End of training
Browse files- README.md +1 -1
- all_results.json +11 -11
- eval_results.json +6 -6
- runs/Oct26_12-15-41_1df5d157110b/events.out.tfevents.1730026191.1df5d157110b.1093.1 +3 -0
- train_results.json +6 -6
- trainer_state.json +2042 -597
README.md
CHANGED
|
@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# whisper-cv-cy-train-all-plus-other-with-excluded-ft-cv-tts
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [DewiBrynJones/whisper-large-v3-ft-cv-cy-train-all-plus-other-with-excluded](https://huggingface.co/DewiBrynJones/whisper-large-v3-ft-cv-cy-train-all-plus-other-with-excluded) on
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.3615
|
| 21 |
- Wer: 0.2008
|
|
|
|
| 15 |
|
| 16 |
# whisper-cv-cy-train-all-plus-other-with-excluded-ft-cv-tts
|
| 17 |
|
| 18 |
+
This model is a fine-tuned version of [DewiBrynJones/whisper-large-v3-ft-cv-cy-train-all-plus-other-with-excluded](https://huggingface.co/DewiBrynJones/whisper-large-v3-ft-cv-cy-train-all-plus-other-with-excluded) on the DewiBrynJones/commonvoice_cy_tts train main dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.3615
|
| 21 |
- Wer: 0.2008
|
all_results.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 5381,
|
| 6 |
-
"eval_samples_per_second": 2.
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
-
"total_flos":
|
| 10 |
-
"train_loss": 0.
|
| 11 |
-
"train_runtime":
|
| 12 |
"train_samples": 69812,
|
| 13 |
-
"train_samples_per_second": 3.
|
| 14 |
-
"train_steps_per_second": 0.
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.582951420714941,
|
| 3 |
+
"eval_loss": 0.36154302954673767,
|
| 4 |
+
"eval_runtime": 2092.8115,
|
| 5 |
"eval_samples": 5381,
|
| 6 |
+
"eval_samples_per_second": 2.571,
|
| 7 |
+
"eval_steps_per_second": 0.161,
|
| 8 |
+
"eval_wer": 0.2007833313451827,
|
| 9 |
+
"total_flos": 1.0870363781804851e+21,
|
| 10 |
+
"train_loss": 0.0910786449804902,
|
| 11 |
+
"train_runtime": 82447.7266,
|
| 12 |
"train_samples": 69812,
|
| 13 |
+
"train_samples_per_second": 3.881,
|
| 14 |
+
"train_steps_per_second": 0.121
|
| 15 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
"eval_samples": 5381,
|
| 6 |
-
"eval_samples_per_second": 2.
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"eval_wer": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.582951420714941,
|
| 3 |
+
"eval_loss": 0.36154302954673767,
|
| 4 |
+
"eval_runtime": 2092.8115,
|
| 5 |
"eval_samples": 5381,
|
| 6 |
+
"eval_samples_per_second": 2.571,
|
| 7 |
+
"eval_steps_per_second": 0.161,
|
| 8 |
+
"eval_wer": 0.2007833313451827
|
| 9 |
}
|
runs/Oct26_12-15-41_1df5d157110b/events.out.tfevents.1730026191.1df5d157110b.1093.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ab9c05435381ed4f818b43f60c3dd8c5699e923bf859b1855c9823536f4badeb
|
| 3 |
+
size 406
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 69812,
|
| 7 |
-
"train_samples_per_second": 3.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.582951420714941,
|
| 3 |
+
"total_flos": 1.0870363781804851e+21,
|
| 4 |
+
"train_loss": 0.0910786449804902,
|
| 5 |
+
"train_runtime": 82447.7266,
|
| 6 |
"train_samples": 69812,
|
| 7 |
+
"train_samples_per_second": 3.881,
|
| 8 |
+
"train_steps_per_second": 0.121
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1472 +1,2917 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 1000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.011457378551787351,
|
| 13 |
-
"grad_norm": 3.
|
| 14 |
"learning_rate": 5.000000000000001e-07,
|
| 15 |
"loss": 0.5663,
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.022914757103574702,
|
| 20 |
-
"grad_norm": 4.
|
| 21 |
"learning_rate": 1.0000000000000002e-06,
|
| 22 |
-
"loss": 0.
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.034372135655362054,
|
| 27 |
-
"grad_norm": 4.
|
| 28 |
"learning_rate": 1.5e-06,
|
| 29 |
"loss": 0.4423,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.045829514207149404,
|
| 34 |
-
"grad_norm": 2.
|
| 35 |
"learning_rate": 2.0000000000000003e-06,
|
| 36 |
"loss": 0.3852,
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.05728689275893675,
|
| 41 |
-
"grad_norm": 2.
|
| 42 |
"learning_rate": 2.5e-06,
|
| 43 |
"loss": 0.3232,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.06874427131072411,
|
| 48 |
-
"grad_norm": 2.
|
| 49 |
"learning_rate": 3e-06,
|
| 50 |
"loss": 0.3331,
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.08020164986251145,
|
| 55 |
-
"grad_norm": 2.
|
| 56 |
"learning_rate": 3.5e-06,
|
| 57 |
"loss": 0.3308,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.09165902841429881,
|
| 62 |
-
"grad_norm": 2.
|
| 63 |
"learning_rate": 4.000000000000001e-06,
|
| 64 |
"loss": 0.3265,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.10311640696608616,
|
| 69 |
-
"grad_norm": 2.
|
| 70 |
"learning_rate": 4.5e-06,
|
| 71 |
"loss": 0.3006,
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.1145737855178735,
|
| 76 |
-
"grad_norm": 2.
|
| 77 |
"learning_rate": 5e-06,
|
| 78 |
"loss": 0.2915,
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.12603116406966086,
|
| 83 |
-
"grad_norm": 2.
|
| 84 |
"learning_rate": 5.500000000000001e-06,
|
| 85 |
"loss": 0.2929,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.13748854262144822,
|
| 90 |
-
"grad_norm": 2.
|
| 91 |
"learning_rate": 6e-06,
|
| 92 |
"loss": 0.2825,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.14894592117323557,
|
| 97 |
-
"grad_norm": 2.
|
| 98 |
"learning_rate": 6.5000000000000004e-06,
|
| 99 |
"loss": 0.3054,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.1604032997250229,
|
| 104 |
-
"grad_norm": 2.
|
| 105 |
"learning_rate": 7e-06,
|
| 106 |
"loss": 0.2833,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.17186067827681026,
|
| 111 |
-
"grad_norm": 2.
|
| 112 |
"learning_rate": 7.500000000000001e-06,
|
| 113 |
"loss": 0.2796,
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.18331805682859761,
|
| 118 |
-
"grad_norm": 1.
|
| 119 |
"learning_rate": 8.000000000000001e-06,
|
| 120 |
"loss": 0.275,
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"epoch": 0.19477543538038497,
|
| 125 |
-
"grad_norm": 2.
|
| 126 |
"learning_rate": 8.5e-06,
|
| 127 |
"loss": 0.2898,
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.20623281393217233,
|
| 132 |
-
"grad_norm": 2.
|
| 133 |
"learning_rate": 9e-06,
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.21769019248395968,
|
| 139 |
-
"grad_norm": 1.
|
| 140 |
"learning_rate": 9.5e-06,
|
| 141 |
"loss": 0.278,
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 0.229147571035747,
|
| 146 |
-
"grad_norm": 2.
|
| 147 |
"learning_rate": 1e-05,
|
| 148 |
"loss": 0.2702,
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.24060494958753437,
|
| 153 |
-
"grad_norm": 2.
|
| 154 |
-
"learning_rate": 9.
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.2520623281393217,
|
| 160 |
-
"grad_norm": 1.
|
| 161 |
-
"learning_rate": 9.
|
| 162 |
"loss": 0.2534,
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.2635197066911091,
|
| 167 |
-
"grad_norm": 1.
|
| 168 |
-
"learning_rate": 9.
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.27497708524289644,
|
| 174 |
-
"grad_norm": 2.
|
| 175 |
-
"learning_rate": 9.
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.2864344637946838,
|
| 181 |
-
"grad_norm": 2.
|
| 182 |
-
"learning_rate": 9.
|
| 183 |
-
"loss": 0.
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.29789184234647115,
|
| 188 |
-
"grad_norm": 1.
|
| 189 |
-
"learning_rate": 9.
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 0.3093492208982585,
|
| 195 |
-
"grad_norm": 2.
|
| 196 |
-
"learning_rate": 9.
|
| 197 |
-
"loss": 0.
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
"epoch": 0.3208065994500458,
|
| 202 |
-
"grad_norm": 1.
|
| 203 |
-
"learning_rate": 9.
|
| 204 |
-
"loss": 0.
|
| 205 |
"step": 700
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.33226397800183316,
|
| 209 |
-
"grad_norm": 2.
|
| 210 |
-
"learning_rate": 9.
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 725
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.3437213565536205,
|
| 216 |
-
"grad_norm": 2.
|
| 217 |
-
"learning_rate": 9.
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 750
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.3551787351054079,
|
| 223 |
-
"grad_norm": 1.
|
| 224 |
-
"learning_rate": 9.
|
| 225 |
-
"loss": 0.
|
| 226 |
"step": 775
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.36663611365719523,
|
| 230 |
-
"grad_norm": 2.
|
| 231 |
-
"learning_rate": 9.
