,
|
| 168 |
"learning_rate": 2.1412499999999995e-05,
|
| 169 |
+
"loss": 0.4051,
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.06,
|
| 174 |
+
"grad_norm": 8.408676147460938,
|
| 175 |
"learning_rate": 2.2349999999999998e-05,
|
| 176 |
+
"loss": 0.3964,
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.0625,
|
| 181 |
+
"grad_norm": 6.321172714233398,
|
| 182 |
"learning_rate": 2.3287499999999997e-05,
|
| 183 |
+
"loss": 0.3565,
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.065,
|
| 188 |
+
"grad_norm": 5.115601062774658,
|
| 189 |
"learning_rate": 2.4225e-05,
|
| 190 |
+
"loss": 0.3746,
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 0.0675,
|
| 195 |
+
"grad_norm": 5.302433967590332,
|
| 196 |
"learning_rate": 2.51625e-05,
|
| 197 |
+
"loss": 0.3636,
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
"epoch": 0.07,
|
| 202 |
+
"grad_norm": 6.260375499725342,
|
| 203 |
"learning_rate": 2.6099999999999997e-05,
|
| 204 |
"loss": 0.3597,
|
| 205 |
"step": 700
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.0725,
|
| 209 |
+
"grad_norm": 8.463918685913086,
|
| 210 |
"learning_rate": 2.7037499999999997e-05,
|
| 211 |
+
"loss": 0.399,
|
| 212 |
"step": 725
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.075,
|
| 216 |
+
"grad_norm": 6.9618659019470215,
|
| 217 |
"learning_rate": 2.7975e-05,
|
| 218 |
+
"loss": 0.5304,
|
| 219 |
"step": 750
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.0775,
|
| 223 |
+
"grad_norm": 6.6647162437438965,
|
| 224 |
"learning_rate": 2.8912499999999998e-05,
|
| 225 |
+
"loss": 0.5135,
|
| 226 |
"step": 775
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.08,
|
| 230 |
+
"grad_norm": 7.29068660736084,
|
| 231 |
"learning_rate": 2.985e-05,
|
| 232 |
+
"loss": 0.5595,
|
| 233 |
"step": 800
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 0.0825,
|
| 237 |
+
"grad_norm": 5.9147186279296875,
|
| 238 |
"learning_rate": 3.0787499999999996e-05,
|
| 239 |
+
"loss": 0.3989,
|
| 240 |
"step": 825
|
| 241 |
},
|
| 242 |
{
|
| 243 |
"epoch": 0.085,
|
| 244 |
+
"grad_norm": 4.840363502502441,
|
| 245 |
"learning_rate": 3.1725e-05,
|
| 246 |
+
"loss": 0.3295,
|
| 247 |
"step": 850
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"epoch": 0.0875,
|
| 251 |
+
"grad_norm": 4.7903571128845215,
|
| 252 |
"learning_rate": 3.2662499999999994e-05,
|
| 253 |
+
"loss": 0.3158,
|
| 254 |
"step": 875
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.09,
|
| 258 |
+
"grad_norm": 6.355573654174805,
|
| 259 |
"learning_rate": 3.36e-05,
|
| 260 |
+
"loss": 0.4046,
|
| 261 |
"step": 900
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.0925,
|
| 265 |
+
"grad_norm": 7.575016975402832,
|
| 266 |
"learning_rate": 3.45375e-05,
|
| 267 |
"loss": 0.4779,
|
| 268 |
"step": 925
|
| 269 |
},
|
| 270 |
{
|
| 271 |
"epoch": 0.095,
|
| 272 |
+
"grad_norm": 6.783919334411621,
|
| 273 |
"learning_rate": 3.5474999999999995e-05,
|
| 274 |
+
"loss": 0.4578,
|
| 275 |
"step": 950
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 0.0975,
|
| 279 |
+
"grad_norm": 7.364860534667969,
|
| 280 |
"learning_rate": 3.64125e-05,
|
| 281 |
+
"loss": 0.6591,
|
| 282 |
"step": 975
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 0.1,
|
| 286 |
+
"grad_norm": 7.520545959472656,
|
| 287 |
"learning_rate": 3.735e-05,
|
| 288 |
+
"loss": 0.586,
|
| 289 |
"step": 1000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.1,
|
| 293 |
+
"eval_loss": 0.6248713731765747,
|
| 294 |
+
"eval_runtime": 39.9662,
|
| 295 |
+
"eval_samples_per_second": 42.311,
|
| 296 |
+
"eval_steps_per_second": 2.652,
|
| 297 |
+
"eval_wer": 34.16385662545272,
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.1025,
|
| 302 |
+
"grad_norm": 6.746083736419678,
|
| 303 |
"learning_rate": 3.74125e-05,
|
| 304 |
"loss": 0.5033,
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.105,
|
| 309 |
+
"grad_norm": 5.35268497467041,
|
| 310 |
"learning_rate": 3.730833333333333e-05,
|
| 311 |
+
"loss": 0.3954,
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 0.1075,
|
| 316 |
+
"grad_norm": 4.840243816375732,
|
| 317 |
"learning_rate": 3.7204166666666665e-05,
|
| 318 |
+
"loss": 0.3024,
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.11,
|
| 323 |
+
"grad_norm": 4.398126602172852,
|
| 324 |
"learning_rate": 3.7099999999999994e-05,
|
| 325 |
+
"loss": 0.2543,
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.1125,
|
| 330 |
+
"grad_norm": 4.504413604736328,
|
| 331 |
"learning_rate": 3.699583333333333e-05,
|
| 332 |
"loss": 0.2439,
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.115,
|
| 337 |
+
"grad_norm": 4.229017734527588,
|
| 338 |
"learning_rate": 3.6891666666666664e-05,
|
| 339 |
"loss": 0.2602,
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.1175,
|
| 344 |
+
"grad_norm": 4.006415367126465,
|
| 345 |
"learning_rate": 3.678749999999999e-05,
|
| 346 |
+
"loss": 0.2588,
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 0.12,
|
| 351 |
+
"grad_norm": 4.367706298828125,
|
| 352 |
"learning_rate": 3.668333333333333e-05,
|
| 353 |
+
"loss": 0.2419,
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.1225,
|
| 358 |
+
"grad_norm": 7.554067611694336,
|
| 359 |
"learning_rate": 3.6579166666666664e-05,
|
| 360 |
+
"loss": 0.3699,
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
"epoch": 0.125,
|
| 365 |
+
"grad_norm": 6.349033355712891,
|
| 366 |
"learning_rate": 3.6475e-05,
|
| 367 |
+
"loss": 0.3811,
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.1275,
|
| 372 |
+
"grad_norm": 6.685649394989014,
|
| 373 |
"learning_rate": 3.6370833333333334e-05,
|
| 374 |
+
"loss": 0.4109,
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.13,
|
| 379 |
+
"grad_norm": 8.145482063293457,
|
| 380 |
"learning_rate": 3.626666666666666e-05,
|
| 381 |
+
"loss": 0.4295,
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
"epoch": 0.1325,
|
| 386 |
+
"grad_norm": 6.097599983215332,
|
| 387 |
"learning_rate": 3.61625e-05,
|
| 388 |
+
"loss": 0.3691,
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.135,
|
| 393 |
+
"grad_norm": 6.148138999938965,
|
| 394 |
"learning_rate": 3.6058333333333333e-05,
|
| 395 |
+
"loss": 0.3754,
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 0.1375,
|
| 400 |
+
"grad_norm": 6.437056541442871,
|
| 401 |
"learning_rate": 3.595416666666666e-05,
|
| 402 |
"loss": 0.3569,
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
"epoch": 0.14,
|
| 407 |
+
"grad_norm": 6.0700201988220215,
|
| 408 |
"learning_rate": 3.585e-05,
|
| 409 |
+
"loss": 0.3443,
|
| 410 |
"step": 1400
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 0.1425,
|
| 414 |
+
"grad_norm": 6.010077476501465,
|
| 415 |
"learning_rate": 3.5745833333333326e-05,
|
| 416 |
+
"loss": 0.3516,
|
| 417 |
"step": 1425
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.145,
|
| 421 |
+
"grad_norm": 4.950198173522949,
|
| 422 |
"learning_rate": 3.564166666666666e-05,
|
| 423 |
+
"loss": 0.3309,
|
| 424 |
"step": 1450
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 0.1475,
|
| 428 |
+
"grad_norm": 4.597211837768555,
|
| 429 |
"learning_rate": 3.5537499999999996e-05,
|
| 430 |
+
"loss": 0.2661,
|
| 431 |
"step": 1475
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.15,
|
| 435 |
+
"grad_norm": 5.0821356773376465,
|
| 436 |
"learning_rate": 3.543333333333333e-05,
|
| 437 |
+
"loss": 0.2346,
|
| 438 |
"step": 1500
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 0.1525,
|
| 442 |
+
"grad_norm": 3.8110575675964355,
|
| 443 |
"learning_rate": 3.532916666666667e-05,
|
| 444 |
+
"loss": 0.2383,
|
| 445 |
"step": 1525
|
| 446 |
},
|
| 447 |
{
|
| 448 |
"epoch": 0.155,
|
| 449 |
+
"grad_norm": 5.058354377746582,
|
| 450 |
"learning_rate": 3.5224999999999996e-05,
|
| 451 |
+
"loss": 0.2311,
|
| 452 |
"step": 1550
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 0.1575,
|
| 456 |
+
"grad_norm": 4.210774898529053,
|
| 457 |
"learning_rate": 3.512083333333333e-05,
|
| 458 |
+
"loss": 0.205,
|
| 459 |
"step": 1575
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.16,
|
| 463 |
+
"grad_norm": 4.318747043609619,
|
| 464 |
"learning_rate": 3.5016666666666666e-05,
|
| 465 |
+
"loss": 0.215,
|
| 466 |
"step": 1600
|
| 467 |
},
|
| 468 |
{
|
| 469 |
"epoch": 0.1625,
|
| 470 |
+
"grad_norm": 5.172638893127441,
|
| 471 |
"learning_rate": 3.49125e-05,
|
| 472 |
+
"loss": 0.3064,
|
| 473 |
"step": 1625
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 0.165,
|
| 477 |
+
"grad_norm": 5.396555423736572,
|
| 478 |
"learning_rate": 3.480833333333333e-05,
|
| 479 |
+
"loss": 0.3197,
|
| 480 |
"step": 1650
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 0.1675,
|
| 484 |
+
"grad_norm": 7.734130382537842,
|
| 485 |
"learning_rate": 3.4704166666666665e-05,
|
| 486 |
+
"loss": 0.3303,
|
| 487 |
"step": 1675
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 0.17,
|
| 491 |
+
"grad_norm": 4.788872241973877,
|
| 492 |
"learning_rate": 3.4599999999999994e-05,
|
| 493 |
+
"loss": 0.3104,
|
| 494 |
"step": 1700
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 0.1725,
|
| 498 |
+
"grad_norm": 4.634150505065918,
|
| 499 |
"learning_rate": 3.449583333333333e-05,
|
| 500 |
+
"loss": 0.26,
|
| 501 |
"step": 1725
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.175,
|
| 505 |
+
"grad_norm": 4.166738510131836,
|
| 506 |
"learning_rate": 3.4391666666666665e-05,
|
| 507 |
+
"loss": 0.2411,
|
| 508 |
"step": 1750
|
| 509 |
},
|
| 510 |
{
|
| 511 |
"epoch": 0.1775,
|
| 512 |
+
"grad_norm": 4.349499702453613,
|
| 513 |
"learning_rate": 3.42875e-05,
|
| 514 |
+
"loss": 0.2139,
|
| 515 |
"step": 1775
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 0.18,
|
| 519 |
+
"grad_norm": 4.308148384094238,
|
| 520 |
"learning_rate": 3.418333333333333e-05,
|
| 521 |
+
"loss": 0.218,
|
| 522 |
"step": 1800
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 0.1825,
|
| 526 |
+
"grad_norm": 3.210516929626465,
|
| 527 |
"learning_rate": 3.4079166666666664e-05,