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 800
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 0.3780934922089826,
|
| 237 |
-
"grad_norm": 2.
|
| 238 |
-
"learning_rate": 9.
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 825
|
| 241 |
},
|
| 242 |
{
|
| 243 |
"epoch": 0.38955087076076994,
|
| 244 |
-
"grad_norm": 1.
|
| 245 |
-
"learning_rate": 9.
|
| 246 |
-
"loss": 0.
|
| 247 |
"step": 850
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"epoch": 0.4010082493125573,
|
| 251 |
-
"grad_norm": 1.
|
| 252 |
-
"learning_rate": 9.
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 875
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.41246562786434465,
|
| 258 |
-
"grad_norm": 1.
|
| 259 |
-
"learning_rate": 9.
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 900
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.423923006416132,
|
| 265 |
-
"grad_norm": 2.
|
| 266 |
-
"learning_rate": 9.
|
| 267 |
-
"loss": 0.
|
| 268 |
"step": 925
|
| 269 |
},
|
| 270 |
{
|
| 271 |
"epoch": 0.43538038496791936,
|
| 272 |
-
"grad_norm": 1.
|
| 273 |
-
"learning_rate":
|
| 274 |
-
"loss": 0.
|
| 275 |
"step": 950
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 0.44683776351970667,
|
| 279 |
-
"grad_norm": 1.
|
| 280 |
-
"learning_rate":
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 975
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 0.458295142071494,
|
| 286 |
-
"grad_norm": 1.
|
| 287 |
-
"learning_rate":
|
| 288 |
-
"loss": 0.
|
| 289 |
"step": 1000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.458295142071494,
|
| 293 |
-
"eval_loss": 0.
|
| 294 |
-
"eval_runtime":
|
| 295 |
-
"eval_samples_per_second": 2.
|
| 296 |
-
"eval_steps_per_second": 0.
|
| 297 |
-
"eval_wer": 0.
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.4697525206232814,
|
| 302 |
-
"grad_norm": 1.
|
| 303 |
-
"learning_rate":
|
| 304 |
-
"loss": 0.
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.48120989917506873,
|
| 309 |
-
"grad_norm": 2.
|
| 310 |
-
"learning_rate":
|
| 311 |
-
"loss": 0.
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 0.4926672777268561,
|
| 316 |
-
"grad_norm": 2.
|
| 317 |
-
"learning_rate":
|
| 318 |
-
"loss": 0.
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.5041246562786434,
|
| 323 |
-
"grad_norm": 2.
|
| 324 |
-
"learning_rate":
|
| 325 |
-
"loss": 0.
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.5155820348304307,
|
| 330 |
-
"grad_norm": 1.
|
| 331 |
-
"learning_rate":
|
| 332 |
-
"loss": 0.
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.5270394133822182,
|
| 337 |
-
"grad_norm": 1.
|
| 338 |
-
"learning_rate":
|
| 339 |
-
"loss": 0.
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.5384967919340055,
|
| 344 |
-
"grad_norm": 1.
|
| 345 |
-
"learning_rate":
|
| 346 |
-
"loss": 0.
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 0.5499541704857929,
|
| 351 |
-
"grad_norm": 1.
|
| 352 |
-
"learning_rate":
|
| 353 |
-
"loss": 0.
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.5614115490375802,
|
| 358 |
-
"grad_norm": 1.
|
| 359 |
-
"learning_rate":
|
| 360 |
-
"loss": 0.
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
"epoch": 0.5728689275893676,
|
| 365 |
-
"grad_norm": 1.
|
| 366 |
-
"learning_rate":
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.5843263061411549,
|
| 372 |
-
"grad_norm": 2.
|
| 373 |
-
"learning_rate":
|
| 374 |
-
"loss": 0.
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.5957836846929423,
|
| 379 |
-
"grad_norm": 1.
|
| 380 |
-
"learning_rate":
|
| 381 |
-
"loss": 0.
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
"epoch": 0.6072410632447296,
|
| 386 |
-
"grad_norm": 2.
|
| 387 |
-
"learning_rate":
|
| 388 |
-
"loss": 0.
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.618698441796517,
|
| 393 |
-
"grad_norm": 1.
|
| 394 |
-
"learning_rate":
|
| 395 |
-
"loss": 0.
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 0.6301558203483043,
|
| 400 |
-
"grad_norm": 2.
|
| 401 |
-
"learning_rate":
|
| 402 |
-
"loss": 0.
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
"epoch": 0.6416131989000916,
|
| 407 |
-
"grad_norm": 1.
|
| 408 |
-
"learning_rate":
|
| 409 |
-
"loss": 0.
|
| 410 |
"step": 1400
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 0.653070577451879,
|
| 414 |
-
"grad_norm": 1.
|
| 415 |
-
"learning_rate":
|
| 416 |
-
"loss": 0.
|
| 417 |
"step": 1425
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.6645279560036663,
|
| 421 |
-
"grad_norm": 1.
|
| 422 |
-
"learning_rate":
|
| 423 |
-
"loss": 0.
|
| 424 |
"step": 1450
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 0.6759853345554537,
|
| 428 |
-
"grad_norm": 1.
|
| 429 |
-
"learning_rate":
|
| 430 |
-
"loss": 0.
|
| 431 |
"step": 1475
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.687442713107241,
|
| 435 |
-
"grad_norm": 1.
|
| 436 |
-
"learning_rate":
|
| 437 |
-
"loss": 0.
|
| 438 |
"step": 1500
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 0.6989000916590284,
|
| 442 |
-
"grad_norm":
|
| 443 |
-
"learning_rate":
|
| 444 |
-
"loss": 0.
|
| 445 |
"step": 1525
|
| 446 |
},
|
| 447 |
{
|
| 448 |
"epoch": 0.7103574702108157,
|
| 449 |
-
"grad_norm": 1.
|
| 450 |
-
"learning_rate":
|
| 451 |
-
"loss": 0.
|
| 452 |
"step": 1550
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 0.7218148487626032,
|
| 456 |
-
"grad_norm": 1.
|
| 457 |
-
"learning_rate":
|
| 458 |
-
"loss": 0.
|
| 459 |
"step": 1575
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.7332722273143905,
|
| 463 |
-
"grad_norm": 1.
|
| 464 |
-
"learning_rate":
|
| 465 |
-
"loss": 0.
|
| 466 |
"step": 1600
|
| 467 |
},
|
| 468 |
{
|
| 469 |
"epoch": 0.7447296058661779,
|
| 470 |
-
"grad_norm": 1.
|
| 471 |
-
"learning_rate":
|
| 472 |
-
"loss": 0.
|
| 473 |
"step": 1625
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 0.7561869844179652,
|
| 477 |
-
"grad_norm": 1.
|
| 478 |
-
"learning_rate":
|
| 479 |
-
"loss": 0.
|
| 480 |
"step": 1650
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 0.7676443629697525,
|
| 484 |
-
"grad_norm": 1.
|
| 485 |
-
"learning_rate":
|
| 486 |
-
"loss": 0.
|
| 487 |
"step": 1675
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 0.7791017415215399,
|
| 491 |
-
"grad_norm": 2.
|
| 492 |
-
"learning_rate":
|
| 493 |
-
"loss": 0.
|
| 494 |
"step": 1700
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 0.7905591200733272,
|
| 498 |
-
"grad_norm": 1.
|
| 499 |
-
"learning_rate":
|
| 500 |
-
"loss": 0.
|
| 501 |
"step": 1725
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.8020164986251146,
|
| 505 |
-
"grad_norm": 1.
|
| 506 |
-
"learning_rate":
|
| 507 |
-
"loss": 0.
|
| 508 |
"step": 1750
|
| 509 |
},
|
| 510 |
{
|
| 511 |
"epoch": 0.8134738771769019,
|
| 512 |
-
"grad_norm": 1.
|
| 513 |
-
"learning_rate":
|
| 514 |
-
"loss": 0.
|
| 515 |
"step": 1775
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 0.8249312557286893,
|
| 519 |
-
"grad_norm": 1.
|
| 520 |
-
"learning_rate":
|
| 521 |
-
"loss": 0.
|
| 522 |
"step": 1800
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 0.8363886342804766,
|
| 526 |
-
"grad_norm": 2.
|
| 527 |
-
"learning_rate":
|
| 528 |
-
"loss": 0.
|
| 529 |
"step": 1825
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 0.847846012832264,
|
| 533 |
-
"grad_norm": 1.
|
| 534 |
-
"learning_rate":
|
| 535 |
-
"loss": 0.
|
| 536 |
"step": 1850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 0.8593033913840513,
|
| 540 |
-
"grad_norm": 1.
|
| 541 |
-
"learning_rate":
|
| 542 |
-
"loss": 0.
|
| 543 |
"step": 1875
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.8707607699358387,
|
| 547 |
-
"grad_norm": 1.
|
| 548 |
-
"learning_rate":
|
| 549 |
-
"loss": 0.
|
| 550 |
"step": 1900
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 0.882218148487626,
|
| 554 |
-
"grad_norm": 1.
|
| 555 |
-
"learning_rate":
|
| 556 |
-
"loss": 0.
|
| 557 |
"step": 1925
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 0.8936755270394133,
|
| 561 |
-
"grad_norm": 1.
|
| 562 |
-
"learning_rate":
|
| 563 |
-
"loss": 0.