|
| 528 |
+
"loss": 0.2069,
|
| 529 |
"step": 1825
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 0.185,
|
| 533 |
+
"grad_norm": 3.348567247390747,
|
| 534 |
"learning_rate": 3.3975e-05,
|
| 535 |
+
"loss": 0.193,
|
| 536 |
"step": 1850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 0.1875,
|
| 540 |
+
"grad_norm": 4.138334274291992,
|
| 541 |
"learning_rate": 3.3870833333333334e-05,
|
| 542 |
+
"loss": 0.1939,
|
| 543 |
"step": 1875
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.19,
|
| 547 |
+
"grad_norm": 4.503148555755615,
|
| 548 |
"learning_rate": 3.376666666666666e-05,
|
| 549 |
+
"loss": 0.1921,
|
| 550 |
"step": 1900
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 0.1925,
|
| 554 |
+
"grad_norm": 3.379345417022705,
|
| 555 |
"learning_rate": 3.36625e-05,
|
| 556 |
+
"loss": 0.1829,
|
| 557 |
"step": 1925
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 0.195,
|
| 561 |
+
"grad_norm": 4.82003927230835,
|
| 562 |
"learning_rate": 3.355833333333333e-05,
|
| 563 |
+
"loss": 0.2564,
|
| 564 |
"step": 1950
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.1975,
|
| 568 |
+
"grad_norm": 5.72125768661499,
|
| 569 |
"learning_rate": 3.345416666666666e-05,
|
| 570 |
+
"loss": 0.3326,
|
| 571 |
"step": 1975
|
| 572 |
},
|
| 573 |
{
|
| 574 |
"epoch": 0.2,
|
| 575 |
+
"grad_norm": 5.574431896209717,
|
| 576 |
"learning_rate": 3.335e-05,
|
| 577 |
+
"loss": 0.3145,
|
| 578 |
"step": 2000
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 0.2,
|
| 582 |
+
"eval_loss": 0.5048365592956543,
|
| 583 |
+
"eval_runtime": 37.7406,
|
| 584 |
+
"eval_samples_per_second": 44.806,
|
| 585 |
+
"eval_steps_per_second": 2.809,
|
| 586 |
+
"eval_wer": 25.259148245285374,
|
| 587 |
"step": 2000
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 0.2025,
|
| 591 |
+
"grad_norm": 4.822042942047119,
|
| 592 |
"learning_rate": 3.324583333333333e-05,
|
| 593 |
+
"loss": 0.2662,
|
| 594 |
"step": 2025
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.205,
|
| 598 |
+
"grad_norm": 4.082330226898193,
|
| 599 |
"learning_rate": 3.314166666666666e-05,
|
| 600 |
+
"loss": 0.1935,
|
| 601 |
"step": 2050
|
| 602 |
},
|
| 603 |
{
|
| 604 |
"epoch": 0.2075,
|
| 605 |
+
"grad_norm": 5.841582775115967,
|
| 606 |
"learning_rate": 3.30375e-05,
|
| 607 |
+
"loss": 0.186,
|
| 608 |
"step": 2075
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 0.21,
|
| 612 |
+
"grad_norm": 3.377840042114258,
|
| 613 |
"learning_rate": 3.293333333333333e-05,
|
| 614 |
+
"loss": 0.1834,
|
| 615 |
"step": 2100
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.2125,
|
| 619 |
+
"grad_norm": 3.8289568424224854,
|
| 620 |
"learning_rate": 3.282916666666667e-05,
|
| 621 |
+
"loss": 0.1594,
|
| 622 |
"step": 2125
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 0.215,
|
| 626 |
+
"grad_norm": 3.3037023544311523,
|
| 627 |
"learning_rate": 3.2724999999999996e-05,
|
| 628 |
+
"loss": 0.1623,
|
| 629 |
"step": 2150
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 0.2175,
|
| 633 |
+
"grad_norm": 4.1908860206604,
|
| 634 |
"learning_rate": 3.262083333333333e-05,
|
| 635 |
+
"loss": 0.1674,
|
| 636 |
"step": 2175
|
| 637 |
},
|
| 638 |
{
|
| 639 |
"epoch": 0.22,
|
| 640 |
+
"grad_norm": 6.681819438934326,
|
| 641 |
"learning_rate": 3.2516666666666666e-05,
|
| 642 |
+
"loss": 0.2856,
|
| 643 |
"step": 2200
|
| 644 |
},
|
| 645 |
{
|
| 646 |
"epoch": 0.2225,
|
| 647 |
+
"grad_norm": 5.509127616882324,
|
| 648 |
"learning_rate": 3.2412499999999995e-05,
|
| 649 |
+
"loss": 0.2918,
|
| 650 |
"step": 2225
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 0.225,
|
| 654 |
+
"grad_norm": 4.549745559692383,
|
| 655 |
"learning_rate": 3.230833333333333e-05,
|
| 656 |
+
"loss": 0.3008,
|
| 657 |
"step": 2250
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 0.2275,
|
| 661 |
+
"grad_norm": 4.073692798614502,
|
| 662 |
"learning_rate": 3.2204166666666666e-05,
|
| 663 |
+
"loss": 0.2116,
|
| 664 |
"step": 2275
|
| 665 |
},
|
| 666 |
{
|
| 667 |
"epoch": 0.23,
|
| 668 |
+
"grad_norm": 3.217360734939575,
|
| 669 |
"learning_rate": 3.2099999999999994e-05,
|
| 670 |
+
"loss": 0.1622,
|
| 671 |
"step": 2300
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 0.2325,
|
| 675 |
+
"grad_norm": 4.068457126617432,
|
| 676 |
"learning_rate": 3.199583333333333e-05,
|
| 677 |
+
"loss": 0.1538,
|
| 678 |
"step": 2325
|
| 679 |
},
|
| 680 |
{
|
| 681 |
"epoch": 0.235,
|
| 682 |
+
"grad_norm": 6.21290922164917,
|
| 683 |
"learning_rate": 3.1891666666666665e-05,
|
| 684 |
+
"loss": 0.1543,
|
| 685 |
"step": 2350
|
| 686 |
},
|
| 687 |
{
|
| 688 |
"epoch": 0.2375,
|
| 689 |
+
"grad_norm": 2.9277803897857666,
|
| 690 |
"learning_rate": 3.17875e-05,
|
| 691 |
+
"loss": 0.158,
|
| 692 |
"step": 2375
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 0.24,
|
| 696 |
+
"grad_norm": 3.4852840900421143,
|
| 697 |
"learning_rate": 3.168333333333333e-05,
|
| 698 |
+
"loss": 0.1626,
|
| 699 |
"step": 2400
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.2425,
|
| 703 |
+
"grad_norm": 4.382246971130371,
|
| 704 |
"learning_rate": 3.1579166666666664e-05,
|
| 705 |
+
"loss": 0.1642,
|
| 706 |
"step": 2425
|
| 707 |
},
|
| 708 |
{
|
| 709 |
"epoch": 0.245,
|
| 710 |
+
"grad_norm": 2.6627678871154785,
|
| 711 |
"learning_rate": 3.1475e-05,
|
| 712 |
+
"loss": 0.1526,
|
| 713 |
"step": 2450
|
| 714 |
},
|
| 715 |
{
|
| 716 |
"epoch": 0.2475,
|
| 717 |
+
"grad_norm": 3.2728075981140137,
|
| 718 |
"learning_rate": 3.1370833333333335e-05,
|
| 719 |
+
"loss": 0.1624,
|
| 720 |
"step": 2475
|
| 721 |
},
|
| 722 |
{
|
| 723 |
"epoch": 0.25,
|
| 724 |
+
"grad_norm": 3.8267433643341064,
|
| 725 |
"learning_rate": 3.126666666666666e-05,
|
| 726 |
+
"loss": 0.1857,
|
| 727 |
"step": 2500
|
| 728 |
},
|
| 729 |
{
|
| 730 |
"epoch": 0.2525,
|
| 731 |
+
"grad_norm": 9.670909881591797,
|
| 732 |
"learning_rate": 3.11625e-05,
|
| 733 |
+
"loss": 0.2492,
|
| 734 |
"step": 2525
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 0.255,
|
| 738 |
+
"grad_norm": 4.275064945220947,
|
| 739 |
"learning_rate": 3.105833333333333e-05,
|
| 740 |
+
"loss": 0.238,
|
| 741 |
"step": 2550
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 0.2575,
|
| 745 |
+
"grad_norm": 4.751267910003662,
|
| 746 |
"learning_rate": 3.095416666666666e-05,
|
| 747 |
+
"loss": 0.2788,
|
| 748 |
"step": 2575
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 0.26,
|
| 752 |
+
"grad_norm": 3.571152925491333,
|
| 753 |
"learning_rate": 3.085e-05,
|
| 754 |
+
"loss": 0.2056,
|
| 755 |
"step": 2600
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 0.2625,
|
| 759 |
+
"grad_norm": 4.068671226501465,
|
| 760 |
"learning_rate": 3.074583333333333e-05,
|
| 761 |
"loss": 0.1741,
|
| 762 |
"step": 2625
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 0.265,
|
| 766 |
+
"grad_norm": 3.274651527404785,
|
| 767 |
"learning_rate": 3.064166666666666e-05,
|
| 768 |
+
"loss": 0.1571,
|
| 769 |
"step": 2650
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 0.2675,
|
| 773 |
+
"grad_norm": 3.0895748138427734,
|
| 774 |
"learning_rate": 3.05375e-05,
|
| 775 |
+
"loss": 0.149,
|
| 776 |
"step": 2675
|
| 777 |
},
|
| 778 |
{
|
| 779 |
"epoch": 0.27,
|
| 780 |
+
"grad_norm": 2.9620044231414795,
|
| 781 |
"learning_rate": 3.0433333333333332e-05,
|
| 782 |
"loss": 0.1584,
|
| 783 |
"step": 2700
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 0.2725,
|
| 787 |
+
"grad_norm": 3.2053639888763428,
|
| 788 |
"learning_rate": 3.0329166666666664e-05,
|
| 789 |
+
"loss": 0.1676,
|
| 790 |
"step": 2725
|
| 791 |
},
|
| 792 |
{
|
| 793 |
"epoch": 0.275,
|
| 794 |
+
"grad_norm": 4.407464981079102,
|
| 795 |
"learning_rate": 3.0225e-05,
|
| 796 |
+
"loss": 0.1799,
|
| 797 |
"step": 2750
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 0.2775,
|
| 801 |
+
"grad_norm": 5.430395603179932,
|
| 802 |
"learning_rate": 3.0120833333333328e-05,
|
| 803 |
+
"loss": 0.3165,
|
| 804 |
"step": 2775
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 0.28,
|
| 808 |
+
"grad_norm": 4.443249702453613,
|
| 809 |
"learning_rate": 3.0016666666666663e-05,
|
| 810 |
+
"loss": 0.2934,
|
| 811 |
"step": 2800
|
| 812 |
},
|
| 813 |
{
|
| 814 |
"epoch": 0.2825,
|
| 815 |
+
"grad_norm": 4.94956111907959,
|
| 816 |
"learning_rate": 2.9912499999999995e-05,
|
| 817 |
+
"loss": 0.2587,
|
| 818 |
"step": 2825
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 0.285,
|
| 822 |
+
"grad_norm": 4.102962493896484,
|
| 823 |
"learning_rate": 2.980833333333333e-05,
|
| 824 |
+
"loss": 0.2637,
|
| 825 |
"step": 2850
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 0.2875,
|
| 829 |
+
"grad_norm": 3.9422686100006104,
|
| 830 |
"learning_rate": 2.9704166666666662e-05,
|
| 831 |
"loss": 0.2561,
|
| 832 |
"step": 2875
|
| 833 |
},
|
| 834 |
{
|
| 835 |
"epoch": 0.29,
|
| 836 |
+
"grad_norm": 4.891707897186279,
|
| 837 |
"learning_rate": 2.9599999999999998e-05,
|
| 838 |
+
"loss": 0.2476,
|
| 839 |
"step": 2900
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 0.2925,
|
| 843 |
+
"grad_norm": 3.3687989711761475,
|
| 844 |
"learning_rate": 2.949583333333333e-05,
|
| 845 |
+
"loss": 0.2015,
|
| 846 |
"step": 2925
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 0.295,
|
| 850 |
+
"grad_norm": 3.280766010284424,
|
| 851 |
"learning_rate": 2.9391666666666665e-05,
|
| 852 |
+
"loss": 0.1618,
|
| 853 |
"step": 2950
|
| 854 |
},
|
| 855 |
{
|
| 856 |
"epoch": 0.2975,
|
| 857 |
+
"grad_norm": 3.6520583629608154,
|
| 858 |
"learning_rate": 2.9287499999999997e-05,
|
| 859 |
+
"loss": 0.1685,
|
| 860 |
"step": 2975
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 0.3,
|
| 864 |
+
"grad_norm": 4.03117561340332,
|
| 865 |
"learning_rate": 2.9183333333333332e-05,
|
| 866 |
+
"loss": 0.225,
|
| 867 |
"step": 3000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 0.3,
|
| 871 |
+
"eval_loss": 0.4838997423648834,
|
| 872 |
+
"eval_runtime": 38.3108,
|
| 873 |
+