|
| 564 |
"step": 1950
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.9051329055912007,
|
| 568 |
-
"grad_norm": 1.
|
| 569 |
-
"learning_rate":
|
| 570 |
-
"loss": 0.
|
| 571 |
"step": 1975
|
| 572 |
},
|
| 573 |
{
|
| 574 |
"epoch": 0.916590284142988,
|
| 575 |
-
"grad_norm": 1.
|
| 576 |
-
"learning_rate":
|
| 577 |
-
"loss": 0.
|
| 578 |
"step": 2000
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 0.916590284142988,
|
| 582 |
-
"eval_loss": 0.
|
| 583 |
-
"eval_runtime":
|
| 584 |
-
"eval_samples_per_second": 2.
|
| 585 |
"eval_steps_per_second": 0.161,
|
| 586 |
-
"eval_wer": 0.
|
| 587 |
"step": 2000
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 0.9280476626947755,
|
| 591 |
-
"grad_norm": 2.
|
| 592 |
-
"learning_rate":
|
| 593 |
-
"loss": 0.
|
| 594 |
"step": 2025
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.9395050412465628,
|
| 598 |
-
"grad_norm": 1.
|
| 599 |
-
"learning_rate":
|
| 600 |
-
"loss": 0.
|
| 601 |
"step": 2050
|
| 602 |
},
|
| 603 |
{
|
| 604 |
"epoch": 0.9509624197983502,
|
| 605 |
-
"grad_norm": 1.
|
| 606 |
-
"learning_rate":
|
| 607 |
-
"loss": 0.
|
| 608 |
"step": 2075
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 0.9624197983501375,
|
| 612 |
-
"grad_norm": 1.
|
| 613 |
-
"learning_rate":
|
| 614 |
-
"loss": 0.
|
| 615 |
"step": 2100
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.9738771769019249,
|
| 619 |
-
"grad_norm": 1.
|
| 620 |
-
"learning_rate":
|
| 621 |
-
"loss": 0.
|
| 622 |
"step": 2125
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 0.9853345554537122,
|
| 626 |
-
"grad_norm": 1.
|
| 627 |
-
"learning_rate":
|
| 628 |
-
"loss": 0.
|
| 629 |
"step": 2150
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 0.9967919340054996,
|
| 633 |
-
"grad_norm": 2.
|
| 634 |
-
"learning_rate":
|
| 635 |
-
"loss": 0.
|
| 636 |
"step": 2175
|
| 637 |
},
|
| 638 |
{
|
| 639 |
"epoch": 1.008249312557287,
|
| 640 |
-
"grad_norm": 1.
|
| 641 |
-
"learning_rate":
|
| 642 |
-
"loss": 0.
|
| 643 |
"step": 2200
|
| 644 |
},
|
| 645 |
{
|
| 646 |
"epoch": 1.0197066911090742,
|
| 647 |
-
"grad_norm": 1.
|
| 648 |
-
"learning_rate":
|
| 649 |
-
"loss": 0.
|
| 650 |
"step": 2225
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 1.0311640696608615,
|
| 654 |
-
"grad_norm": 1.
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 0.
|
| 657 |
"step": 2250
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 1.042621448212649,
|
| 661 |
-
"grad_norm": 1.
|
| 662 |
-
"learning_rate":
|
| 663 |
-
"loss": 0.
|
| 664 |
"step": 2275
|
| 665 |
},
|
| 666 |
{
|
| 667 |
"epoch": 1.0540788267644363,
|
| 668 |
-
"grad_norm": 1.
|
| 669 |
-
"learning_rate":
|
| 670 |
-
"loss": 0.
|
| 671 |
"step": 2300
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 1.0655362053162236,
|
| 675 |
-
"grad_norm": 1.
|
| 676 |
-
"learning_rate":
|
| 677 |
-
"loss": 0.
|
| 678 |
"step": 2325
|
| 679 |
},
|
| 680 |
{
|
| 681 |
"epoch": 1.076993583868011,
|
| 682 |
-
"grad_norm": 1.
|
| 683 |
-
"learning_rate":
|
| 684 |
-
"loss": 0.
|
| 685 |
"step": 2350
|
| 686 |
},
|
| 687 |
{
|
| 688 |
"epoch": 1.0884509624197984,
|
| 689 |
-
"grad_norm": 1.
|
| 690 |
-
"learning_rate":
|
| 691 |
-
"loss": 0.
|
| 692 |
"step": 2375
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 1.0999083409715857,
|
| 696 |
-
"grad_norm": 1.
|
| 697 |
-
"learning_rate":
|
| 698 |
-
"loss": 0.
|
| 699 |
"step": 2400
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 1.111365719523373,
|
| 703 |
-
"grad_norm": 1.
|
| 704 |
-
"learning_rate":
|
| 705 |
-
"loss": 0.
|
| 706 |
"step": 2425
|
| 707 |
},
|
| 708 |
{
|
| 709 |
"epoch": 1.1228230980751603,
|
| 710 |
-
"grad_norm": 1.
|
| 711 |
-
"learning_rate":
|
| 712 |
-
"loss": 0.
|
| 713 |
"step": 2450
|
| 714 |
},
|
| 715 |
{
|
| 716 |
"epoch": 1.1342804766269476,
|
| 717 |
-
"grad_norm": 1.
|
| 718 |
-
"learning_rate":
|
| 719 |
-
"loss": 0.
|
| 720 |
"step": 2475
|
| 721 |
},
|
| 722 |
{
|
| 723 |
"epoch": 1.1457378551787352,
|
| 724 |
-
"grad_norm": 1.
|
| 725 |
-
"learning_rate":
|
| 726 |
-
"loss": 0.
|
| 727 |
"step": 2500
|
| 728 |
},
|
| 729 |
{
|
| 730 |
"epoch": 1.1571952337305225,
|
| 731 |
-
"grad_norm": 1.
|
| 732 |
-
"learning_rate":
|
| 733 |
-
"loss": 0.
|
| 734 |
"step": 2525
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 1.1686526122823098,
|
| 738 |
-
"grad_norm": 1.
|
| 739 |
-
"learning_rate":
|
| 740 |
-
"loss": 0.
|
| 741 |
"step": 2550
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 1.180109990834097,
|
| 745 |
-
"grad_norm": 1.
|
| 746 |
-
"learning_rate":
|
| 747 |
-
"loss": 0.
|
| 748 |
"step": 2575
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 1.1915673693858846,
|
| 752 |
-
"grad_norm": 1.
|
| 753 |
-
"learning_rate":
|
| 754 |
-
"loss": 0.
|
| 755 |
"step": 2600
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 1.203024747937672,
|
| 759 |
-
"grad_norm": 1.
|
| 760 |
-
"learning_rate":
|
| 761 |
-
"loss": 0.
|
| 762 |
"step": 2625
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 1.2144821264894592,
|
| 766 |
-
"grad_norm": 1.
|
| 767 |
-
"learning_rate":
|
| 768 |
-
"loss": 0.
|
| 769 |
"step": 2650
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 1.2259395050412465,
|
| 773 |
-
"grad_norm": 1.
|
| 774 |
-
"learning_rate":
|
| 775 |
-
"loss": 0.
|
| 776 |
"step": 2675
|
| 777 |
},
|
| 778 |
{
|
| 779 |
"epoch": 1.2373968835930338,
|
| 780 |
-
"grad_norm": 1.
|
| 781 |
-
"learning_rate":
|
| 782 |
-
"loss": 0.
|
| 783 |
"step": 2700
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 1.2488542621448213,
|
| 787 |
-
"grad_norm": 1.
|
| 788 |
-
"learning_rate":
|
| 789 |
-
"loss": 0.
|
| 790 |
"step": 2725
|
| 791 |
},
|
| 792 |
{
|
| 793 |
"epoch": 1.2603116406966086,
|
| 794 |
-
"grad_norm": 1.
|
| 795 |
-
"learning_rate":
|
| 796 |
-
"loss": 0.
|
| 797 |
"step": 2750
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 1.271769019248396,
|
| 801 |
-
"grad_norm": 1.
|
| 802 |
-
"learning_rate":
|
| 803 |
-
"loss": 0.
|
| 804 |
"step": 2775
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 1.2832263978001834,
|
| 808 |
-
"grad_norm": 2.
|
| 809 |
-
"learning_rate":
|
| 810 |
-
"loss": 0.
|
| 811 |
"step": 2800
|
| 812 |
},
|
| 813 |
{
|
| 814 |
"epoch": 1.2946837763519707,
|
| 815 |
-
"grad_norm": 1.
|
| 816 |
-
"learning_rate":
|
| 817 |
-
"loss": 0.
|
| 818 |
"step": 2825
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 1.306141154903758,
|
| 822 |
-
"grad_norm": 1.
|
| 823 |
-
"learning_rate":
|
| 824 |
-
"loss": 0.
|
| 825 |
"step": 2850
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 1.3175985334555453,
|
| 829 |
-
"grad_norm": 1.
|
| 830 |
-
"learning_rate":
|
| 831 |
-
"loss": 0.
|
| 832 |
"step": 2875
|
| 833 |
},
|
| 834 |
{
|
| 835 |
"epoch": 1.3290559120073326,
|
| 836 |
-
"grad_norm": 1.
|
| 837 |
-
"learning_rate":
|
| 838 |
-
"loss": 0.