"eval_samples_per_second": 44.139,
|
| 874 |
+
"eval_steps_per_second": 2.767,
|
| 875 |
+
"eval_wer": 22.05570126139628,
|
| 876 |
"step": 3000
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 0.3025,
|
| 880 |
+
"grad_norm": 5.466115951538086,
|
| 881 |
"learning_rate": 2.9079166666666664e-05,
|
| 882 |
+
"loss": 0.2752,
|
| 883 |
"step": 3025
|
| 884 |
},
|
| 885 |
{
|
| 886 |
"epoch": 0.305,
|
| 887 |
+
"grad_norm": 6.833763599395752,
|
| 888 |
"learning_rate": 2.8974999999999996e-05,
|
| 889 |
+
"loss": 0.3203,
|
| 890 |
"step": 3050
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 0.3075,
|
| 894 |
+
"grad_norm": 3.3256447315216064,
|
| 895 |
"learning_rate": 2.8870833333333328e-05,
|
| 896 |
+
"loss": 0.2206,
|
| 897 |
"step": 3075
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 0.31,
|
| 901 |
+
"grad_norm": 3.841505289077759,
|
| 902 |
"learning_rate": 2.8766666666666663e-05,
|
| 903 |
+
"loss": 0.1708,
|
| 904 |
"step": 3100
|
| 905 |
},
|
| 906 |
{
|
| 907 |
"epoch": 0.3125,
|
| 908 |
+
"grad_norm": 3.2223970890045166,
|
| 909 |
"learning_rate": 2.8662499999999995e-05,
|
| 910 |
+
"loss": 0.1507,
|
| 911 |
"step": 3125
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 0.315,
|
| 915 |
+
"grad_norm": 4.438643932342529,
|
| 916 |
"learning_rate": 2.855833333333333e-05,
|
| 917 |
+
"loss": 0.1827,
|
| 918 |
"step": 3150
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 0.3175,
|
| 922 |
+
"grad_norm": 4.88558292388916,
|
| 923 |
"learning_rate": 2.8454166666666663e-05,
|
| 924 |
+
"loss": 0.2668,
|
| 925 |
"step": 3175
|
| 926 |
},
|
| 927 |
{
|
| 928 |
"epoch": 0.32,
|
| 929 |
+
"grad_norm": 5.326033115386963,
|
| 930 |
"learning_rate": 2.8349999999999998e-05,
|
| 931 |
+
"loss": 0.247,
|
| 932 |
"step": 3200
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 0.3225,
|
| 936 |
+
"grad_norm": 4.686404228210449,
|
| 937 |
"learning_rate": 2.824583333333333e-05,
|
| 938 |
+
"loss": 0.2666,
|
| 939 |
"step": 3225
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 0.325,
|
| 943 |
+
"grad_norm": 5.8094401359558105,
|
| 944 |
"learning_rate": 2.8141666666666665e-05,
|
| 945 |
+
"loss": 0.2178,
|
| 946 |
"step": 3250
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 0.3275,
|
| 950 |
+
"grad_norm": 5.154630661010742,
|
| 951 |
"learning_rate": 2.80375e-05,
|
| 952 |
+
"loss": 0.2241,
|
| 953 |
"step": 3275
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 0.33,
|
| 957 |
+
"grad_norm": 3.6322267055511475,
|
| 958 |
"learning_rate": 2.7933333333333332e-05,
|
| 959 |
+
"loss": 0.2393,
|
| 960 |
"step": 3300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 0.3325,
|
| 964 |
+
"grad_norm": 3.901859998703003,
|
| 965 |
"learning_rate": 2.7829166666666668e-05,
|
| 966 |
+
"loss": 0.188,
|
| 967 |
"step": 3325
|
| 968 |
},
|
| 969 |
{
|
| 970 |
"epoch": 0.335,
|
| 971 |
+
"grad_norm": 3.8515241146087646,
|
| 972 |
"learning_rate": 2.7724999999999996e-05,
|
| 973 |
+
"loss": 0.1522,
|
| 974 |
"step": 3350
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 0.3375,
|
| 978 |
+
"grad_norm": 3.150157928466797,
|
| 979 |
"learning_rate": 2.7620833333333328e-05,
|
| 980 |
+
"loss": 0.1356,
|
| 981 |
"step": 3375
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 0.34,
|
| 985 |
+
"grad_norm": 3.471564769744873,
|
| 986 |
"learning_rate": 2.7516666666666664e-05,
|
| 987 |
+
"loss": 0.1435,
|
| 988 |
"step": 3400
|
| 989 |
},
|
| 990 |
{
|
| 991 |
"epoch": 0.3425,
|
| 992 |
+
"grad_norm": 3.938312292098999,
|
| 993 |
"learning_rate": 2.7412499999999995e-05,
|
| 994 |
+
"loss": 0.1445,
|
| 995 |
"step": 3425
|
| 996 |
},
|
| 997 |
{
|
| 998 |
"epoch": 0.345,
|
| 999 |
+
"grad_norm": 2.5613808631896973,
|
| 1000 |
"learning_rate": 2.730833333333333e-05,
|
| 1001 |
+
"loss": 0.1363,
|
| 1002 |
"step": 3450
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 0.3475,
|
| 1006 |
+
"grad_norm": 3.3892834186553955,
|
| 1007 |
"learning_rate": 2.7204166666666663e-05,
|
| 1008 |
+
"loss": 0.1322,
|
| 1009 |
"step": 3475
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 0.35,
|
| 1013 |
+
"grad_norm": 4.404175281524658,
|
| 1014 |
"learning_rate": 2.7099999999999998e-05,
|
| 1015 |
+
"loss": 0.1602,
|
| 1016 |
"step": 3500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 0.3525,
|
| 1020 |
+
"grad_norm": 3.336073398590088,
|
| 1021 |
"learning_rate": 2.6995833333333333e-05,
|
| 1022 |
+
"loss": 0.1488,
|
| 1023 |
"step": 3525
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 0.355,
|
| 1027 |
+
"grad_norm": 5.919867992401123,
|
| 1028 |
"learning_rate": 2.6891666666666665e-05,
|
| 1029 |
+
"loss": 0.1476,
|
| 1030 |
"step": 3550
|
| 1031 |
},
|
| 1032 |
{
|
| 1033 |
"epoch": 0.3575,
|
| 1034 |
+
"grad_norm": 5.114974021911621,
|
| 1035 |
"learning_rate": 2.67875e-05,
|
| 1036 |
+
"loss": 0.2274,
|
| 1037 |
"step": 3575
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 0.36,
|
| 1041 |
+
"grad_norm": 4.443377494812012,
|
| 1042 |
"learning_rate": 2.6683333333333333e-05,
|
| 1043 |
+
"loss": 0.2627,
|
| 1044 |
"step": 3600
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 0.3625,
|
| 1048 |
+
"grad_norm": 4.095592021942139,
|
| 1049 |
"learning_rate": 2.657916666666666e-05,
|
| 1050 |
+
"loss": 0.2451,
|
| 1051 |
"step": 3625
|
| 1052 |
},
|
| 1053 |
{
|
| 1054 |
"epoch": 0.365,
|
| 1055 |
+
"grad_norm": 2.603665828704834,
|
| 1056 |
"learning_rate": 2.6474999999999996e-05,
|
| 1057 |
+
"loss": 0.1531,
|
| 1058 |
"step": 3650
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 0.3675,
|
| 1062 |
+
"grad_norm": 3.089383125305176,
|
| 1063 |
"learning_rate": 2.637083333333333e-05,
|
| 1064 |
+
"loss": 0.1334,
|
| 1065 |
"step": 3675
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 0.37,
|
| 1069 |
+
"grad_norm": 3.058241605758667,
|
| 1070 |
"learning_rate": 2.6266666666666664e-05,
|
| 1071 |
+
"loss": 0.1322,
|
| 1072 |
"step": 3700
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 0.3725,
|
| 1076 |
+
"grad_norm": 3.3943984508514404,
|
| 1077 |
"learning_rate": 2.61625e-05,
|
| 1078 |
+
"loss": 0.159,
|
| 1079 |
"step": 3725
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 0.375,
|
| 1083 |
+
"grad_norm": 4.98250150680542,
|
| 1084 |
"learning_rate": 2.605833333333333e-05,
|
| 1085 |
+
"loss": 0.2044,
|
| 1086 |
"step": 3750
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 0.3775,
|
| 1090 |
+
"grad_norm": 3.9143283367156982,
|
| 1091 |
"learning_rate": 2.5954166666666666e-05,
|
| 1092 |
+
"loss": 0.2495,
|
| 1093 |
"step": 3775
|
| 1094 |
},
|
| 1095 |
{
|
| 1096 |
"epoch": 0.38,
|
| 1097 |
+
"grad_norm": 5.307981014251709,
|
| 1098 |
"learning_rate": 2.5849999999999998e-05,
|
| 1099 |
+
"loss": 0.2502,
|
| 1100 |
"step": 3800
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 0.3825,
|
| 1104 |
+
"grad_norm": 4.974552631378174,
|
| 1105 |
"learning_rate": 2.5745833333333333e-05,
|
| 1106 |
+
"loss": 0.2502,
|
| 1107 |
"step": 3825
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 0.385,
|
| 1111 |
+
"grad_norm": 5.2868547439575195,
|
| 1112 |
"learning_rate": 2.5641666666666665e-05,
|
| 1113 |
+
"loss": 0.2662,
|
| 1114 |
"step": 3850
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
"epoch": 0.3875,
|
| 1118 |
+
"grad_norm": 4.530261039733887,
|
| 1119 |
"learning_rate": 2.55375e-05,
|
| 1120 |
+
"loss": 0.2359,
|
| 1121 |
"step": 3875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 0.39,
|
| 1125 |
+
"grad_norm": 3.932204484939575,
|
| 1126 |
"learning_rate": 2.543333333333333e-05,
|
| 1127 |
+
"loss": 0.1757,
|
| 1128 |
"step": 3900
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 0.3925,
|
| 1132 |
+
"grad_norm": 3.0662739276885986,
|
| 1133 |
"learning_rate": 2.5329166666666665e-05,
|
| 1134 |
+
"loss": 0.148,
|
| 1135 |
"step": 3925
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
"epoch": 0.395,
|
| 1139 |
+
"grad_norm": 3.4663736820220947,
|
| 1140 |
"learning_rate": 2.5224999999999997e-05,
|
| 1141 |
+
"loss": 0.1411,
|
| 1142 |
"step": 3950
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 0.3975,
|
| 1146 |
+
"grad_norm": 6.448814868927002,
|
| 1147 |
"learning_rate": 2.5120833333333332e-05,
|
| 1148 |
+
"loss": 0.2409,
|
| 1149 |
"step": 3975
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 0.4,
|
| 1153 |
+
"grad_norm": 4.826459884643555,
|
| 1154 |
"learning_rate": 2.5016666666666664e-05,
|
| 1155 |
+
"loss": 0.3003,
|
| 1156 |
"step": 4000
|
| 1157 |
},
|
| 1158 |
{
|
| 1159 |
"epoch": 0.4,
|
| 1160 |
+
"eval_loss": 0.45400944352149963,
|
| 1161 |
+
"eval_runtime": 38.2487,
|
| 1162 |
+
"eval_samples_per_second": 44.211,
|
| 1163 |
+
"eval_steps_per_second": 2.771,
|
| 1164 |
+
"eval_wer": 20.307231172723867,
|
| 1165 |
"step": 4000
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
"epoch": 0.4025,
|
| 1169 |
+
"grad_norm": 3.8398826122283936,
|
| 1170 |
"learning_rate": 2.49125e-05,
|
| 1171 |
+
"loss": 0.2244,
|
| 1172 |
"step": 4025
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 0.405,
|
| 1176 |
+
"grad_norm": 5.592238426208496,
|
| 1177 |
"learning_rate": 2.480833333333333e-05,
|
| 1178 |
+
"loss": 0.3029,
|
| 1179 |
"step": 4050
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 0.4075,
|
| 1183 |
+
"grad_norm": 4.635577201843262,
|
| 1184 |
"learning_rate": 2.4704166666666666e-05,
|
| 1185 |
+
"loss": 0.2194,
|
| 1186 |
"step": 4075
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 0.41,
|
| 1190 |
+
"grad_norm": 3.2737910747528076,
|
| 1191 |
"learning_rate": 2.4599999999999998e-05,
|
| 1192 |
+
"loss": 0.2102,
|
| 1193 |
"step": 4100
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
"epoch": 0.4125,
|
| 1197 |
+
"grad_norm": 2.5521626472473145,
|
| 1198 |
"learning_rate": 2.4495833333333334e-05,
|
| 1199 |
+
"loss": 0.1517,
|
| 1200 |
"step": 4125
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 0.415,
|
| 1204 |
+
"grad_norm": 3.2408368587493896,
|
| 1205 |
"learning_rate": 2.4391666666666666e-05,
|
| 1206 |
+
"loss": 0.1331,