|
| 839 |
"step": 2900
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 1.34051329055912,
|
| 843 |
-
"grad_norm": 1.
|
| 844 |
-
"learning_rate":
|
| 845 |
-
"loss": 0.
|
| 846 |
"step": 2925
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 1.3519706691109075,
|
| 850 |
-
"grad_norm": 1.
|
| 851 |
-
"learning_rate":
|
| 852 |
-
"loss": 0.
|
| 853 |
"step": 2950
|
| 854 |
},
|
| 855 |
{
|
| 856 |
"epoch": 1.3634280476626948,
|
| 857 |
-
"grad_norm": 1.
|
| 858 |
-
"learning_rate":
|
| 859 |
-
"loss": 0.
|
| 860 |
"step": 2975
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 1.374885426214482,
|
| 864 |
-
"grad_norm": 1.
|
| 865 |
-
"learning_rate":
|
| 866 |
-
"loss": 0.
|
| 867 |
"step": 3000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 1.374885426214482,
|
| 871 |
-
"eval_loss": 0.
|
| 872 |
-
"eval_runtime":
|
| 873 |
-
"eval_samples_per_second": 2.
|
| 874 |
-
"eval_steps_per_second": 0.
|
| 875 |
-
"eval_wer": 0.
|
| 876 |
"step": 3000
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 1.3863428047662696,
|
| 880 |
-
"grad_norm": 1.
|
| 881 |
-
"learning_rate":
|
| 882 |
-
"loss": 0.
|
| 883 |
"step": 3025
|
| 884 |
},
|
| 885 |
{
|
| 886 |
"epoch": 1.3978001833180569,
|
| 887 |
-
"grad_norm": 1.
|
| 888 |
-
"learning_rate":
|
| 889 |
-
"loss": 0.
|
| 890 |
"step": 3050
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 1.4092575618698442,
|
| 894 |
-
"grad_norm": 1.
|
| 895 |
-
"learning_rate":
|
| 896 |
-
"loss": 0.
|
| 897 |
"step": 3075
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 1.4207149404216315,
|
| 901 |
-
"grad_norm": 1.
|
| 902 |
-
"learning_rate":
|
| 903 |
-
"loss": 0.
|
| 904 |
"step": 3100
|
| 905 |
},
|
| 906 |
{
|
| 907 |
"epoch": 1.4321723189734188,
|
| 908 |
-
"grad_norm": 1.
|
| 909 |
-
"learning_rate":
|
| 910 |
-
"loss": 0.
|
| 911 |
"step": 3125
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 1.4436296975252063,
|
| 915 |
-
"grad_norm":
|
| 916 |
-
"learning_rate":
|
| 917 |
-
"loss": 0.
|
| 918 |
"step": 3150
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 1.4550870760769936,
|
| 922 |
-
"grad_norm": 1.
|
| 923 |
-
"learning_rate":
|
| 924 |
-
"loss": 0.
|
| 925 |
"step": 3175
|
| 926 |
},
|
| 927 |
{
|
| 928 |
"epoch": 1.466544454628781,
|
| 929 |
-
"grad_norm": 1.
|
| 930 |
-
"learning_rate":
|
| 931 |
-
"loss": 0.
|
| 932 |
"step": 3200
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 1.4780018331805682,
|
| 936 |
-
"grad_norm": 1.
|
| 937 |
-
"learning_rate":
|
| 938 |
-
"loss": 0.
|
| 939 |
"step": 3225
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 1.4894592117323557,
|
| 943 |
-
"grad_norm": 1.
|
| 944 |
-
"learning_rate":
|
| 945 |
-
"loss": 0.
|
| 946 |
"step": 3250
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 1.500916590284143,
|
| 950 |
-
"grad_norm": 1.
|
| 951 |
-
"learning_rate":
|
| 952 |
-
"loss": 0.
|
| 953 |
"step": 3275
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 1.5123739688359303,
|
| 957 |
-
"grad_norm": 1.
|
| 958 |
-
"learning_rate":
|
| 959 |
-
"loss": 0.
|
| 960 |
"step": 3300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 1.5238313473877176,
|
| 964 |
-
"grad_norm": 1.
|
| 965 |
-
"learning_rate":
|
| 966 |
-
"loss": 0.
|
| 967 |
"step": 3325
|
| 968 |
},
|
| 969 |
{
|
| 970 |
"epoch": 1.535288725939505,
|
| 971 |
-
"grad_norm": 1.
|
| 972 |
-
"learning_rate":
|
| 973 |
-
"loss": 0.
|
| 974 |
"step": 3350
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 1.5467461044912922,
|
| 978 |
-
"grad_norm": 1.
|
| 979 |
-
"learning_rate":
|
| 980 |
-
"loss": 0.
|
| 981 |
"step": 3375
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 1.5582034830430798,
|
| 985 |
-
"grad_norm": 1.
|
| 986 |
-
"learning_rate":
|
| 987 |
-
"loss": 0.
|
| 988 |
"step": 3400
|
| 989 |
},
|
| 990 |
{
|
| 991 |
"epoch": 1.569660861594867,
|
| 992 |
-
"grad_norm": 1.
|
| 993 |
-
"learning_rate":
|
| 994 |
-
"loss": 0.
|
| 995 |
"step": 3425
|
| 996 |
},
|
| 997 |
{
|
| 998 |
"epoch": 1.5811182401466546,
|
| 999 |
-
"grad_norm": 1.
|
| 1000 |
-
"learning_rate":
|
| 1001 |
-
"loss": 0.
|
| 1002 |
"step": 3450
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 1.5925756186984419,
|
| 1006 |
-
"grad_norm": 1.
|
| 1007 |
-
"learning_rate":
|
| 1008 |
-
"loss": 0.
|
| 1009 |
"step": 3475
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 1.6040329972502292,
|
| 1013 |
-
"grad_norm": 1.
|
| 1014 |
-
"learning_rate":
|
| 1015 |
-
"loss": 0.
|
| 1016 |
"step": 3500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 1.6154903758020165,
|
| 1020 |
-
"grad_norm": 1.
|
| 1021 |
-
"learning_rate":
|
| 1022 |
-
"loss": 0.
|
| 1023 |
"step": 3525
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 1.6269477543538038,
|
| 1027 |
-
"grad_norm": 1.
|
| 1028 |
-
"learning_rate":
|
| 1029 |
-
"loss": 0.
|
| 1030 |
"step": 3550
|
| 1031 |
},
|
| 1032 |
{
|
| 1033 |
"epoch": 1.638405132905591,
|
| 1034 |
-
"grad_norm": 1.
|
| 1035 |
-
"learning_rate":
|
| 1036 |
-
"loss": 0.
|
| 1037 |
"step": 3575
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 1.6498625114573784,
|
| 1041 |
-
"grad_norm": 1.
|
| 1042 |
-
"learning_rate":
|
| 1043 |
-
"loss": 0.
|
| 1044 |
"step": 3600
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 1.661319890009166,
|
| 1048 |
-
"grad_norm": 1.
|
| 1049 |
-
"learning_rate":
|
| 1050 |
-
"loss": 0.
|
| 1051 |
"step": 3625
|
| 1052 |
},
|
| 1053 |
{
|
| 1054 |
"epoch": 1.6727772685609532,
|
| 1055 |
-
"grad_norm": 1.
|
| 1056 |
-
"learning_rate":
|
| 1057 |
-
"loss": 0.
|
| 1058 |
"step": 3650
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 1.6842346471127407,
|
| 1062 |
-
"grad_norm": 1.
|
| 1063 |
-
"learning_rate":
|
| 1064 |
-
"loss": 0.
|
| 1065 |
"step": 3675
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 1.695692025664528,
|
| 1069 |
-
"grad_norm": 1.
|
| 1070 |
-
"learning_rate":
|
| 1071 |
-
"loss": 0.
|
| 1072 |
"step": 3700
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 1.7071494042163153,
|
| 1076 |
-
"grad_norm": 1.
|
| 1077 |
-
"learning_rate":
|
| 1078 |
-
"loss": 0.
|
| 1079 |
"step": 3725
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 1.7186067827681026,
|
| 1083 |
-
"grad_norm": 1.
|
| 1084 |
-
"learning_rate":
|
| 1085 |
-
"loss": 0.
|
| 1086 |
"step": 3750
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 1.73006416131989,
|
| 1090 |
-
"grad_norm": 1.
|
| 1091 |
-
"learning_rate":
|
| 1092 |
-
"loss": 0.
|
| 1093 |
"step": 3775
|
| 1094 |
},
|
| 1095 |
{
|
| 1096 |
"epoch": 1.7415215398716772,
|
| 1097 |
-
"grad_norm": 1.
|
| 1098 |
-
"learning_rate":
|
| 1099 |
-
"loss": 0.
|
| 1100 |
"step": 3800
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 1.7529789184234648,
|
| 1104 |
-
"grad_norm": 1.
|
| 1105 |
-
"learning_rate":
|
| 1106 |
-
"loss": 0.
|
| 1107 |
"step": 3825
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 1.764436296975252,
|
| 1111 |
-
"grad_norm": 1.
|
| 1112 |
-
"learning_rate":
|
| 1113 |
-
"loss": 0.
|
| 1114 |
"step": 3850
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
"epoch": 1.7758936755270394,
|
| 1118 |
-
"grad_norm": 1.
|
| 1119 |
-
"learning_rate":
|
| 1120 |
-
"loss": 0.