|
| 1207 |
"step": 4150
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
"epoch": 0.4175,
|
| 1211 |
+
"grad_norm": 3.243713855743408,
|
| 1212 |
"learning_rate": 2.4287499999999997e-05,
|
| 1213 |
+
"loss": 0.1385,
|
| 1214 |
"step": 4175
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 0.42,
|
| 1218 |
+
"grad_norm": 2.6619813442230225,
|
| 1219 |
"learning_rate": 2.418333333333333e-05,
|
| 1220 |
+
"loss": 0.1307,
|
| 1221 |
"step": 4200
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 0.4225,
|
| 1225 |
+
"grad_norm": 2.726177930831909,
|
| 1226 |
"learning_rate": 2.4079166666666665e-05,
|
| 1227 |
+
"loss": 0.1399,
|
| 1228 |
"step": 4225
|
| 1229 |
},
|
| 1230 |
{
|
| 1231 |
"epoch": 0.425,
|
| 1232 |
+
"grad_norm": 3.1518099308013916,
|
| 1233 |
"learning_rate": 2.3974999999999997e-05,
|
| 1234 |
+
"loss": 0.1344,
|
| 1235 |
"step": 4250
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 0.4275,
|
| 1239 |
+
"grad_norm": 3.95133376121521,
|
| 1240 |
"learning_rate": 2.3870833333333332e-05,
|
| 1241 |
+
"loss": 0.1512,
|
| 1242 |
"step": 4275
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 0.43,
|
| 1246 |
+
"grad_norm": 4.805422306060791,
|
| 1247 |
"learning_rate": 2.3766666666666664e-05,
|
| 1248 |
+
"loss": 0.2212,
|
| 1249 |
"step": 4300
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 0.4325,
|
| 1253 |
+
"grad_norm": 4.867978572845459,
|
| 1254 |
"learning_rate": 2.36625e-05,
|
| 1255 |
+
"loss": 0.2005,
|
| 1256 |
"step": 4325
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 0.435,
|
| 1260 |
+
"grad_norm": 3.448626756668091,
|
| 1261 |
"learning_rate": 2.355833333333333e-05,
|
| 1262 |
+
"loss": 0.2085,
|
| 1263 |
"step": 4350
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 0.4375,
|
| 1267 |
+
"grad_norm": 2.675185441970825,
|
| 1268 |
"learning_rate": 2.3454166666666666e-05,
|
| 1269 |
+
"loss": 0.1263,
|
| 1270 |
"step": 4375
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 0.44,
|
| 1274 |
+
"grad_norm": 3.467017412185669,
|
| 1275 |
"learning_rate": 2.335e-05,
|
| 1276 |
+
"loss": 0.1122,
|
| 1277 |
"step": 4400
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 0.4425,
|
| 1281 |
+
"grad_norm": 3.4933815002441406,
|
| 1282 |
"learning_rate": 2.3245833333333334e-05,
|
| 1283 |
+
"loss": 0.1266,
|
| 1284 |
"step": 4425
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 0.445,
|
| 1288 |
+
"grad_norm": 3.4990930557250977,
|
| 1289 |
"learning_rate": 2.3141666666666666e-05,
|
| 1290 |
"loss": 0.1466,
|
| 1291 |
"step": 4450
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 0.4475,
|
| 1295 |
+
"grad_norm": 3.077288866043091,
|
| 1296 |
"learning_rate": 2.3037499999999998e-05,
|
| 1297 |
+
"loss": 0.1486,
|
| 1298 |
"step": 4475
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 0.45,
|
| 1302 |
+
"grad_norm": 3.375373363494873,
|
| 1303 |
"learning_rate": 2.293333333333333e-05,
|
| 1304 |
+
"loss": 0.1391,
|
| 1305 |
"step": 4500
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 0.4525,
|
| 1309 |
+
"grad_norm": 4.026995658874512,
|
| 1310 |
"learning_rate": 2.2829166666666665e-05,
|
| 1311 |
"loss": 0.1807,
|
| 1312 |
"step": 4525
|
| 1313 |
},
|
| 1314 |
{
|
| 1315 |
"epoch": 0.455,
|
| 1316 |
+
"grad_norm": 4.551440238952637,
|
| 1317 |
"learning_rate": 2.2724999999999997e-05,
|
| 1318 |
+
"loss": 0.2373,
|
| 1319 |
"step": 4550
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 0.4575,
|
| 1323 |
+
"grad_norm": 5.197234153747559,
|
| 1324 |
"learning_rate": 2.2620833333333332e-05,
|
| 1325 |
+
"loss": 0.2225,
|
| 1326 |
"step": 4575
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 0.46,
|
| 1330 |
+
"grad_norm": 4.593672752380371,
|
| 1331 |
"learning_rate": 2.2516666666666664e-05,
|
| 1332 |
+
"loss": 0.2394,
|
| 1333 |
"step": 4600
|
| 1334 |
},
|
| 1335 |
{
|
| 1336 |
"epoch": 0.4625,
|
| 1337 |
+
"grad_norm": 4.616493225097656,
|
| 1338 |
"learning_rate": 2.24125e-05,
|
| 1339 |
+
"loss": 0.2431,
|
| 1340 |
"step": 4625
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 0.465,
|
| 1344 |
+
"grad_norm": 5.233091831207275,
|
| 1345 |
"learning_rate": 2.230833333333333e-05,
|
| 1346 |
+
"loss": 0.2357,
|
| 1347 |
"step": 4650
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 0.4675,
|
| 1351 |
+
"grad_norm": 4.357009410858154,
|
| 1352 |
"learning_rate": 2.2204166666666667e-05,
|
| 1353 |
+
"loss": 0.2041,
|
| 1354 |
"step": 4675
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
"epoch": 0.47,
|
| 1358 |
+
"grad_norm": 3.481781482696533,
|
| 1359 |
"learning_rate": 2.21e-05,
|
| 1360 |
+
"loss": 0.1817,
|
| 1361 |
"step": 4700
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 0.4725,
|
| 1365 |
+
"grad_norm": 4.179861545562744,
|
| 1366 |
"learning_rate": 2.1995833333333334e-05,
|
| 1367 |
"loss": 0.1953,
|
| 1368 |
"step": 4725
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 0.475,
|
| 1372 |
+
"grad_norm": 3.9264073371887207,
|
| 1373 |
"learning_rate": 2.1891666666666662e-05,
|
| 1374 |
+
"loss": 0.1927,
|
| 1375 |
"step": 4750
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 0.4775,
|
| 1379 |
+
"grad_norm": 3.4452853202819824,
|
| 1380 |
"learning_rate": 2.1787499999999998e-05,
|
| 1381 |
+
"loss": 0.1443,
|
| 1382 |
"step": 4775
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 0.48,
|
| 1386 |
+
"grad_norm": 2.9945247173309326,
|
| 1387 |
"learning_rate": 2.168333333333333e-05,
|
| 1388 |
+
"loss": 0.1587,
|
| 1389 |
"step": 4800
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 0.4825,
|
| 1393 |
+
"grad_norm": 3.666703701019287,
|
| 1394 |
"learning_rate": 2.1579166666666665e-05,
|
| 1395 |
+
"loss": 0.1314,
|
| 1396 |
"step": 4825
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 0.485,
|
| 1400 |
+
"grad_norm": 3.7032406330108643,
|
| 1401 |
"learning_rate": 2.1474999999999997e-05,
|
| 1402 |
+
"loss": 0.1524,
|
| 1403 |
"step": 4850
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 0.4875,
|
| 1407 |
+
"grad_norm": 3.9737789630889893,
|
| 1408 |
"learning_rate": 2.1370833333333332e-05,
|
| 1409 |
+
"loss": 0.1605,
|
| 1410 |
"step": 4875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 0.49,
|
| 1414 |
+
"grad_norm": 4.30121374130249,
|
| 1415 |
"learning_rate": 2.1266666666666664e-05,
|
| 1416 |
"loss": 0.1834,
|
| 1417 |
"step": 4900
|
| 1418 |
},
|
| 1419 |
{
|
| 1420 |
"epoch": 0.4925,
|
| 1421 |
+
"grad_norm": 3.051647424697876,
|
| 1422 |
"learning_rate": 2.11625e-05,
|
| 1423 |
+
"loss": 0.1651,
|
| 1424 |
"step": 4925
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 0.495,
|
| 1428 |
+
"grad_norm": 3.1889472007751465,
|
| 1429 |
"learning_rate": 2.105833333333333e-05,
|
| 1430 |
+
"loss": 0.1392,
|
| 1431 |
"step": 4950
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 0.4975,
|
| 1435 |
+
"grad_norm": 3.226409912109375,
|
| 1436 |
"learning_rate": 2.0954166666666667e-05,
|
| 1437 |
+
"loss": 0.1286,
|
| 1438 |
"step": 4975
|
| 1439 |
},
|
| 1440 |
{
|
| 1441 |
"epoch": 0.5,
|
| 1442 |
+
"grad_norm": 3.729316234588623,
|
| 1443 |
"learning_rate": 2.085e-05,
|
| 1444 |
+
"loss": 0.132,
|
| 1445 |
"step": 5000
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 0.5,
|
| 1449 |
+
"eval_loss": 0.45741865038871765,
|
| 1450 |
+
"eval_runtime": 38.6756,
|
| 1451 |
+
"eval_samples_per_second": 43.723,
|
| 1452 |
+
"eval_steps_per_second": 2.741,
|
| 1453 |
+
"eval_wer": 19.014612214312475,
|
| 1454 |
"step": 5000
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
"epoch": 0.5025,
|
| 1458 |
+
"grad_norm": 3.4599833488464355,
|
| 1459 |
"learning_rate": 2.074583333333333e-05,
|
| 1460 |
+
"loss": 0.1868,
|
| 1461 |
"step": 5025
|
| 1462 |
},
|
| 1463 |
{
|
| 1464 |
"epoch": 0.505,
|
| 1465 |
+
"grad_norm": 3.7839012145996094,
|
| 1466 |
"learning_rate": 2.0641666666666662e-05,
|
| 1467 |
+
"loss": 0.1635,
|
| 1468 |
"step": 5050
|
| 1469 |
},
|
| 1470 |
{
|
| 1471 |
"epoch": 0.5075,
|
| 1472 |
+
"grad_norm": 4.264291286468506,
|
| 1473 |
"learning_rate": 2.0537499999999998e-05,
|
| 1474 |
+
"loss": 0.2522,
|
| 1475 |
"step": 5075
|
| 1476 |
},
|
| 1477 |
{
|
| 1478 |
"epoch": 0.51,
|
| 1479 |
+
"grad_norm": 4.526363849639893,
|
| 1480 |
"learning_rate": 2.043333333333333e-05,
|
| 1481 |
+
"loss": 0.2296,
|
| 1482 |
"step": 5100
|
| 1483 |
},
|
| 1484 |
{
|
| 1485 |
"epoch": 0.5125,
|
| 1486 |
+
"grad_norm": 4.191585063934326,
|
| 1487 |
"learning_rate": 2.0329166666666665e-05,
|
| 1488 |
+
"loss": 0.1969,
|
| 1489 |
"step": 5125
|
| 1490 |
},
|
| 1491 |
{
|
| 1492 |
"epoch": 0.515,
|
| 1493 |
+
"grad_norm": 5.047796249389648,
|
| 1494 |
"learning_rate": 2.0224999999999997e-05,
|
| 1495 |
+
"loss": 0.204,
|
| 1496 |
"step": 5150
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
"epoch": 0.5175,
|
| 1500 |
+
"grad_norm": 2.665344476699829,
|
| 1501 |
"learning_rate": 2.0120833333333332e-05,
|
| 1502 |
+
"loss": 0.1704,
|
| 1503 |
"step": 5175
|
| 1504 |
},
|
| 1505 |
{
|
| 1506 |
"epoch": 0.52,
|
| 1507 |
+
"grad_norm": 2.922651529312134,
|
| 1508 |
"learning_rate": 2.0016666666666664e-05,
|
| 1509 |
+
"loss": 0.117,
|
| 1510 |
"step": 5200
|
| 1511 |
},
|
| 1512 |
{
|
| 1513 |
"epoch": 0.5225,
|
| 1514 |
+
"grad_norm": 2.9956483840942383,
|
| 1515 |
"learning_rate": 1.99125e-05,
|
| 1516 |
+
"loss": 0.123,
|
| 1517 |
"step": 5225
|
| 1518 |
},
|
| 1519 |
{
|
| 1520 |
"epoch": 0.525,
|
| 1521 |
+
"grad_norm": 2.745727777481079,
|
| 1522 |
"learning_rate": 1.980833333333333e-05,
|
| 1523 |
+
"loss": 0.1223,
|
| 1524 |
"step": 5250
|
| 1525 |
},
|
| 1526 |
{
|
| 1527 |
"epoch": 0.5275,
|
| 1528 |
+
"grad_norm": 2.3079681396484375,
|
| 1529 |
"learning_rate": 1.9704166666666667e-05,
|
| 1530 |
+
"loss": 0.1229,
|
| 1531 |
"step": 5275
|
| 1532 |
},
|
| 1533 |
{
|
| 1534 |
"epoch": 0.53,
|
| 1535 |
+
"grad_norm": 3.4931304454803467,
|
| 1536 |
"learning_rate": 1.9599999999999995e-05,
|
| 1537 |
"loss": 0.1287,
|
| 1538 |
"step": 5300
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
"epoch": 0.5325,