|
| 1121 |
"step": 3875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 1.7873510540788269,
|
| 1125 |
-
"grad_norm":
|
| 1126 |
-
"learning_rate":
|
| 1127 |
-
"loss": 0.
|
| 1128 |
"step": 3900
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 1.7988084326306142,
|
| 1132 |
-
"grad_norm": 1.
|
| 1133 |
-
"learning_rate":
|
| 1134 |
-
"loss": 0.
|
| 1135 |
"step": 3925
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
"epoch": 1.8102658111824015,
|
| 1139 |
-
"grad_norm": 1.
|
| 1140 |
-
"learning_rate":
|
| 1141 |
-
"loss": 0.
|
| 1142 |
"step": 3950
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 1.8217231897341888,
|
| 1146 |
-
"grad_norm": 1.
|
| 1147 |
-
"learning_rate":
|
| 1148 |
-
"loss": 0.
|
| 1149 |
"step": 3975
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 1.833180568285976,
|
| 1153 |
-
"grad_norm": 1.
|
| 1154 |
-
"learning_rate":
|
| 1155 |
-
"loss": 0.
|
| 1156 |
"step": 4000
|
| 1157 |
},
|
| 1158 |
{
|
| 1159 |
"epoch": 1.833180568285976,
|
| 1160 |
-
"eval_loss": 0.
|
| 1161 |
-
"eval_runtime":
|
| 1162 |
-
"eval_samples_per_second": 2.
|
| 1163 |
-
"eval_steps_per_second": 0.
|
| 1164 |
-
"eval_wer": 0.
|
| 1165 |
"step": 4000
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
"epoch": 1.8446379468377634,
|
| 1169 |
-
"grad_norm": 1.
|
| 1170 |
-
"learning_rate":
|
| 1171 |
-
"loss": 0.
|
| 1172 |
"step": 4025
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 1.856095325389551,
|
| 1176 |
-
"grad_norm": 1.
|
| 1177 |
-
"learning_rate":
|
| 1178 |
-
"loss": 0.
|
| 1179 |
"step": 4050
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 1.8675527039413382,
|
| 1183 |
-
"grad_norm": 1.
|
| 1184 |
-
"learning_rate":
|
| 1185 |
-
"loss": 0.
|
| 1186 |
"step": 4075
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 1.8790100824931257,
|
| 1190 |
-
"grad_norm":
|
| 1191 |
-
"learning_rate":
|
| 1192 |
-
"loss": 0.
|
| 1193 |
"step": 4100
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
"epoch": 1.890467461044913,
|
| 1197 |
-
"grad_norm": 1.
|
| 1198 |
-
"learning_rate":
|
| 1199 |
-
"loss": 0.
|
| 1200 |
"step": 4125
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 1.9019248395967003,
|
| 1204 |
-
"grad_norm": 1.
|
| 1205 |
-
"learning_rate":
|
| 1206 |
-
"loss": 0.
|
| 1207 |
"step": 4150
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
"epoch": 1.9133822181484876,
|
| 1211 |
-
"grad_norm": 1.
|
| 1212 |
-
"learning_rate":
|
| 1213 |
-
"loss": 0.
|
| 1214 |
"step": 4175
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 1.924839596700275,
|
| 1218 |
-
"grad_norm": 1.
|
| 1219 |
-
"learning_rate":
|
| 1220 |
-
"loss": 0.
|
| 1221 |
"step": 4200
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 1.9362969752520622,
|
| 1225 |
-
"grad_norm": 1.
|
| 1226 |
-
"learning_rate":
|
| 1227 |
-
"loss": 0.
|
| 1228 |
"step": 4225
|
| 1229 |
},
|
| 1230 |
{
|
| 1231 |
"epoch": 1.9477543538038495,
|
| 1232 |
-
"grad_norm": 1.
|
| 1233 |
-
"learning_rate":
|
| 1234 |
-
"loss": 0.
|
| 1235 |
"step": 4250
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 1.959211732355637,
|
| 1239 |
-
"grad_norm": 1.
|
| 1240 |
-
"learning_rate":
|
| 1241 |
-
"loss": 0.
|
| 1242 |
"step": 4275
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 1.9706691109074244,
|
| 1246 |
-
"grad_norm": 1.
|
| 1247 |
-
"learning_rate":
|
| 1248 |
-
"loss": 0.
|
| 1249 |
"step": 4300
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 1.9821264894592119,
|
| 1253 |
-
"grad_norm": 1.
|
| 1254 |
-
"learning_rate":
|
| 1255 |
-
"loss": 0.
|
| 1256 |
"step": 4325
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 1.9935838680109992,
|
| 1260 |
-
"grad_norm": 1.
|
| 1261 |
-
"learning_rate":
|
| 1262 |
-
"loss": 0.
|
| 1263 |
"step": 4350
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 2.0050412465627865,
|
| 1267 |
-
"grad_norm": 0.
|
| 1268 |
-
"learning_rate":
|
| 1269 |
-
"loss": 0.
|
| 1270 |
"step": 4375
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 2.016498625114574,
|
| 1274 |
-
"grad_norm": 0.
|
| 1275 |
-
"learning_rate":
|
| 1276 |
-
"loss": 0.
|
| 1277 |
"step": 4400
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 2.027956003666361,
|
| 1281 |
-
"grad_norm": 1.
|
| 1282 |
-
"learning_rate":
|
| 1283 |
-
"loss": 0.
|
| 1284 |
"step": 4425
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 2.0394133822181484,
|
| 1288 |
-
"grad_norm": 1.
|
| 1289 |
-
"learning_rate":
|
| 1290 |
-
"loss": 0.
|
| 1291 |
"step": 4450
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 2.0508707607699357,
|
| 1295 |
-
"grad_norm":
|
| 1296 |
-
"learning_rate":
|
| 1297 |
-
"loss": 0.
|
| 1298 |
"step": 4475
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 2.062328139321723,
|
| 1302 |
-
"grad_norm": 1.
|
| 1303 |
-
"learning_rate":
|
| 1304 |
-
"loss": 0.
|
| 1305 |
"step": 4500
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 2.0737855178735107,
|
| 1309 |
-
"grad_norm": 1.
|
| 1310 |
-
"learning_rate":
|
| 1311 |
-
"loss": 0.
|
| 1312 |
"step": 4525
|
| 1313 |
},
|
| 1314 |
{
|
| 1315 |
"epoch": 2.085242896425298,
|
| 1316 |
-
"grad_norm": 1.
|
| 1317 |
-
"learning_rate":
|
| 1318 |
-
"loss": 0.
|
| 1319 |
"step": 4550
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 2.0967002749770853,
|
| 1323 |
-
"grad_norm":
|
| 1324 |
-
"learning_rate":
|
| 1325 |
-
"loss": 0.
|
| 1326 |
"step": 4575
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 2.1081576535288726,
|
| 1330 |
-
"grad_norm": 1.
|
| 1331 |
-
"learning_rate":
|
| 1332 |
-
"loss": 0.
|
| 1333 |
"step": 4600
|
| 1334 |
},
|
| 1335 |
{
|
| 1336 |
"epoch": 2.11961503208066,
|
| 1337 |
-
"grad_norm": 1.
|
| 1338 |
-
"learning_rate":
|
| 1339 |
-
"loss": 0.
|
| 1340 |
"step": 4625
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 2.1310724106324472,
|
| 1344 |
-
"grad_norm": 1.
|
| 1345 |
-
"learning_rate":
|
| 1346 |
-
"loss": 0.
|
| 1347 |
"step": 4650
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 2.1425297891842345,
|
| 1351 |
-
"grad_norm": 1.
|
| 1352 |
-
"learning_rate":
|
| 1353 |
-
"loss": 0.
|
| 1354 |
"step": 4675
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
"epoch": 2.153987167736022,
|
| 1358 |
-
"grad_norm": 0.
|
| 1359 |
-
"learning_rate":
|
| 1360 |
-
"loss": 0.
|
| 1361 |
"step": 4700
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 2.1654445462878096,
|
| 1365 |
-
"grad_norm": 1.
|
| 1366 |
-
"learning_rate":
|
| 1367 |
-
"loss": 0.
|
| 1368 |
"step": 4725
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 2.176901924839597,
|
| 1372 |
-
"grad_norm": 1.
|
| 1373 |
-
"learning_rate": 5.
|
| 1374 |
-
"loss": 0.
|
| 1375 |
"step": 4750
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 2.188359303391384,
|
| 1379 |
-
"grad_norm": 1.
|
| 1380 |
-
"learning_rate": 5.
|
| 1381 |
-
"loss": 0.
|
| 1382 |
"step": 4775
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 2.1998166819431715,
|
| 1386 |
-
"grad_norm":
|
| 1387 |
-
"learning_rate":
|
| 1388 |
-
"loss": 0.
|
| 1389 |
"step": 4800
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 2.211274060494959,
|
| 1393 |
-
"grad_norm": 1.
|
| 1394 |
-
"learning_rate":
|
| 1395 |
-
"loss": 0.
|
| 1396 |
"step": 4825
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 2.222731439046746,
|
| 1400 |
-
"grad_norm": 1.
|
| 1401 |
-
"learning_rate":
|
| 1402 |
-
"loss": 0.