|
| 1542 |
+
"grad_norm": 3.3155760765075684,
|
| 1543 |
"learning_rate": 1.949583333333333e-05,
|
| 1544 |
+
"loss": 0.1284,
|
| 1545 |
"step": 5325
|
| 1546 |
},
|
| 1547 |
{
|
| 1548 |
"epoch": 0.535,
|
| 1549 |
+
"grad_norm": 4.483437538146973,
|
| 1550 |
"learning_rate": 1.9391666666666663e-05,
|
| 1551 |
+
"loss": 0.1695,
|
| 1552 |
"step": 5350
|
| 1553 |
},
|
| 1554 |
{
|
| 1555 |
"epoch": 0.5375,
|
| 1556 |
+
"grad_norm": 4.182435512542725,
|
| 1557 |
"learning_rate": 1.9287499999999998e-05,
|
| 1558 |
+
"loss": 0.2029,
|
| 1559 |
"step": 5375
|
| 1560 |
},
|
| 1561 |
{
|
| 1562 |
"epoch": 0.54,
|
| 1563 |
+
"grad_norm": 3.4684391021728516,
|
| 1564 |
"learning_rate": 1.918333333333333e-05,
|
| 1565 |
+
"loss": 0.1907,
|
| 1566 |
"step": 5400
|
| 1567 |
},
|
| 1568 |
{
|
| 1569 |
"epoch": 0.5425,
|
| 1570 |
+
"grad_norm": 4.1064019203186035,
|
| 1571 |
"learning_rate": 1.9079166666666665e-05,
|
| 1572 |
+
"loss": 0.1847,
|
| 1573 |
"step": 5425
|
| 1574 |
},
|
| 1575 |
{
|
| 1576 |
"epoch": 0.545,
|
| 1577 |
+
"grad_norm": 3.6676344871520996,
|
| 1578 |
"learning_rate": 1.8974999999999997e-05,
|
| 1579 |
+
"loss": 0.1644,
|
| 1580 |
"step": 5450
|
| 1581 |
},
|
| 1582 |
{
|
| 1583 |
"epoch": 0.5475,
|
| 1584 |
+
"grad_norm": 3.6105685234069824,
|
| 1585 |
"learning_rate": 1.8870833333333332e-05,
|
| 1586 |
+
"loss": 0.1368,
|
| 1587 |
"step": 5475
|
| 1588 |
},
|
| 1589 |
{
|
| 1590 |
"epoch": 0.55,
|
| 1591 |
+
"grad_norm": 3.2299535274505615,
|
| 1592 |
"learning_rate": 1.8766666666666664e-05,
|
| 1593 |
+
"loss": 0.138,
|
| 1594 |
"step": 5500
|
| 1595 |
},
|
| 1596 |
{
|
| 1597 |
"epoch": 0.5525,
|
| 1598 |
+
"grad_norm": 5.327032566070557,
|
| 1599 |
"learning_rate": 1.8662499999999996e-05,
|
| 1600 |
+
"loss": 0.1786,
|
| 1601 |
"step": 5525
|
| 1602 |
},
|
| 1603 |
{
|
| 1604 |
"epoch": 0.555,
|
| 1605 |
+
"grad_norm": 4.6841583251953125,
|
| 1606 |
"learning_rate": 1.855833333333333e-05,
|
| 1607 |
+
"loss": 0.2034,
|
| 1608 |
"step": 5550
|
| 1609 |
},
|
| 1610 |
{
|
| 1611 |
"epoch": 0.5575,
|
| 1612 |
+
"grad_norm": 4.341950416564941,
|
| 1613 |
"learning_rate": 1.8454166666666663e-05,
|
| 1614 |
+
"loss": 0.1611,
|
| 1615 |
"step": 5575
|
| 1616 |
},
|
| 1617 |
{
|
| 1618 |
"epoch": 0.56,
|
| 1619 |
+
"grad_norm": 2.254387378692627,
|
| 1620 |
"learning_rate": 1.835e-05,
|
| 1621 |
+
"loss": 0.1087,
|
| 1622 |
"step": 5600
|
| 1623 |
},
|
| 1624 |
{
|
| 1625 |
"epoch": 0.5625,
|
| 1626 |
+
"grad_norm": 3.0610556602478027,
|
| 1627 |
"learning_rate": 1.8245833333333334e-05,
|
| 1628 |
+
"loss": 0.0877,
|
| 1629 |
"step": 5625
|
| 1630 |
},
|
| 1631 |
{
|
| 1632 |
"epoch": 0.565,
|
| 1633 |
+
"grad_norm": 2.6597042083740234,
|
| 1634 |
"learning_rate": 1.8141666666666663e-05,
|
| 1635 |
+
"loss": 0.0893,
|
| 1636 |
"step": 5650
|
| 1637 |
},
|
| 1638 |
{
|
| 1639 |
"epoch": 0.5675,
|
| 1640 |
+
"grad_norm": 2.638615369796753,
|
| 1641 |
"learning_rate": 1.8037499999999998e-05,
|
| 1642 |
+
"loss": 0.1155,
|
| 1643 |
"step": 5675
|
| 1644 |
},
|
| 1645 |
{
|
| 1646 |
"epoch": 0.57,
|
| 1647 |
+
"grad_norm": 3.690338373184204,
|
| 1648 |
"learning_rate": 1.793333333333333e-05,
|
| 1649 |
+
"loss": 0.1252,
|
| 1650 |
"step": 5700
|
| 1651 |
},
|
| 1652 |
{
|
| 1653 |
"epoch": 0.5725,
|
| 1654 |
+
"grad_norm": 2.847730875015259,
|
| 1655 |
"learning_rate": 1.7829166666666665e-05,
|
| 1656 |
+
"loss": 0.1228,
|
| 1657 |
"step": 5725
|
| 1658 |
},
|
| 1659 |
{
|
| 1660 |
"epoch": 0.575,
|
| 1661 |
+
"grad_norm": 2.7520906925201416,
|
| 1662 |
"learning_rate": 1.7725e-05,
|
| 1663 |
+
"loss": 0.1162,
|
| 1664 |
"step": 5750
|
| 1665 |
},
|
| 1666 |
{
|
| 1667 |
"epoch": 0.5775,
|
| 1668 |
+
"grad_norm": 2.643385648727417,
|
| 1669 |
"learning_rate": 1.7620833333333332e-05,
|
| 1670 |
+
"loss": 0.1168,
|
| 1671 |
"step": 5775
|
| 1672 |
},
|
| 1673 |
{
|
| 1674 |
"epoch": 0.58,
|
| 1675 |
+
"grad_norm": 2.70414400100708,
|
| 1676 |
"learning_rate": 1.7516666666666664e-05,
|
| 1677 |
+
"loss": 0.1092,
|
| 1678 |
"step": 5800
|
| 1679 |
},
|
| 1680 |
{
|
| 1681 |
"epoch": 0.5825,
|
| 1682 |
+
"grad_norm": 2.194267988204956,
|
| 1683 |
"learning_rate": 1.7412499999999996e-05,
|
| 1684 |
+
"loss": 0.0949,
|
| 1685 |
"step": 5825
|
| 1686 |
},
|
| 1687 |
{
|
| 1688 |
"epoch": 0.585,
|
| 1689 |
+
"grad_norm": 3.784864902496338,
|
| 1690 |
"learning_rate": 1.730833333333333e-05,
|
| 1691 |
+
"loss": 0.1041,
|
| 1692 |
"step": 5850
|
| 1693 |
},
|
| 1694 |
{
|
| 1695 |
"epoch": 0.5875,
|
| 1696 |
+
"grad_norm": 3.313042402267456,
|
| 1697 |
"learning_rate": 1.7204166666666667e-05,
|
| 1698 |
+
"loss": 0.1105,
|
| 1699 |
"step": 5875
|
| 1700 |
},
|
| 1701 |
{
|
| 1702 |
"epoch": 0.59,
|
| 1703 |
+
"grad_norm": 3.7411179542541504,
|
| 1704 |
"learning_rate": 1.71e-05,
|
| 1705 |
+
"loss": 0.1396,
|
| 1706 |
"step": 5900
|
| 1707 |
},
|
| 1708 |
{
|
| 1709 |
"epoch": 0.5925,
|
| 1710 |
+
"grad_norm": 5.693575859069824,
|
| 1711 |
"learning_rate": 1.6995833333333334e-05,
|
| 1712 |
"loss": 0.2024,
|
| 1713 |
"step": 5925
|
| 1714 |
},
|
| 1715 |
{
|
| 1716 |
"epoch": 0.595,
|
| 1717 |
+
"grad_norm": 3.940126419067383,
|
| 1718 |
"learning_rate": 1.6891666666666663e-05,
|
| 1719 |
+
"loss": 0.1936,
|
| 1720 |
"step": 5950
|
| 1721 |
},
|
| 1722 |
{
|
| 1723 |
"epoch": 0.5975,
|
| 1724 |
+
"grad_norm": 4.423816680908203,
|
| 1725 |
"learning_rate": 1.6787499999999998e-05,
|
| 1726 |
+
"loss": 0.1702,
|
| 1727 |
"step": 5975
|
| 1728 |
},
|
| 1729 |
{
|
| 1730 |
"epoch": 0.6,
|
| 1731 |
+
"grad_norm": 3.7899911403656006,
|
| 1732 |
"learning_rate": 1.6683333333333333e-05,
|
| 1733 |
+
"loss": 0.1588,
|
| 1734 |
"step": 6000
|
| 1735 |
},
|
| 1736 |
{
|
| 1737 |
"epoch": 0.6,
|
| 1738 |
+
"eval_loss": 0.4380021393299103,
|
| 1739 |
+
"eval_runtime": 37.918,
|
| 1740 |
+
"eval_samples_per_second": 44.596,
|
| 1741 |
+
"eval_steps_per_second": 2.796,
|
| 1742 |
+
"eval_wer": 17.821905832396652,
|
| 1743 |
"step": 6000
|
| 1744 |
},
|
| 1745 |
{
|
| 1746 |
"epoch": 0.6025,
|
| 1747 |
+
"grad_norm": 3.4838547706604004,
|
| 1748 |
"learning_rate": 1.6579166666666665e-05,
|
| 1749 |
+
"loss": 0.1486,
|
| 1750 |
"step": 6025
|
| 1751 |
},
|
| 1752 |
{
|
| 1753 |
"epoch": 0.605,
|
| 1754 |
+
"grad_norm": 3.3172895908355713,
|
| 1755 |
"learning_rate": 1.6475e-05,
|
| 1756 |
+
"loss": 0.1349,
|
| 1757 |
"step": 6050
|
| 1758 |
},
|
| 1759 |
{
|
| 1760 |
"epoch": 0.6075,
|
| 1761 |
+
"grad_norm": 2.6951234340667725,
|
| 1762 |
"learning_rate": 1.6370833333333333e-05,
|
| 1763 |
+
"loss": 0.1201,
|
| 1764 |
"step": 6075
|
| 1765 |
},
|
| 1766 |
{
|
| 1767 |
"epoch": 0.61,
|
| 1768 |
+
"grad_norm": 2.4754366874694824,
|
| 1769 |
"learning_rate": 1.6266666666666665e-05,
|
| 1770 |
+
"loss": 0.1109,
|
| 1771 |
"step": 6100
|
| 1772 |
},
|
| 1773 |
{
|
| 1774 |
"epoch": 0.6125,
|
| 1775 |
+
"grad_norm": 3.2808432579040527,
|
| 1776 |
"learning_rate": 1.61625e-05,
|
| 1777 |
+
"loss": 0.115,
|
| 1778 |
"step": 6125
|
| 1779 |
},
|
| 1780 |
{
|
| 1781 |
"epoch": 0.615,
|
| 1782 |
+
"grad_norm": 3.401995897293091,
|
| 1783 |
"learning_rate": 1.6058333333333332e-05,
|
| 1784 |
+
"loss": 0.1293,
|
| 1785 |
"step": 6150
|
| 1786 |
},
|
| 1787 |
{
|
| 1788 |
"epoch": 0.6175,
|
| 1789 |
+
"grad_norm": 3.313483476638794,
|
| 1790 |
"learning_rate": 1.5954166666666667e-05,
|
| 1791 |
+
"loss": 0.1285,
|
| 1792 |
"step": 6175
|
| 1793 |
},
|
| 1794 |
{
|
| 1795 |
"epoch": 0.62,
|
| 1796 |
+
"grad_norm": 4.194374084472656,
|
| 1797 |
"learning_rate": 1.585e-05,
|
| 1798 |
+
"loss": 0.129,
|
| 1799 |
"step": 6200
|
| 1800 |
},
|
| 1801 |
{
|
| 1802 |
"epoch": 0.6225,
|
| 1803 |
+
"grad_norm": 3.408337354660034,
|
| 1804 |
"learning_rate": 1.574583333333333e-05,
|
| 1805 |
+
"loss": 0.1212,
|
| 1806 |
"step": 6225
|
| 1807 |
},
|
| 1808 |
{
|
| 1809 |
"epoch": 0.625,
|
| 1810 |
+
"grad_norm": 3.5416505336761475,
|
| 1811 |
"learning_rate": 1.5641666666666666e-05,
|
| 1812 |
+
"loss": 0.1146,
|
| 1813 |
"step": 6250
|
| 1814 |
},
|
| 1815 |
{
|
| 1816 |
"epoch": 0.6275,
|
| 1817 |
+
"grad_norm": 3.736720561981201,
|
| 1818 |
"learning_rate": 1.5537499999999998e-05,
|
| 1819 |
+
"loss": 0.133,
|
| 1820 |
"step": 6275
|
| 1821 |
},
|
| 1822 |
{
|
| 1823 |
"epoch": 0.63,
|
| 1824 |
+
"grad_norm": 4.025073051452637,
|
| 1825 |
"learning_rate": 1.5433333333333334e-05,
|
| 1826 |
+
"loss": 0.1445,
|
| 1827 |
"step": 6300
|
| 1828 |
},
|
| 1829 |
{
|
| 1830 |
"epoch": 0.6325,
|
| 1831 |
+
"grad_norm": 3.1371307373046875,
|
| 1832 |
"learning_rate": 1.5329166666666665e-05,
|
| 1833 |
+
"loss": 0.1943,
|
| 1834 |
"step": 6325
|
| 1835 |
},
|
| 1836 |
{
|
| 1837 |
"epoch": 0.635,
|
| 1838 |
+
"grad_norm": 3.793010711669922,
|
| 1839 |
"learning_rate": 1.5224999999999999e-05,
|
| 1840 |
+
"loss": 0.176,
|
| 1841 |
"step": 6350
|
| 1842 |
},
|
| 1843 |
{
|
| 1844 |
"epoch": 0.6375,
|
| 1845 |
+
"grad_norm": 3.112964630126953,
|
| 1846 |
"learning_rate": 1.5120833333333331e-05,
|
| 1847 |
+
"loss": 0.1647,
|
| 1848 |
"step": 6375
|
| 1849 |
},
|
| 1850 |
{
|
| 1851 |
"epoch": 0.64,
|
| 1852 |
+
"grad_norm": 2.877613067626953,
|
| 1853 |
"learning_rate": 1.5016666666666665e-05,
|
| 1854 |
+
"loss": 0.1252,
|
| 1855 |
"step": 6400
|
| 1856 |
},
|
| 1857 |
{
|
| 1858 |
"epoch": 0.6425,
|
| 1859 |
+
"grad_norm": 2.556334972381592,
|
| 1860 |
"learning_rate": 1.4912499999999998e-05,
|
| 1861 |
+
"loss": 0.1137,
|
| 1862 |
"step": 6425
|
| 1863 |
},
|
| 1864 |
{
|
| 1865 |
"epoch": 0.645,
|
| 1866 |
+
"grad_norm": 5.248802661895752,
|
| 1867 |
"learning_rate": 1.4808333333333332e-05,
|
| 1868 |
+
"loss": 0.1087,
|
| 1869 |
"step": 6450
|
| 1870 |
},
|
| 1871 |
{
|
| 1872 |
"epoch": 0.6475,
|
| 1873 |
+
"grad_norm": 3.568530321121216,
|
| 1874 |
"learning_rate": 1.4704166666666666e-05,
|
| 1875 |
+
"loss": 0.1548,
|
| 1876 |
"step": 6475
|
| 1877 |
},
|
| 1878 |
{
|
| 1879 |
"epoch": 0.65,
|
| 1880 |
+
"grad_norm": 3.2436602115631104,
|
| 1881 |
"learning_rate": 1.4599999999999997e-05,
|
| 1882 |
+
"loss": 0.1494,