|
| 1403 |
"step": 4850
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 2.2341888175985334,
|
| 1407 |
-
"grad_norm":
|
| 1408 |
-
"learning_rate":
|
| 1409 |
-
"loss": 0.
|
| 1410 |
"step": 4875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 2.2456461961503207,
|
| 1414 |
-
"grad_norm":
|
| 1415 |
-
"learning_rate":
|
| 1416 |
-
"loss": 0.
|
| 1417 |
"step": 4900
|
| 1418 |
},
|
| 1419 |
{
|
| 1420 |
"epoch": 2.257103574702108,
|
| 1421 |
-
"grad_norm": 1.
|
| 1422 |
-
"learning_rate":
|
| 1423 |
-
"loss": 0.
|
| 1424 |
"step": 4925
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 2.2685609532538953,
|
| 1428 |
-
"grad_norm": 1.
|
| 1429 |
-
"learning_rate":
|
| 1430 |
-
"loss": 0.
|
| 1431 |
"step": 4950
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 2.280018331805683,
|
| 1435 |
-
"grad_norm": 1.
|
| 1436 |
-
"learning_rate": 5.
|
| 1437 |
-
"loss": 0.
|
| 1438 |
"step": 4975
|
| 1439 |
},
|
| 1440 |
{
|
| 1441 |
"epoch": 2.2914757103574703,
|
| 1442 |
-
"grad_norm": 1.
|
| 1443 |
-
"learning_rate":
|
| 1444 |
-
"loss": 0.
|
| 1445 |
"step": 5000
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 2.2914757103574703,
|
| 1449 |
-
"eval_loss": 0.
|
| 1450 |
-
"eval_runtime":
|
| 1451 |
-
"eval_samples_per_second": 2.
|
| 1452 |
-
"eval_steps_per_second": 0.
|
| 1453 |
-
"eval_wer": 0.
|
| 1454 |
"step": 5000
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
-
"epoch": 2.
|
| 1458 |
-
"
|
| 1459 |
-
"
|
| 1460 |
-
"
|
| 1461 |
-
"
|
| 1462 |
-
|
| 1463 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1464 |
}
|
| 1465 |
],
|
| 1466 |
"logging_steps": 25,
|
| 1467 |
-
"max_steps":
|
| 1468 |
"num_input_tokens_seen": 0,
|
| 1469 |
-
"num_train_epochs":
|
| 1470 |
"save_steps": 1000,
|
| 1471 |
"stateful_callbacks": {
|
| 1472 |
"TrainerControl": {
|
|
@@ -1480,7 +2925,7 @@
|
|
| 1480 |
"attributes": {}
|
| 1481 |
}
|
| 1482 |
},
|
| 1483 |
-
"total_flos":
|
| 1484 |
"train_batch_size": 16,
|
| 1485 |
"trial_name": null,
|
| 1486 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.582951420714941,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.011457378551787351,
|
| 13 |
+
"grad_norm": 3.87955641746521,
|
| 14 |
"learning_rate": 5.000000000000001e-07,
|
| 15 |
"loss": 0.5663,
|
| 16 |
"step": 25
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.022914757103574702,
|
| 20 |
+
"grad_norm": 4.039402484893799,
|
| 21 |
"learning_rate": 1.0000000000000002e-06,
|
| 22 |
+
"loss": 0.5105,
|
| 23 |
"step": 50
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.034372135655362054,
|
| 27 |
+
"grad_norm": 4.18735933303833,
|
| 28 |
"learning_rate": 1.5e-06,
|
| 29 |
"loss": 0.4423,
|
| 30 |
"step": 75
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.045829514207149404,
|
| 34 |
+
"grad_norm": 2.755176544189453,
|
| 35 |
"learning_rate": 2.0000000000000003e-06,
|
| 36 |
"loss": 0.3852,
|
| 37 |
"step": 100
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.05728689275893675,
|
| 41 |
+
"grad_norm": 2.9501564502716064,
|
| 42 |
"learning_rate": 2.5e-06,
|
| 43 |
"loss": 0.3232,
|
| 44 |
"step": 125
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.06874427131072411,
|
| 48 |
+
"grad_norm": 2.5360376834869385,
|
| 49 |
"learning_rate": 3e-06,
|
| 50 |
"loss": 0.3331,
|
| 51 |
"step": 150
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.08020164986251145,
|
| 55 |
+
"grad_norm": 2.275416612625122,
|
| 56 |
"learning_rate": 3.5e-06,
|
| 57 |
"loss": 0.3308,
|
| 58 |
"step": 175
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.09165902841429881,
|
| 62 |
+
"grad_norm": 2.565553665161133,
|
| 63 |
"learning_rate": 4.000000000000001e-06,
|
| 64 |
"loss": 0.3265,
|
| 65 |
"step": 200
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.10311640696608616,
|
| 69 |
+
"grad_norm": 2.5722615718841553,
|
| 70 |
"learning_rate": 4.5e-06,
|
| 71 |
"loss": 0.3006,
|
| 72 |
"step": 225
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.1145737855178735,
|
| 76 |
+
"grad_norm": 2.477465867996216,
|
| 77 |
"learning_rate": 5e-06,
|
| 78 |
"loss": 0.2915,
|
| 79 |
"step": 250
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.12603116406966086,
|
| 83 |
+
"grad_norm": 2.316929578781128,
|
| 84 |
"learning_rate": 5.500000000000001e-06,
|
| 85 |
"loss": 0.2929,
|
| 86 |
"step": 275
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.13748854262144822,
|
| 90 |
+
"grad_norm": 2.3275156021118164,
|
| 91 |
"learning_rate": 6e-06,
|
| 92 |
"loss": 0.2825,
|
| 93 |
"step": 300
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.14894592117323557,
|
| 97 |
+
"grad_norm": 2.438490629196167,
|
| 98 |
"learning_rate": 6.5000000000000004e-06,
|
| 99 |
"loss": 0.3054,
|
| 100 |
"step": 325
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.1604032997250229,
|
| 104 |
+
"grad_norm": 2.051579236984253,
|
| 105 |
"learning_rate": 7e-06,
|
| 106 |
"loss": 0.2833,
|
| 107 |
"step": 350
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.17186067827681026,
|
| 111 |
+
"grad_norm": 2.419551134109497,
|
| 112 |
"learning_rate": 7.500000000000001e-06,
|
| 113 |
"loss": 0.2796,
|
| 114 |
"step": 375
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.18331805682859761,
|
| 118 |
+
"grad_norm": 1.9953334331512451,
|
| 119 |
"learning_rate": 8.000000000000001e-06,
|
| 120 |
"loss": 0.275,
|
| 121 |
"step": 400
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"epoch": 0.19477543538038497,
|
| 125 |
+
"grad_norm": 2.12074613571167,
|
| 126 |
"learning_rate": 8.5e-06,
|
| 127 |
"loss": 0.2898,
|
| 128 |
"step": 425
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.20623281393217233,
|
| 132 |
+
"grad_norm": 2.4646286964416504,
|
| 133 |
"learning_rate": 9e-06,
|
| 134 |
+
"loss": 0.2645,
|
| 135 |
"step": 450
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.21769019248395968,
|
| 139 |
+
"grad_norm": 1.9933630228042603,
|
| 140 |
"learning_rate": 9.5e-06,
|
| 141 |
"loss": 0.278,
|
| 142 |
"step": 475
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 0.229147571035747,
|
| 146 |
+
"grad_norm": 2.180187940597534,
|
| 147 |
"learning_rate": 1e-05,
|
| 148 |
"loss": 0.2702,
|
| 149 |
"step": 500
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.24060494958753437,
|
| 153 |
+
"grad_norm": 2.0871520042419434,
|
| 154 |
+
"learning_rate": 9.973684210526316e-06,
|
| 155 |
+
"loss": 0.2455,
|
| 156 |
"step": 525
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 0.2520623281393217,
|
| 160 |
+
"grad_norm": 1.7063077688217163,
|
| 161 |
+
"learning_rate": 9.947368421052632e-06,
|
| 162 |
"loss": 0.2534,
|
| 163 |
"step": 550
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.2635197066911091,
|
| 167 |
+
"grad_norm": 1.6936042308807373,
|
| 168 |
+
"learning_rate": 9.921052631578947e-06,
|
| 169 |
+
"loss": 0.2399,
|
| 170 |
"step": 575
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.27497708524289644,
|
| 174 |
+
"grad_norm": 2.111734390258789,
|
| 175 |
+
"learning_rate": 9.894736842105264e-06,
|
| 176 |
+
"loss": 0.2585,
|
| 177 |
"step": 600
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.2864344637946838,
|
| 181 |
+
"grad_norm": 2.468724250793457,
|
| 182 |
+
"learning_rate": 9.868421052631579e-06,
|
| 183 |
+
"loss": 0.2369,
|
| 184 |
"step": 625
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.29789184234647115,
|
| 188 |
+
"grad_norm": 1.8268728256225586,
|
| 189 |
+
"learning_rate": 9.842105263157896e-06,
|
| 190 |
+
"loss": 0.2368,
|
| 191 |
"step": 650
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"epoch": 0.3093492208982585,
|
| 195 |
+
"grad_norm": 2.1481707096099854,
|
| 196 |
+
"learning_rate": 9.815789473684212e-06,
|
| 197 |
+
"loss": 0.236,
|
| 198 |
"step": 675
|
| 199 |
},
|
| 200 |
{
|
| 201 |
"epoch": 0.3208065994500458,
|
| 202 |
+
"grad_norm": 1.7079181671142578,
|
| 203 |
+
"learning_rate": 9.789473684210527e-06,
|
| 204 |
+
"loss": 0.2297,
|
| 205 |
"step": 700
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.33226397800183316,
|
| 209 |
+
"grad_norm": 2.141742467880249,
|
| 210 |
+
"learning_rate": 9.763157894736844e-06,
|
| 211 |
+
"loss": 0.229,
|
| 212 |
"step": 725
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.3437213565536205,
|
| 216 |
+
"grad_norm": 2.0361740589141846,
|
| 217 |
+
"learning_rate": 9.736842105263159e-06,
|
| 218 |
+
"loss": 0.2446,
|
| 219 |
"step": 750
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.3551787351054079,
|
| 223 |
+
"grad_norm": 1.885888934135437,
|
| 224 |
+
"learning_rate": 9.710526315789474e-06,
|
| 225 |
+
"loss": 0.2246,
|
| 226 |
"step": 775
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.36663611365719523,
|
| 230 |
+
"grad_norm": 2.1196212768554688,
|
| 231 |
+
"learning_rate": 9.68421052631579e-06,
|
| 232 |
+
"loss": 0.2343,
|
| 233 |
"step": 800
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"epoch": 0.3780934922089826,
|
| 237 |
+
"grad_norm": 2.0630996227264404,
|
| 238 |
+
"learning_rate": 9.657894736842106e-06,
|
| 239 |
+
"loss": 0.2359,
|
| 240 |
"step": 825
|
| 241 |
},
|
| 242 |
{
|
| 243 |
"epoch": 0.38955087076076994,
|
| 244 |
+
"grad_norm": 1.9666824340820312,
|
| 245 |
+
"learning_rate": 9.631578947368422e-06,
|
| 246 |
+
"loss": 0.2354,
|
| 247 |
"step": 850
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"epoch": 0.4010082493125573,
|
| 251 |
+
"grad_norm": 1.8715167045593262,
|
| 252 |
+
"learning_rate": 9.605263157894737e-06,
|
| 253 |
+
"loss": 0.2133,
|
| 254 |
"step": 875
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.41246562786434465,
|
| 258 |
+
"grad_norm": 1.959590196609497,
|
| 259 |
+
"learning_rate": 9.578947368421054e-06,
|
| 260 |
+
"loss": 0.2202,
|
| 261 |
"step": 900
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.423923006416132,
|
| 265 |
+
"grad_norm": 2.479285955429077,
|
| 266 |
+
"learning_rate": 9.552631578947369e-06,
|
| 267 |
+
"loss": 0.2145,
|
| 268 |
"step": 925
|
| 269 |
},
|
| 270 |
{
|
| 271 |
"epoch": 0.43538038496791936,
|
| 272 |
+
"grad_norm": 1.8657755851745605,
|
| 273 |
+
"learning_rate": 9.526315789473684e-06,
|
| 274 |
+
"loss": 0.2153,
|
| 275 |
"step": 950
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 0.44683776351970667,
|
| 279 |
+
"grad_norm": 1.8653265237808228,
|
| 280 |
+
"learning_rate": 9.5e-06,
|
| 281 |
+
"loss": 0.2136,
|
| 282 |
"step": 975
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 0.458295142071494,
|
| 286 |
+
"grad_norm": 1.9904170036315918,
|
| 287 |
+
"learning_rate": 9.473684210526315e-06,
|
| 288 |
+
"loss": 0.2314,
|
| 289 |
"step": 1000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.458295142071494,
|
| 293 |
+
"eval_loss": 0.25911733508110046,
|
| 294 |
+
"eval_runtime": 2127.3148,
|
| 295 |
+
"eval_samples_per_second": 2.529,
|
| 296 |
+
"eval_steps_per_second": 0.158,
|
| 297 |
+
"eval_wer": 0.19994830808382044,
|
| 298 |
"step": 1000
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.4697525206232814,
|
| 302 |
+
"grad_norm": 1.6627790927886963,
|
| 303 |
+
"learning_rate": 9.447368421052632e-06,
|
| 304 |
+
"loss": 0.2428,
|
| 305 |
"step": 1025
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.48120989917506873,
|
| 309 |
+
"grad_norm": 2.021806240081787,
|
| 310 |
+
"learning_rate": 9.421052631578949e-06,
|
| 311 |
+
"loss": 0.2223,
|
| 312 |
"step": 1050
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 0.4926672777268561,
|
| 316 |
+
"grad_norm": 2.055922746658325,
|
| 317 |
+
"learning_rate": 9.394736842105264e-06,
|
| 318 |
+
"loss": 0.2165,
|
| 319 |
"step": 1075
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.5041246562786434,
|
| 323 |
+
"grad_norm": 2.1315300464630127,
|
| 324 |
+
"learning_rate": 9.36842105263158e-06,
|
| 325 |
+
"loss": 0.2133,
|
| 326 |
"step": 1100
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.5155820348304307,
|
| 330 |
+
"grad_norm": 1.7185808420181274,
|
| 331 |
+
"learning_rate": 9.342105263157895e-06,
|
| 332 |
+
"loss": 0.1915,
|
| 333 |
"step": 1125
|
| 334 |
},
|
| 335 |
{
|
| 336 |
"epoch": 0.5270394133822182,
|
| 337 |
+
"grad_norm": 1.6289584636688232,
|
| 338 |
+
"learning_rate": 9.315789473684212e-06,
|
| 339 |
+
"loss": 0.2001,
|
| 340 |
"step": 1150
|
| 341 |
},
|
| 342 |
{
|
| 343 |
"epoch": 0.5384967919340055,
|
| 344 |
+
"grad_norm": 1.6599018573760986,
|
| 345 |
+
"learning_rate": 9.289473684210527e-06,
|
| 346 |
+
"loss": 0.1922,
|
| 347 |
"step": 1175
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 0.5499541704857929,
|
| 351 |
+
"grad_norm": 1.7362765073776245,
|
| 352 |
+
"learning_rate": 9.263157894736842e-06,
|
| 353 |
+
"loss": 0.2047,
|
| 354 |
"step": 1200
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.5614115490375802,
|
| 358 |
+
"grad_norm": 1.9757411479949951,
|
| 359 |
+
"learning_rate": 9.236842105263159e-06,
|
| 360 |
+
"loss": 0.2059,
|
| 361 |
"step": 1225
|
| 362 |
},
|
| 363 |
{
|
| 364 |
"epoch": 0.5728689275893676,
|
| 365 |
+
"grad_norm": 1.9076176881790161,
|
| 366 |
+
"learning_rate": 9.210526315789474e-06,
|
| 367 |
+
"loss": 0.2225,
|
| 368 |
"step": 1250
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.5843263061411549,
|
| 372 |
+
"grad_norm": 2.0330986976623535,
|
| 373 |
+
"learning_rate": 9.18421052631579e-06,
|
| 374 |
+
"loss": 0.2159,
|
| 375 |
"step": 1275
|
| 376 |
},
|
| 377 |
{
|
| 378 |
"epoch": 0.5957836846929423,
|
| 379 |
+
"grad_norm": 1.8117249011993408,
|
| 380 |
+
"learning_rate": 9.157894736842105e-06,
|
| 381 |
+
"loss": 0.2049,
|
| 382 |
"step": 1300
|
| 383 |
},
|
| 384 |
{
|
| 385 |
"epoch": 0.6072410632447296,
|
| 386 |
+
"grad_norm": 2.2551121711730957,
|
| 387 |
+
"learning_rate": 9.131578947368422e-06,
|
| 388 |
+
"loss": 0.2128,
|
| 389 |
"step": 1325
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.618698441796517,
|
| 393 |
+
"grad_norm": 1.7910491228103638,
|
| 394 |
+
"learning_rate": 9.105263157894739e-06,
|
| 395 |
+
"loss": 0.1951,
|
| 396 |
"step": 1350
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 0.6301558203483043,
|
| 400 |
+
"grad_norm": 2.0170273780822754,
|
| 401 |
+
"learning_rate": 9.078947368421054e-06,
|
| 402 |
+
"loss": 0.1945,
|
| 403 |
"step": 1375
|
| 404 |
},
|
| 405 |
{
|
| 406 |
"epoch": 0.6416131989000916,
|
| 407 |
+
"grad_norm": 1.7710468769073486,
|
| 408 |
+
"learning_rate": 9.05263157894737e-06,
|
| 409 |
+
"loss": 0.1983,
|
| 410 |
"step": 1400
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 0.653070577451879,
|
| 414 |
+
"grad_norm": 1.4496461153030396,
|
| 415 |
+
"learning_rate": 9.026315789473685e-06,
|
| 416 |
+
"loss": 0.1827,
|
| 417 |
"step": 1425
|
| 418 |
},
|
| 419 |
{
|