|
| 1883 |
"step": 6500
|
| 1884 |
},
|
| 1885 |
{
|
| 1886 |
"epoch": 0.6525,
|
| 1887 |
+
"grad_norm": 5.127964019775391,
|
| 1888 |
"learning_rate": 1.4495833333333331e-05,
|
| 1889 |
+
"loss": 0.1935,
|
| 1890 |
"step": 6525
|
| 1891 |
},
|
| 1892 |
{
|
| 1893 |
"epoch": 0.655,
|
| 1894 |
+
"grad_norm": 3.848689079284668,
|
| 1895 |
"learning_rate": 1.4391666666666665e-05,
|
| 1896 |
+
"loss": 0.1781,
|
| 1897 |
"step": 6550
|
| 1898 |
},
|
| 1899 |
{
|
| 1900 |
"epoch": 0.6575,
|
| 1901 |
+
"grad_norm": 3.3769173622131348,
|
| 1902 |
"learning_rate": 1.4287499999999998e-05,
|
| 1903 |
+
"loss": 0.178,
|
| 1904 |
"step": 6575
|
| 1905 |
},
|
| 1906 |
{
|
| 1907 |
"epoch": 0.66,
|
| 1908 |
+
"grad_norm": 3.279507875442505,
|
| 1909 |
"learning_rate": 1.4183333333333332e-05,
|
| 1910 |
+
"loss": 0.1559,
|
| 1911 |
"step": 6600
|
| 1912 |
},
|
| 1913 |
{
|
| 1914 |
"epoch": 0.6625,
|
| 1915 |
+
"grad_norm": 4.212521553039551,
|
| 1916 |
"learning_rate": 1.4079166666666666e-05,
|
| 1917 |
+
"loss": 0.1667,
|
| 1918 |
"step": 6625
|
| 1919 |
},
|
| 1920 |
{
|
| 1921 |
"epoch": 0.665,
|
| 1922 |
+
"grad_norm": 3.3866727352142334,
|
| 1923 |
"learning_rate": 1.3974999999999997e-05,
|
| 1924 |
+
"loss": 0.1966,
|
| 1925 |
"step": 6650
|
| 1926 |
},
|
| 1927 |
{
|
| 1928 |
"epoch": 0.6675,
|
| 1929 |
+
"grad_norm": 3.52656888961792,
|
| 1930 |
"learning_rate": 1.3870833333333331e-05,
|
| 1931 |
+
"loss": 0.1675,
|
| 1932 |
"step": 6675
|
| 1933 |
},
|
| 1934 |
{
|
| 1935 |
"epoch": 0.67,
|
| 1936 |
+
"grad_norm": 2.9528002738952637,
|
| 1937 |
"learning_rate": 1.3766666666666665e-05,
|
| 1938 |
+
"loss": 0.1612,
|
| 1939 |
"step": 6700
|
| 1940 |
},
|
| 1941 |
{
|
| 1942 |
"epoch": 0.6725,
|
| 1943 |
+
"grad_norm": 3.4069855213165283,
|
| 1944 |
"learning_rate": 1.3662499999999998e-05,
|
| 1945 |
+
"loss": 0.1162,
|
| 1946 |
"step": 6725
|
| 1947 |
},
|
| 1948 |
{
|
| 1949 |
"epoch": 0.675,
|
| 1950 |
+
"grad_norm": 3.048247814178467,
|
| 1951 |
"learning_rate": 1.3558333333333334e-05,
|
| 1952 |
"loss": 0.105,
|
| 1953 |
"step": 6750
|
| 1954 |
},
|
| 1955 |
{
|
| 1956 |
"epoch": 0.6775,
|
| 1957 |
+
"grad_norm": 3.930851459503174,
|
| 1958 |
"learning_rate": 1.3454166666666664e-05,
|
| 1959 |
+
"loss": 0.1272,
|
| 1960 |
"step": 6775
|
| 1961 |
},
|
| 1962 |
{
|
| 1963 |
"epoch": 0.68,
|
| 1964 |
+
"grad_norm": 4.210860729217529,
|
| 1965 |
"learning_rate": 1.3349999999999998e-05,
|
| 1966 |
+
"loss": 0.2174,
|
| 1967 |
"step": 6800
|
| 1968 |
},
|
| 1969 |
{
|
| 1970 |
"epoch": 0.6825,
|
| 1971 |
+
"grad_norm": 6.0455498695373535,
|
| 1972 |
"learning_rate": 1.3245833333333331e-05,
|
| 1973 |
+
"loss": 0.194,
|
| 1974 |
"step": 6825
|
| 1975 |
},
|
| 1976 |
{
|
| 1977 |
"epoch": 0.685,
|
| 1978 |
+
"grad_norm": 5.270689487457275,
|
| 1979 |
"learning_rate": 1.3141666666666665e-05,
|
| 1980 |
+
"loss": 0.2003,
|
| 1981 |
"step": 6850
|
| 1982 |
},
|
| 1983 |
{
|
| 1984 |
"epoch": 0.6875,
|
| 1985 |
+
"grad_norm": 4.297138690948486,
|
| 1986 |
"learning_rate": 1.30375e-05,
|
| 1987 |
+
"loss": 0.1923,
|
| 1988 |
"step": 6875
|
| 1989 |
},
|
| 1990 |
{
|
| 1991 |
"epoch": 0.69,
|
| 1992 |
+
"grad_norm": 4.304157257080078,
|
| 1993 |
"learning_rate": 1.2933333333333334e-05,
|
| 1994 |
+
"loss": 0.1884,
|
| 1995 |
"step": 6900
|
| 1996 |
},
|
| 1997 |
{
|
| 1998 |
"epoch": 0.6925,
|
| 1999 |
+
"grad_norm": 3.3254096508026123,
|
| 2000 |
"learning_rate": 1.2829166666666664e-05,
|
| 2001 |
+
"loss": 0.1713,
|
| 2002 |
"step": 6925
|
| 2003 |
},
|
| 2004 |
{
|
| 2005 |
"epoch": 0.695,
|
| 2006 |
+
"grad_norm": 4.185650825500488,
|
| 2007 |
"learning_rate": 1.2724999999999998e-05,
|
| 2008 |
+
"loss": 0.1756,
|
| 2009 |
"step": 6950
|
| 2010 |
},
|
| 2011 |
{
|
| 2012 |
"epoch": 0.6975,
|
| 2013 |
+
"grad_norm": 4.220348358154297,
|
| 2014 |
"learning_rate": 1.2620833333333333e-05,
|
| 2015 |
+
"loss": 0.1958,
|
| 2016 |
"step": 6975
|
| 2017 |
},
|
| 2018 |
{
|
| 2019 |
"epoch": 0.7,
|
| 2020 |
+
"grad_norm": 4.203312873840332,
|
| 2021 |
"learning_rate": 1.2516666666666667e-05,
|
| 2022 |
+
"loss": 0.1841,
|
| 2023 |
"step": 7000
|
| 2024 |
},
|
| 2025 |
{
|
| 2026 |
"epoch": 0.7,
|
| 2027 |
+
"eval_loss": 0.4394695460796356,
|
| 2028 |
+
"eval_runtime": 37.5658,
|
| 2029 |
+
"eval_samples_per_second": 45.014,
|
| 2030 |
+
"eval_steps_per_second": 2.822,
|
| 2031 |
+
"eval_wer": 16.666666666666664,
|
| 2032 |
"step": 7000
|
| 2033 |
},
|
| 2034 |
{
|
| 2035 |
"epoch": 0.7025,
|
| 2036 |
+
"grad_norm": 6.67822265625,
|
| 2037 |
"learning_rate": 1.2416666666666666e-05,
|
| 2038 |
+
"loss": 0.3408,
|
| 2039 |
"step": 7025
|
| 2040 |
},
|
| 2041 |
{
|
| 2042 |
"epoch": 0.705,
|
| 2043 |
+
"grad_norm": 8.843494415283203,
|
| 2044 |
"learning_rate": 1.2312499999999998e-05,
|
| 2045 |
+
"loss": 0.6013,
|
| 2046 |
"step": 7050
|
| 2047 |
},
|
| 2048 |
{
|
| 2049 |
"epoch": 0.7075,
|
| 2050 |
+
"grad_norm": 5.624876499176025,
|
| 2051 |
"learning_rate": 1.2208333333333331e-05,
|
| 2052 |
+
"loss": 0.5257,
|
| 2053 |
"step": 7075
|
| 2054 |
},
|
| 2055 |
{
|
| 2056 |
"epoch": 0.71,
|
| 2057 |
+
"grad_norm": 4.507925987243652,
|
| 2058 |
"learning_rate": 1.2104166666666665e-05,
|
| 2059 |
+
"loss": 0.4356,
|
| 2060 |
"step": 7100
|
| 2061 |
},
|
| 2062 |
{
|
| 2063 |
"epoch": 0.7125,
|
| 2064 |
+
"grad_norm": 4.30970573425293,
|
| 2065 |
"learning_rate": 1.1999999999999999e-05,
|
| 2066 |
+
"loss": 0.2381,
|
| 2067 |
"step": 7125
|
| 2068 |
},
|
| 2069 |
{
|
| 2070 |
"epoch": 0.715,
|
| 2071 |
+
"grad_norm": 5.139924049377441,
|
| 2072 |
"learning_rate": 1.1895833333333332e-05,
|
| 2073 |
+
"loss": 0.185,
|
| 2074 |
"step": 7150
|
| 2075 |
},
|
| 2076 |
{
|
| 2077 |
"epoch": 0.7175,
|
| 2078 |
+
"grad_norm": 2.9631428718566895,
|
| 2079 |
"learning_rate": 1.1791666666666666e-05,
|
| 2080 |
+
"loss": 0.17,
|
| 2081 |
"step": 7175
|
| 2082 |
},
|
| 2083 |
{
|
| 2084 |
"epoch": 0.72,
|
| 2085 |
+
"grad_norm": 3.6727359294891357,
|
| 2086 |
"learning_rate": 1.1687499999999998e-05,
|
| 2087 |
+
"loss": 0.1323,
|
| 2088 |
"step": 7200
|
| 2089 |
},
|
| 2090 |
{
|
| 2091 |
"epoch": 0.7225,
|
| 2092 |
+
"grad_norm": 2.7629499435424805,
|
| 2093 |
"learning_rate": 1.1583333333333331e-05,
|
| 2094 |
+
"loss": 0.128,
|
| 2095 |
"step": 7225
|
| 2096 |
},
|
| 2097 |
{
|
| 2098 |
"epoch": 0.725,
|
| 2099 |
+
"grad_norm": 3.040910482406616,
|
| 2100 |
"learning_rate": 1.1479166666666665e-05,
|
| 2101 |
+
"loss": 0.1371,
|
| 2102 |
"step": 7250
|
| 2103 |
},
|
| 2104 |
{
|
| 2105 |
"epoch": 0.7275,
|
| 2106 |
+
"grad_norm": 2.7423293590545654,
|
| 2107 |
"learning_rate": 1.1374999999999999e-05,
|
| 2108 |
+
"loss": 0.1128,
|
| 2109 |
"step": 7275
|
| 2110 |
},
|
| 2111 |
{
|
| 2112 |
"epoch": 0.73,
|
| 2113 |
+
"grad_norm": 2.2156248092651367,
|
| 2114 |
"learning_rate": 1.1270833333333332e-05,
|
| 2115 |
+
"loss": 0.0964,
|
| 2116 |
"step": 7300
|
| 2117 |
},
|
| 2118 |
{
|
| 2119 |
"epoch": 0.7325,
|
| 2120 |
+
"grad_norm": 2.783275604248047,
|
| 2121 |
"learning_rate": 1.1166666666666664e-05,
|
| 2122 |
+
"loss": 0.101,
|
| 2123 |
"step": 7325
|
| 2124 |
},
|
| 2125 |
{
|
| 2126 |
"epoch": 0.735,
|
| 2127 |
+
"grad_norm": 4.380770206451416,
|
| 2128 |
"learning_rate": 1.1062499999999998e-05,
|
| 2129 |
+
"loss": 0.132,
|
| 2130 |
"step": 7350
|
| 2131 |
},
|
| 2132 |
{
|
| 2133 |
"epoch": 0.7375,
|
| 2134 |
+
"grad_norm": 3.904127597808838,
|
| 2135 |
"learning_rate": 1.0958333333333331e-05,
|
| 2136 |
+
"loss": 0.1807,
|
| 2137 |
"step": 7375
|
| 2138 |
},
|
| 2139 |
{
|
| 2140 |
"epoch": 0.74,
|
| 2141 |
+
"grad_norm": 5.420929431915283,
|
| 2142 |
"learning_rate": 1.0854166666666665e-05,
|
| 2143 |
+
"loss": 0.1436,
|
| 2144 |
"step": 7400
|
| 2145 |
},
|
| 2146 |
{
|
| 2147 |
"epoch": 0.7425,
|
| 2148 |
+
"grad_norm": 2.929569721221924,
|
| 2149 |
"learning_rate": 1.075e-05,
|
| 2150 |
+
"loss": 0.1648,
|
| 2151 |
"step": 7425
|
| 2152 |
},
|
| 2153 |
{
|
| 2154 |
"epoch": 0.745,
|
| 2155 |
+
"grad_norm": 3.0535011291503906,
|
| 2156 |
"learning_rate": 1.0645833333333334e-05,
|
| 2157 |
+
"loss": 0.1166,
|
| 2158 |
"step": 7450
|
| 2159 |
},
|
| 2160 |
{
|
| 2161 |
"epoch": 0.7475,
|
| 2162 |
+
"grad_norm": 2.630368232727051,
|
| 2163 |
"learning_rate": 1.0541666666666664e-05,
|
| 2164 |
+
"loss": 0.1081,
|
| 2165 |
"step": 7475
|
| 2166 |
},
|
| 2167 |
{
|
| 2168 |
"epoch": 0.75,
|
| 2169 |
+
"grad_norm": 4.1855621337890625,
|
| 2170 |
"learning_rate": 1.0437499999999998e-05,
|
| 2171 |
+
"loss": 0.1033,
|
| 2172 |
"step": 7500
|
| 2173 |
},
|
| 2174 |
{
|
| 2175 |
"epoch": 0.7525,
|
| 2176 |
+
"grad_norm": 4.28782844543457,
|
| 2177 |
"learning_rate": 1.0333333333333332e-05,
|
| 2178 |
+
"loss": 0.2037,
|
| 2179 |
"step": 7525
|
| 2180 |
},
|
| 2181 |
{
|
| 2182 |
"epoch": 0.755,
|
| 2183 |
+
"grad_norm": 5.351161956787109,
|
| 2184 |
"learning_rate": 1.0229166666666667e-05,
|
| 2185 |
+
"loss": 0.2651,
|
| 2186 |
"step": 7550
|
| 2187 |
},
|
| 2188 |
{
|
| 2189 |
"epoch": 0.7575,
|
| 2190 |
+
"grad_norm": 5.073176383972168,
|
| 2191 |
"learning_rate": 1.0125e-05,
|
| 2192 |
+
"loss": 0.2933,
|
| 2193 |
"step": 7575
|
| 2194 |
},
|
| 2195 |
{
|
| 2196 |
"epoch": 0.76,
|
| 2197 |
+
"grad_norm": 2.562958002090454,
|
| 2198 |
"learning_rate": 1.0020833333333334e-05,
|
| 2199 |
+
"loss": 0.1829,
|
| 2200 |
"step": 7600
|
| 2201 |
},
|
| 2202 |
{
|
| 2203 |
"epoch": 0.7625,
|
| 2204 |
+
"grad_norm": 4.164181709289551,
|
| 2205 |
"learning_rate": 9.916666666666664e-06,
|
| 2206 |
+
"loss": 0.1309,
|
| 2207 |
"step": 7625
|
| 2208 |
},
|
| 2209 |
{
|
| 2210 |
"epoch": 0.765,
|
| 2211 |
+
"grad_norm": 2.616945743560791,
|
| 2212 |
"learning_rate": 9.8125e-06,
|
| 2213 |
+
"loss": 0.1077,
|
| 2214 |
"step": 7650
|
| 2215 |
},
|
| 2216 |
{
|
| 2217 |
"epoch": 0.7675,
|
| 2218 |
+
"grad_norm": 4.083223342895508,
|
| 2219 |
"learning_rate": 9.708333333333333e-06,
|
| 2220 |
+
"loss": 0.1442,
|
| 2221 |
"step": 7675
|
| 2222 |
},
|
| 2223 |
{
|
| 2224 |
"epoch": 0.77,
|
| 2225 |
+
"grad_norm": 4.046257972717285,
|
| 2226 |
"learning_rate": 9.604166666666667e-06,
|
| 2227 |
+
"loss": 0.1725,
|
| 2228 |
"step": 7700
|
| 2229 |
},
|
| 2230 |
{
|
| 2231 |
"epoch": 0.7725,
|
| 2232 |
+
"grad_norm": 4.065029621124268,
|
| 2233 |
"learning_rate": 9.5e-06,
|
| 2234 |
+
"loss": 0.1833,
|
| 2235 |
"step": 7725
|
| 2236 |
},
|
| 2237 |
{
|
| 2238 |
"epoch": 0.775,
|
| 2239 |
+
"grad_norm": 3.321071147918701,
|
| 2240 |
"learning_rate": 9.39583333333333e-06,
|
| 2241 |
"loss": 0.1671,
|
| 2242 |
"step": 7750
|
| 2243 |
},
|
| 2244 |
{
|
| 2245 |
"epoch": 0.7775,
|
| 2246 |
+
"grad_norm": 3.388200044631958,
|
| 2247 |
"learning_rate": 9.291666666666666e-06,
|
| 2248 |
+
"loss": 0.1196,
|
| 2249 |
"step": 7775
|
| 2250 |
},
|
| 2251 |
{
|
| 2252 |
"epoch": 0.78,
|
| 2253 |
+
"grad_norm": 4.393247604370117,
|
| 2254 |
"learning_rate": 9.1875e-06,
|
| 2255 |
+
"loss": 0.1399,
|
| 2256 |
"step": 7800
|
| 2257 |
},
|
| 2258 |
{
|
| 2259 |
"epoch": 0.7825,
|
| 2260 |
+
"grad_norm": 3.463327646255493,
|
| 2261 |
"learning_rate": 9.083333333333333e-06,
|
| 2262 |
+
"loss": 0.3125,
|
| 2263 |
"step": 7825
|
| 2264 |
},
|
| 2265 |
{
|
| 2266 |
"epoch": 0.785,
|
| 2267 |
+
"grad_norm": 2.6306939125061035,
|
| 2268 |
"learning_rate": 8.979166666666665e-06,
|
| 2269 |
+
"loss": 0.153,
|
| 2270 |
"step": 7850
|
| 2271 |
},
|
| 2272 |
{
|
| 2273 |
"epoch": 0.7875,
|
| 2274 |
+
"grad_norm": 2.3278794288635254,
|
| 2275 |
"learning_rate": 8.874999999999999e-06,
|
| 2276 |
+
"loss": 0.1141,
|
| 2277 |
"step": 7875
|
| 2278 |
},
|
| 2279 |
{
|
| 2280 |
"epoch": 0.79,
|
| 2281 |
+
"grad_norm": 2.850830554962158,
|
| 2282 |
"learning_rate": 8.770833333333333e-06,
|
| 2283 |
+
"loss": 0.1096,
|
| 2284 |
"step": 7900
|
| 2285 |
},
|
| 2286 |
{
|
| 2287 |
"epoch": 0.7925,
|
| 2288 |
+
"grad_norm": 2.4107863903045654,
|
| 2289 |
"learning_rate": 8.666666666666666e-06,
|
| 2290 |
+
"loss": 0.1117,
|
| 2291 |
"step": 7925
|
| 2292 |
},
|
| 2293 |
{
|
| 2294 |
"epoch": 0.795,
|
| 2295 |
+
"grad_norm": 3.420828104019165,
|
| 2296 |
"learning_rate": 8.5625e-06,
|
| 2297 |
+
"loss": 0.1071,
|
| 2298 |
"step": 7950
|
| 2299 |
},
|
| 2300 |
{
|
| 2301 |
"epoch": 0.7975,
|
| 2302 |
+
"grad_norm": 3.005918025970459,
|
| 2303 |
"learning_rate": 8.458333333333333e-06,
|
| 2304 |
+
"loss": 0.1146,
|
| 2305 |
"step": 7975
|
| 2306 |
},
|
| 2307 |
{
|
| 2308 |
"epoch": 0.8,
|
| 2309 |
+
"grad_norm": 3.3296029567718506,
|
| 2310 |
"learning_rate": 8.354166666666665e-06,
|
| 2311 |
+
"loss": 0.143,
|
| 2312 |
"step": 8000
|
| 2313 |
},
|
| 2314 |
{
|
| 2315 |
"epoch": 0.8,
|
| 2316 |
+
"eval_loss": 0.3718983232975006,
|
| 2317 |
+
"eval_runtime": 38.8288,
|
| 2318 |
+
"eval_samples_per_second": 43.55,
|
| 2319 |
+
"eval_steps_per_second": 2.73,
|
| 2320 |
+
"eval_wer": 15.448982140626951,
|
| 2321 |
"step": 8000
|
| 2322 |
},
|
| 2323 |
{
|
| 2324 |
"epoch": 0.8025,
|
| 2325 |
+
"grad_norm": 4.029442310333252,
|
| 2326 |
"learning_rate": 8.249999999999999e-06,
|
| 2327 |
+
"loss": 0.1663,
|
| 2328 |
"step": 8025
|
| 2329 |
},
|
| 2330 |
{
|
| 2331 |
"epoch": 0.805,
|
| 2332 |
+
"grad_norm": 3.7387983798980713,
|
| 2333 |
"learning_rate": 8.145833333333333e-06,
|
| 2334 |
+
"loss": 0.1808,
|
| 2335 |
"step": 8050
|
| 2336 |
},
|
| 2337 |
{
|
| 2338 |
"epoch": 0.8075,
|
| 2339 |
+
"grad_norm": 3.0396807193756104,
|
| 2340 |
"learning_rate": 8.041666666666666e-06,
|
| 2341 |
+
"loss": 0.1351,
|
| 2342 |
"step": 8075
|
| 2343 |
},
|
| 2344 |
{
|
| 2345 |
"epoch": 0.81,
|
| 2346 |
+
"grad_norm": 2.9825220108032227,
|
| 2347 |
"learning_rate": 7.9375e-06,
|
| 2348 |
+
"loss": 0.1215,
|
| 2349 |
"step": 8100
|
| 2350 |
},
|
| 2351 |
{
|
| 2352 |
"epoch": 0.8125,
|
| 2353 |
+
"grad_norm": 2.495398998260498,
|
| 2354 |
"learning_rate": 7.833333333333333e-06,
|
| 2355 |
+
"loss": 0.1054,
|
| 2356 |
"step": 8125
|
| 2357 |
},
|
| 2358 |
{
|
| 2359 |
"epoch": 0.815,
|
| 2360 |
+
"grad_norm": 3.750054121017456,
|
| 2361 |
"learning_rate": 7.729166666666665e-06,
|
| 2362 |
+
"loss": 0.1049,
|
| 2363 |
"step": 8150
|
| 2364 |
},
|
| 2365 |
{
|
| 2366 |
"epoch": 0.8175,
|
| 2367 |
+
"grad_norm": 2.1268110275268555,
|
| 2368 |
"learning_rate": 7.625e-06,
|
| 2369 |
+
"loss": 0.0932,
|
| 2370 |
"step": 8175
|
| 2371 |
},
|
| 2372 |
{
|
| 2373 |
"epoch": 0.82,
|
| 2374 |
+
"grad_norm": 3.2984964847564697,
|
| 2375 |
"learning_rate": 7.520833333333332e-06,
|
| 2376 |
+
"loss": 0.0759,
|
| 2377 |
"step": 8200
|
| 2378 |
},
|
| 2379 |
{
|
| 2380 |
"epoch": 0.8225,
|
| 2381 |
+
"grad_norm": 2.470698356628418,
|
| 2382 |
"learning_rate": 7.416666666666666e-06,
|
| 2383 |
+
"loss": 0.0983,
|
| 2384 |
"step": 8225
|
| 2385 |
},
|
| 2386 |
{
|
| 2387 |
"epoch": 0.825,
|
| 2388 |
+
"grad_norm": 3.2774899005889893,
|
| 2389 |
"learning_rate": 7.3125e-06,
|
| 2390 |
+
"loss": 0.085,
|
| 2391 |
"step": 8250
|
| 2392 |
},
|
| 2393 |
{
|
| 2394 |
"epoch": 0.8275,
|
| 2395 |
+
"grad_norm": 2.46321177482605,
|
| 2396 |
"learning_rate": 7.208333333333333e-06,
|
| 2397 |
+
"loss": 0.0916,
|
| 2398 |
"step": 8275
|
| 2399 |
},
|
| 2400 |
{
|
| 2401 |
"epoch": 0.83,
|
| 2402 |
+
"grad_norm": 2.0468456745147705,
|
| 2403 |
"learning_rate": 7.104166666666666e-06,
|
| 2404 |
+
"loss": 0.0951,
|
| 2405 |
"step": 8300
|
| 2406 |
},
|
| 2407 |
{
|
| 2408 |
"epoch": 0.8325,
|
| 2409 |
+
"grad_norm": 3.481823444366455,
|
| 2410 |
"learning_rate": 7e-06,
|
| 2411 |
+
"loss": 0.1368,
|
| 2412 |
"step": 8325
|
| 2413 |
},
|
| 2414 |
{
|
| 2415 |
"epoch": 0.835,
|
| 2416 |
+
"grad_norm": 4.122819423675537,
|
| 2417 |
"learning_rate": 6.895833333333333e-06,
|
| 2418 |
+
"loss": 0.1719,
|
| 2419 |
"step": 8350
|
| 2420 |
},
|
| 2421 |
{
|
| 2422 |
"epoch": 0.8375,
|
| 2423 |
+
"grad_norm": 4.750814914703369,
|
| 2424 |
"learning_rate": 6.791666666666666e-06,
|
| 2425 |
+
"loss": 0.2073,
|
| 2426 |
"step": 8375
|
| 2427 |
},
|
| 2428 |
{
|
| 2429 |
"epoch": 0.84,
|
| 2430 |
+
"grad_norm": 2.771611452102661,
|
| 2431 |
"learning_rate": 6.6875e-06,
|
| 2432 |
+
"loss": 0.1592,
|
| 2433 |
"step": 8400
|
| 2434 |
},
|
| 2435 |
{
|
| 2436 |
"epoch": 0.8425,
|
| 2437 |
+
"grad_norm": 4.536616325378418,
|
| 2438 |
"learning_rate": 6.583333333333333e-06,
|
| 2439 |
+
"loss": 0.1153,
|
| 2440 |
"step": 8425
|
| 2441 |
},
|
| 2442 |
{
|
| 2443 |
"epoch": 0.845,
|
| 2444 |
+
"grad_norm": 2.6289072036743164,
|
| 2445 |
"learning_rate": 6.479166666666666e-06,
|
| 2446 |
"loss": 0.1098,
|
| 2447 |
"step": 8450
|
| 2448 |
},
|
| 2449 |
{
|
| 2450 |
"epoch": 0.8475,
|
| 2451 |
+
"grad_norm": 3.1746957302093506,
|
| 2452 |
"learning_rate": 6.375e-06,
|
| 2453 |
+
"loss": 0.1216,
|
| 2454 |
"step": 8475
|
| 2455 |
},
|
| 2456 |
{
|
| 2457 |
"epoch": 0.85,
|
| 2458 |
+
"grad_norm": 3.029151439666748,
|
| 2459 |
"learning_rate": 6.270833333333333e-06,
|
| 2460 |
+
"loss": 0.1374,
|
| 2461 |
"step": 8500
|
| 2462 |
},
|
| 2463 |
{
|
| 2464 |
"epoch": 0.8525,
|
| 2465 |
+
"grad_norm": 3.8162496089935303,
|
| 2466 |
"learning_rate": 6.166666666666666e-06,
|
| 2467 |
+
"loss": 0.1408,
|
| 2468 |
"step": 8525
|
| 2469 |
},
|
| 2470 |
{
|
| 2471 |
"epoch": 0.855,
|
| 2472 |
+
"grad_norm": 5.568809509277344,
|
| 2473 |
"learning_rate": 6.062499999999999e-06,
|
| 2474 |
+
"loss": 0.1948,
|
| 2475 |
"step": 8550
|
| 2476 |
},
|
| 2477 |
{
|
| 2478 |
"epoch": 0.8575,
|
| 2479 |
+
"grad_norm": 5.649611473083496,
|
| 2480 |
"learning_rate": 5.958333333333333e-06,
|
| 2481 |
+
"loss": 0.3217,
|
| 2482 |
"step": 8575
|
| 2483 |
},
|
| 2484 |
{
|
| 2485 |
"epoch": 0.86,
|
| 2486 |
+
"grad_norm": 7.035798072814941,
|
| 2487 |
"learning_rate": 5.854166666666666e-06,
|
| 2488 |
+
"loss": 0.4378,
|
| 2489 |
"step": 8600
|
| 2490 |
},
|
| 2491 |
{
|
| 2492 |
"epoch": 0.8625,
|
| 2493 |
+
"grad_norm": 7.135374546051025,
|
| 2494 |
"learning_rate": 5.749999999999999e-06,
|
| 2495 |
+
"loss": 0.5134,
|
| 2496 |
"step": 8625
|
| 2497 |
},
|
| 2498 |
{
|
| 2499 |
"epoch": 0.865,
|
| 2500 |
+
"grad_norm": 3.085472345352173,
|
| 2501 |
"learning_rate": 5.645833333333333e-06,
|
| 2502 |
+
"loss": 0.3075,
|
| 2503 |
"step": 8650
|
| 2504 |
},
|
| 2505 |
{
|
| 2506 |
"epoch": 0.8675,
|
| 2507 |
+
"grad_norm": 4.60648250579834,
|
| 2508 |
"learning_rate": 5.541666666666666e-06,
|
| 2509 |
+
"loss": 0.2117,
|
| 2510 |
"step": 8675
|
| 2511 |
},
|
| 2512 |
{
|
| 2513 |
"epoch": 0.87,
|
| 2514 |
+
"grad_norm": 5.39269495010376,
|
| 2515 |
"learning_rate": 5.437499999999999e-06,
|
| 2516 |
+
"loss": 0.2447,
|
| 2517 |
"step": 8700
|
| 2518 |
},
|
| 2519 |
{
|
| 2520 |
"epoch": 0.8725,
|
| 2521 |
+
"grad_norm": 3.4270975589752197,
|
| 2522 |
"learning_rate": 5.333333333333333e-06,
|
| 2523 |
+
"loss": 0.1657,
|
| 2524 |
"step": 8725
|
| 2525 |
},
|
| 2526 |
{
|
| 2527 |
"epoch": 0.875,
|
| 2528 |
+
"grad_norm": 2.6738994121551514,
|
| 2529 |
"learning_rate": 5.2291666666666664e-06,
|
| 2530 |
+
"loss": 0.1087,
|
| 2531 |
"step": 8750
|
| 2532 |
},
|
| 2533 |
{
|
| 2534 |
"epoch": 0.8775,
|
| 2535 |
+
"grad_norm": 2.7085695266723633,
|
| 2536 |
"learning_rate": 5.124999999999999e-06,
|
| 2537 |
+
"loss": 0.1076,
|
| 2538 |
"step": 8775
|
| 2539 |
},
|
| 2540 |
{
|
| 2541 |
"epoch": 0.88,
|
| 2542 |
+
"grad_norm": 3.1228623390197754,
|
| 2543 |
"learning_rate": 5.020833333333333e-06,
|
| 2544 |
+
"loss": 0.1216,
|
| 2545 |
"step": 8800
|
| 2546 |
},
|
| 2547 |
{
|
| 2548 |
"epoch": 0.8825,
|
| 2549 |
+
"grad_norm": 5.188724994659424,
|
| 2550 |
"learning_rate": 4.9166666666666665e-06,
|
| 2551 |
+
"loss": 0.154,
|
| 2552 |
"step": 8825
|
| 2553 |
},
|
| 2554 |
{
|
| 2555 |
"epoch": 0.885,
|
| 2556 |
+
"grad_norm": 3.700319528579712,
|
| 2557 |
"learning_rate": 4.812499999999999e-06,
|
| 2558 |
+
"loss": 0.1491,
|
| 2559 |
"step": 8850
|
| 2560 |
},
|
| 2561 |
{
|
| 2562 |
"epoch": 0.8875,
|
| 2563 |
+
"grad_norm": 3.185534954071045,
|
| 2564 |
"learning_rate": 4.708333333333333e-06,
|
| 2565 |
+
"loss": 0.1517,
|
| 2566 |
"step": 8875
|
| 2567 |
},
|
| 2568 |
{
|
| 2569 |
"epoch": 0.89,
|
| 2570 |
+
"grad_norm": 3.0981645584106445,
|
| 2571 |
"learning_rate": 4.6041666666666665e-06,
|
| 2572 |
+
"loss": 0.1092,
|
| 2573 |
"step": 8900
|
| 2574 |
},
|
| 2575 |
{
|
| 2576 |
"epoch": 0.8925,
|
| 2577 |
+
"grad_norm": 3.850358247756958,
|
| 2578 |
"learning_rate": 4.499999999999999e-06,
|
| 2579 |
+
"loss": 0.107,
|
| 2580 |
"step": 8925
|
| 2581 |
},
|
| 2582 |
{
|
| 2583 |
"epoch": 0.895,
|
| 2584 |
+
"grad_norm": 3.169626235961914,
|
| 2585 |
"learning_rate": 4.395833333333333e-06,
|
| 2586 |
+
"loss": 0.115,
|
| 2587 |
"step": 8950
|
| 2588 |
},
|
| 2589 |
{
|
| 2590 |
"epoch": 0.8975,
|
| 2591 |
+
"grad_norm": 2.9183895587921143,
|
| 2592 |
"learning_rate": 4.2916666666666665e-06,
|
| 2593 |
+
"loss": 0.0914,
|
| 2594 |
"step": 8975
|
| 2595 |
},
|
| 2596 |
{
|
| 2597 |
"epoch": 0.9,
|
| 2598 |
+
"grad_norm": 2.877211570739746,
|
| 2599 |
"learning_rate": 4.187499999999999e-06,
|
| 2600 |
+
"loss": 0.0967,
|
| 2601 |
"step": 9000
|
| 2602 |
},
|
| 2603 |
{
|
| 2604 |
"epoch": 0.9,
|
| 2605 |
+
"eval_loss": 0.3684903383255005,
|
| 2606 |
+
"eval_runtime": 39.0288,
|
| 2607 |
+
"eval_samples_per_second": 43.327,
|
| 2608 |
+
"eval_steps_per_second": 2.716,
|
| 2609 |
+
"eval_wer": 15.136755339078306,
|
| 2610 |
"step": 9000
|
| 2611 |
},
|
| 2612 |
{
|
| 2613 |
"epoch": 0.9025,
|
| 2614 |
+
"grad_norm": 3.196434497833252,
|
| 2615 |
"learning_rate": 4.083333333333333e-06,
|
| 2616 |
"loss": 0.1113,
|
| 2617 |
"step": 9025
|
| 2618 |
},
|
| 2619 |
{
|
| 2620 |
"epoch": 0.905,
|
| 2621 |
+
"grad_norm": 3.862992525100708,
|
| 2622 |
"learning_rate": 3.9791666666666665e-06,
|
| 2623 |
+
"loss": 0.1354,
|
| 2624 |
"step": 9050
|
| 2625 |
},
|
| 2626 |
{
|
| 2627 |
"epoch": 0.9075,
|
| 2628 |
+
"grad_norm": 4.684475898742676,
|
| 2629 |
"learning_rate": 3.874999999999999e-06,
|
| 2630 |
"loss": 0.1739,
|
| 2631 |
"step": 9075
|
| 2632 |
},
|
| 2633 |
{
|
| 2634 |
"epoch": 0.91,
|
| 2635 |
+
"grad_norm": 3.8043711185455322,
|
| 2636 |
"learning_rate": 3.770833333333333e-06,
|
| 2637 |
+
"loss": 0.1725,
|
| 2638 |
"step": 9100
|
| 2639 |
},
|
| 2640 |
{
|
| 2641 |
"epoch": 0.9125,
|
| 2642 |
+
"grad_norm": 3.2041592597961426,
|
| 2643 |
"learning_rate": 3.6666666666666666e-06,
|
| 2644 |
+
"loss": 0.1385,
|
| 2645 |
"step": 9125
|
| 2646 |
},
|
| 2647 |
{
|
| 2648 |
"epoch": 0.915,
|
| 2649 |
+
"grad_norm": 2.448774576187134,
|
| 2650 |
"learning_rate": 3.5624999999999998e-06,
|
| 2651 |
+
"loss": 0.1294,
|
| 2652 |
"step": 9150
|
| 2653 |
},
|
| 2654 |
{
|
| 2655 |
"epoch": 0.9175,
|
| 2656 |
+
"grad_norm": 3.0288240909576416,
|
| 2657 |
"learning_rate": 3.458333333333333e-06,
|
| 2658 |
+
"loss": 0.1252,
|
| 2659 |
"step": 9175
|
| 2660 |
},
|
| 2661 |
{
|
| 2662 |
"epoch": 0.92,
|
| 2663 |
+
"grad_norm": 3.7271015644073486,
|
| 2664 |
"learning_rate": 3.354166666666666e-06,
|
| 2665 |
+
"loss": 0.1297,
|
| 2666 |
"step": 9200
|
| 2667 |
},
|
| 2668 |
{
|
| 2669 |
"epoch": 0.9225,
|
| 2670 |
+
"grad_norm": 3.9993860721588135,
|
| 2671 |
"learning_rate": 3.25e-06,
|
| 2672 |
+
"loss": 0.1503,
|
| 2673 |
"step": 9225
|
| 2674 |
},
|
| 2675 |
{
|
| 2676 |
"epoch": 0.925,
|
| 2677 |
+
"grad_norm": 4.390246391296387,
|
| 2678 |
"learning_rate": 3.145833333333333e-06,
|
| 2679 |
+
"loss": 0.1744,
|
| 2680 |
"step": 9250
|
| 2681 |
},
|
| 2682 |
{
|
| 2683 |
"epoch": 0.9275,
|
| 2684 |
+
"grad_norm": 3.318673849105835,
|
| 2685 |
"learning_rate": 3.041666666666666e-06,
|
| 2686 |
+
"loss": 0.1926,
|
| 2687 |
"step": 9275
|
| 2688 |
},
|
| 2689 |
{
|
| 2690 |
"epoch": 0.93,
|
| 2691 |
+
"grad_norm": 2.6013784408569336,
|
| 2692 |
"learning_rate": 2.9375e-06,
|
| 2693 |
+
"loss": 0.1183,
|
| 2694 |
"step": 9300
|
| 2695 |
},
|
| 2696 |
{
|
| 2697 |
"epoch": 0.9325,
|
| 2698 |
+
"grad_norm": 2.507906675338745,
|
| 2699 |
"learning_rate": 2.833333333333333e-06,
|
| 2700 |
+
"loss": 0.1198,
|
| 2701 |
"step": 9325
|
| 2702 |
},
|
| 2703 |
{
|
| 2704 |
"epoch": 0.935,
|
| 2705 |
+
"grad_norm": 2.4156315326690674,
|
| 2706 |
"learning_rate": 2.7291666666666662e-06,
|
| 2707 |
+
"loss": 0.0964,
|
| 2708 |
"step": 9350
|
| 2709 |
},
|
| 2710 |
{
|
| 2711 |
"epoch": 0.9375,
|
| 2712 |
+
"grad_norm": 3.535658359527588,
|
| 2713 |
"learning_rate": 2.625e-06,
|
| 2714 |
+
"loss": 0.125,
|
| 2715 |
"step": 9375
|
| 2716 |
},
|
| 2717 |
{
|
| 2718 |
"epoch": 0.94,
|
| 2719 |
+
"grad_norm": 3.782447099685669,
|
| 2720 |
"learning_rate": 2.520833333333333e-06,
|
| 2721 |
+
"loss": 0.15,
|
| 2722 |
"step": 9400
|
| 2723 |
},
|
| 2724 |
{
|
| 2725 |
"epoch": 0.9425,
|
| 2726 |
+
"grad_norm": 3.4775750637054443,
|
| 2727 |
"learning_rate": 2.4166666666666663e-06,
|
| 2728 |
"loss": 0.169,
|
| 2729 |
"step": 9425
|
| 2730 |
},
|
| 2731 |
{
|
| 2732 |
"epoch": 0.945,
|
| 2733 |
+
"grad_norm": 2.6878416538238525,
|
| 2734 |
"learning_rate": 2.3125e-06,
|
| 2735 |
+
"loss": 0.2931,
|
| 2736 |
"step": 9450
|
| 2737 |
},
|
| 2738 |
{
|
| 2739 |
"epoch": 0.9475,
|
| 2740 |
+
"grad_norm": 3.1754825115203857,
|
| 2741 |
"learning_rate": 2.208333333333333e-06,
|
| 2742 |
+
"loss": 0.1597,
|
| 2743 |
"step": 9475
|
| 2744 |
},
|
| 2745 |
{
|
| 2746 |
"epoch": 0.95,
|
| 2747 |
+
"grad_norm": 4.0340256690979,
|
| 2748 |
"learning_rate": 2.1041666666666667e-06,
|
| 2749 |
+
"loss": 0.1347,
|
| 2750 |
"step": 9500
|
| 2751 |
},
|
| 2752 |
{
|
| 2753 |
"epoch": 0.9525,
|
| 2754 |
+
"grad_norm": 4.122541904449463,
|
| 2755 |
"learning_rate": 2e-06,
|
| 2756 |
+
"loss": 0.1548,
|
| 2757 |
"step": 9525
|
| 2758 |
},
|
| 2759 |
{
|
| 2760 |
"epoch": 0.955,
|
| 2761 |
+
"grad_norm": 3.8141324520111084,
|
| 2762 |
"learning_rate": 1.8958333333333331e-06,
|
| 2763 |
+
"loss": 0.1713,
|
| 2764 |
"step": 9550
|
| 2765 |
},
|
| 2766 |
{
|
| 2767 |
"epoch": 0.9575,
|
| 2768 |
+
"grad_norm": 4.313726902008057,
|
| 2769 |
"learning_rate": 1.7916666666666665e-06,
|
| 2770 |
+
"loss": 0.1928,
|
| 2771 |
"step": 9575
|
| 2772 |
},
|
| 2773 |
{
|
| 2774 |
"epoch": 0.96,
|
| 2775 |
+
"grad_norm": 3.0914995670318604,
|
| 2776 |
"learning_rate": 1.6874999999999997e-06,
|
| 2777 |
+
"loss": 0.1501,
|
| 2778 |
"step": 9600
|
| 2779 |
},
|
| 2780 |
{
|
| 2781 |
"epoch": 0.9625,
|
| 2782 |
+
"grad_norm": 2.9334723949432373,
|
| 2783 |
"learning_rate": 1.5833333333333331e-06,
|
| 2784 |
+
"loss": 0.1221,
|
| 2785 |
"step": 9625
|
| 2786 |
},
|
| 2787 |
{
|
| 2788 |
"epoch": 0.965,
|
| 2789 |
+
"grad_norm": 2.4183266162872314,
|
| 2790 |
"learning_rate": 1.4791666666666663e-06,
|
| 2791 |
+
"loss": 0.0928,
|
| 2792 |
"step": 9650
|
| 2793 |
},
|
| 2794 |
{
|
| 2795 |
"epoch": 0.9675,
|
| 2796 |
+
"grad_norm": 2.296832323074341,
|
| 2797 |
"learning_rate": 1.375e-06,
|
| 2798 |
"loss": 0.0959,
|
| 2799 |
"step": 9675
|
| 2800 |
},
|
| 2801 |
{
|
| 2802 |
"epoch": 0.97,
|
| 2803 |
+
"grad_norm": 2.63551926612854,
|
| 2804 |
"learning_rate": 1.2708333333333334e-06,
|
| 2805 |
+
"loss": 0.0966,
|
| 2806 |
"step": 9700
|
| 2807 |
},
|
| 2808 |
{
|
| 2809 |
"epoch": 0.9725,
|
| 2810 |
+
"grad_norm": 1.874773621559143,
|
| 2811 |
"learning_rate": 1.1666666666666666e-06,
|
| 2812 |
+
"loss": 0.0958,
|
| 2813 |
"step": 9725
|
| 2814 |
},
|
| 2815 |
{
|
| 2816 |
"epoch": 0.975,
|
| 2817 |
+
"grad_norm": 2.397674322128296,
|
| 2818 |
"learning_rate": 1.0624999999999998e-06,
|
| 2819 |
+
"loss": 0.1009,
|
| 2820 |
"step": 9750
|
| 2821 |
},
|
| 2822 |
{
|
| 2823 |
"epoch": 0.9775,
|
| 2824 |
+
"grad_norm": 3.000075101852417,
|
| 2825 |
"learning_rate": 9.583333333333334e-07,
|
| 2826 |
+
"loss": 0.1095,
|
| 2827 |
"step": 9775
|
| 2828 |
},
|
| 2829 |
{
|
| 2830 |
"epoch": 0.98,
|
| 2831 |
+
"grad_norm": 2.8197531700134277,
|
| 2832 |
"learning_rate": 8.541666666666666e-07,
|
| 2833 |
+
"loss": 0.1535,
|
| 2834 |
"step": 9800
|
| 2835 |
},
|
| 2836 |
{
|
| 2837 |
"epoch": 0.9825,
|
| 2838 |
+
"grad_norm": 3.8931257724761963,
|
| 2839 |
"learning_rate": 7.499999999999999e-07,
|
| 2840 |
+
"loss": 0.1639,
|
| 2841 |
"step": 9825
|
| 2842 |
},
|
| 2843 |
{
|
| 2844 |
"epoch": 0.985,
|
| 2845 |
+
"grad_norm": 3.463376045227051,
|
| 2846 |
"learning_rate": 6.458333333333332e-07,
|
| 2847 |
+
"loss": 0.1494,
|
| 2848 |
"step": 9850
|
| 2849 |
},
|
| 2850 |
{
|
| 2851 |
"epoch": 0.9875,
|
| 2852 |
+
"grad_norm": 2.9888062477111816,
|
| 2853 |
"learning_rate": 5.416666666666666e-07,
|
| 2854 |
+
"loss": 0.1483,
|
| 2855 |
"step": 9875
|
| 2856 |
},
|
| 2857 |
{
|
| 2858 |
"epoch": 0.99,
|
| 2859 |
+
"grad_norm": 7.088611602783203,
|
| 2860 |
"learning_rate": 4.375e-07,
|
| 2861 |
+
"loss": 0.2288,
|
| 2862 |
"step": 9900
|
| 2863 |
},
|
| 2864 |
{
|
| 2865 |
"epoch": 0.9925,
|
| 2866 |
+
"grad_norm": 2.751157283782959,
|
| 2867 |
"learning_rate": 3.333333333333333e-07,
|
| 2868 |
+
"loss": 0.1954,
|
| 2869 |
"step": 9925
|
| 2870 |
},
|
| 2871 |
{
|
| 2872 |
"epoch": 0.995,
|
| 2873 |
+
"grad_norm": 3.0588839054107666,
|
| 2874 |
"learning_rate": 2.2916666666666666e-07,
|
| 2875 |
+
"loss": 0.1068,
|
| 2876 |
"step": 9950
|
| 2877 |
},
|
| 2878 |
{
|
| 2879 |
"epoch": 0.9975,
|
| 2880 |
+
"grad_norm": 2.712277412414551,
|
| 2881 |
"learning_rate": 1.25e-07,
|
| 2882 |
+
"loss": 0.0997,
|
| 2883 |
"step": 9975
|
| 2884 |
},
|
| 2885 |
{
|
| 2886 |
"epoch": 1.0,
|
| 2887 |
+
"grad_norm": 2.309202194213867,
|
| 2888 |
"learning_rate": 2.083333333333333e-08,
|
| 2889 |
+
"loss": 0.1059,
|
| 2890 |
"step": 10000
|
| 2891 |
},
|
| 2892 |
{
|
| 2893 |
"epoch": 1.0,
|
| 2894 |
+
"eval_loss": 0.37193799018859863,
|
| 2895 |
+
"eval_runtime": 39.6288,
|
| 2896 |
+
"eval_samples_per_second": 42.671,
|
| 2897 |
+
"eval_steps_per_second": 2.675,
|
| 2898 |
+
"eval_wer": 14.849506681653555,
|
| 2899 |
"step": 10000
|
| 2900 |
},
|
| 2901 |
{
|
| 2902 |
"epoch": 1.0,
|
| 2903 |
"step": 10000,
|
| 2904 |
"total_flos": 7.8780432384e+18,
|
| 2905 |
+
"train_loss": 0.2454334835767746,
|
| 2906 |
+
"train_runtime": 3974.1996,
|
| 2907 |
+
"train_samples_per_second": 80.519,
|
| 2908 |
+
"train_steps_per_second": 2.516
|
| 2909 |
}
|
| 2910 |
],
|
| 2911 |
"logging_steps": 25,
|