| 420 |
"epoch": 0.6645279560036663,
|
| 421 |
+
"grad_norm": 1.5348992347717285,
|
| 422 |
+
"learning_rate": 9e-06,
|
| 423 |
+
"loss": 0.2059,
|
| 424 |
"step": 1450
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 0.6759853345554537,
|
| 428 |
+
"grad_norm": 1.9110026359558105,
|
| 429 |
+
"learning_rate": 8.973684210526317e-06,
|
| 430 |
+
"loss": 0.1965,
|
| 431 |
"step": 1475
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 0.687442713107241,
|
| 435 |
+
"grad_norm": 1.5054250955581665,
|
| 436 |
+
"learning_rate": 8.947368421052632e-06,
|
| 437 |
+
"loss": 0.1805,
|
| 438 |
"step": 1500
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 0.6989000916590284,
|
| 442 |
+
"grad_norm": 1.9980241060256958,
|
| 443 |
+
"learning_rate": 8.921052631578949e-06,
|
| 444 |
+
"loss": 0.1976,
|
| 445 |
"step": 1525
|
| 446 |
},
|
| 447 |
{
|
| 448 |
"epoch": 0.7103574702108157,
|
| 449 |
+
"grad_norm": 1.8609012365341187,
|
| 450 |
+
"learning_rate": 8.894736842105264e-06,
|
| 451 |
+
"loss": 0.1897,
|
| 452 |
"step": 1550
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 0.7218148487626032,
|
| 456 |
+
"grad_norm": 1.942555546760559,
|
| 457 |
+
"learning_rate": 8.86842105263158e-06,
|
| 458 |
+
"loss": 0.1947,
|
| 459 |
"step": 1575
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.7332722273143905,
|
| 463 |
+
"grad_norm": 1.7984477281570435,
|
| 464 |
+
"learning_rate": 8.842105263157895e-06,
|
| 465 |
+
"loss": 0.1923,
|
| 466 |
"step": 1600
|
| 467 |
},
|
| 468 |
{
|
| 469 |
"epoch": 0.7447296058661779,
|
| 470 |
+
"grad_norm": 1.7936090230941772,
|
| 471 |
+
"learning_rate": 8.81578947368421e-06,
|
| 472 |
+
"loss": 0.1748,
|
| 473 |
"step": 1625
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 0.7561869844179652,
|
| 477 |
+
"grad_norm": 1.8037903308868408,
|
| 478 |
+
"learning_rate": 8.789473684210527e-06,
|
| 479 |
+
"loss": 0.1803,
|
| 480 |
"step": 1650
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 0.7676443629697525,
|
| 484 |
+
"grad_norm": 1.7054656744003296,
|
| 485 |
+
"learning_rate": 8.763157894736842e-06,
|
| 486 |
+
"loss": 0.2022,
|
| 487 |
"step": 1675
|
| 488 |
},
|
| 489 |
{
|
| 490 |
"epoch": 0.7791017415215399,
|
| 491 |
+
"grad_norm": 2.1285829544067383,
|
| 492 |
+
"learning_rate": 8.736842105263158e-06,
|
| 493 |
+
"loss": 0.197,
|
| 494 |
"step": 1700
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 0.7905591200733272,
|
| 498 |
+
"grad_norm": 1.6678701639175415,
|
| 499 |
+
"learning_rate": 8.710526315789475e-06,
|
| 500 |
+
"loss": 0.1834,
|
| 501 |
"step": 1725
|
| 502 |
},
|
| 503 |
{
|
| 504 |
"epoch": 0.8020164986251146,
|
| 505 |
+
"grad_norm": 1.7868952751159668,
|
| 506 |
+
"learning_rate": 8.68421052631579e-06,
|
| 507 |
+
"loss": 0.171,
|
| 508 |
"step": 1750
|
| 509 |
},
|
| 510 |
{
|
| 511 |
"epoch": 0.8134738771769019,
|
| 512 |
+
"grad_norm": 1.9292125701904297,
|
| 513 |
+
"learning_rate": 8.657894736842107e-06,
|
| 514 |
+
"loss": 0.1744,
|
| 515 |
"step": 1775
|
| 516 |
},
|
| 517 |
{
|
| 518 |
"epoch": 0.8249312557286893,
|
| 519 |
+
"grad_norm": 1.4363629817962646,
|
| 520 |
+
"learning_rate": 8.631578947368422e-06,
|
| 521 |
+
"loss": 0.1647,
|
| 522 |
"step": 1800
|
| 523 |
},
|
| 524 |
{
|
| 525 |
"epoch": 0.8363886342804766,
|
| 526 |
+
"grad_norm": 2.056227445602417,
|
| 527 |
+
"learning_rate": 8.605263157894738e-06,
|
| 528 |
+
"loss": 0.1775,
|
| 529 |
"step": 1825
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 0.847846012832264,
|
| 533 |
+
"grad_norm": 1.6606287956237793,
|
| 534 |
+
"learning_rate": 8.578947368421053e-06,
|
| 535 |
+
"loss": 0.1682,
|
| 536 |
"step": 1850
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 0.8593033913840513,
|
| 540 |
+
"grad_norm": 1.817449927330017,
|
| 541 |
+
"learning_rate": 8.552631578947368e-06,
|
| 542 |
+
"loss": 0.1891,
|
| 543 |
"step": 1875
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 0.8707607699358387,
|
| 547 |
+
"grad_norm": 1.7576777935028076,
|
| 548 |
+
"learning_rate": 8.526315789473685e-06,
|
| 549 |
+
"loss": 0.1777,
|
| 550 |
"step": 1900
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 0.882218148487626,
|
| 554 |
+
"grad_norm": 1.8435312509536743,
|
| 555 |
+
"learning_rate": 8.5e-06,
|
| 556 |
+
"loss": 0.1969,
|
| 557 |
"step": 1925
|
| 558 |
},
|
| 559 |
{
|
| 560 |
"epoch": 0.8936755270394133,
|
| 561 |
+
"grad_norm": 1.6576586961746216,
|
| 562 |
+
"learning_rate": 8.473684210526317e-06,
|
| 563 |
+
"loss": 0.189,
|
| 564 |
"step": 1950
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.9051329055912007,
|
| 568 |
+
"grad_norm": 1.918747901916504,
|
| 569 |
+
"learning_rate": 8.447368421052632e-06,
|
| 570 |
+
"loss": 0.1663,
|
| 571 |
"step": 1975
|
| 572 |
},
|
| 573 |
{
|
| 574 |
"epoch": 0.916590284142988,
|
| 575 |
+
"grad_norm": 1.894875407218933,
|
| 576 |
+
"learning_rate": 8.421052631578948e-06,
|
| 577 |
+
"loss": 0.1819,
|
| 578 |
"step": 2000
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 0.916590284142988,
|
| 582 |
+
"eval_loss": 0.258757084608078,
|
| 583 |
+
"eval_runtime": 2096.7645,
|
| 584 |
+
"eval_samples_per_second": 2.566,
|
| 585 |
"eval_steps_per_second": 0.161,
|
| 586 |
+
"eval_wer": 0.20462046204620463,
|
| 587 |
"step": 2000
|
| 588 |
},
|
| 589 |
{
|
| 590 |
"epoch": 0.9280476626947755,
|
| 591 |
+
"grad_norm": 2.047499895095825,
|
| 592 |
+
"learning_rate": 8.394736842105263e-06,
|
| 593 |
+
"loss": 0.1804,
|
| 594 |
"step": 2025
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.9395050412465628,
|
| 598 |
+
"grad_norm": 1.6285953521728516,
|
| 599 |
+
"learning_rate": 8.36842105263158e-06,
|
| 600 |
+
"loss": 0.1771,
|
| 601 |
"step": 2050
|
| 602 |
},
|
| 603 |
{
|
| 604 |
"epoch": 0.9509624197983502,
|
| 605 |
+
"grad_norm": 1.5991675853729248,
|
| 606 |
+
"learning_rate": 8.342105263157897e-06,
|
| 607 |
+
"loss": 0.1848,
|
| 608 |
"step": 2075
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 0.9624197983501375,
|
| 612 |
+
"grad_norm": 1.5647332668304443,
|
| 613 |
+
"learning_rate": 8.315789473684212e-06,
|
| 614 |
+
"loss": 0.1672,
|
| 615 |
"step": 2100
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 0.9738771769019249,
|
| 619 |
+
"grad_norm": 1.4593747854232788,
|
| 620 |
+
"learning_rate": 8.289473684210526e-06,
|
| 621 |
+
"loss": 0.177,
|
| 622 |
"step": 2125
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 0.9853345554537122,
|
| 626 |
+
"grad_norm": 1.270159363746643,
|
| 627 |
+
"learning_rate": 8.263157894736843e-06,
|
| 628 |
+
"loss": 0.1727,
|
| 629 |
"step": 2150
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 0.9967919340054996,
|
| 633 |
+
"grad_norm": 2.3153159618377686,
|
| 634 |
+
"learning_rate": 8.236842105263158e-06,
|
| 635 |
+
"loss": 0.1712,
|
| 636 |
"step": 2175
|
| 637 |
},
|
| 638 |
{
|
| 639 |
"epoch": 1.008249312557287,
|
| 640 |
+
"grad_norm": 1.4837404489517212,
|
| 641 |
+
"learning_rate": 8.210526315789475e-06,
|
| 642 |
+
"loss": 0.1186,
|
| 643 |
"step": 2200
|
| 644 |
},
|
| 645 |
{
|
| 646 |
"epoch": 1.0197066911090742,
|
| 647 |
+
"grad_norm": 1.4995671510696411,
|
| 648 |
+
"learning_rate": 8.18421052631579e-06,
|
| 649 |
+
"loss": 0.115,
|
| 650 |
"step": 2225
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 1.0311640696608615,
|
| 654 |
+
"grad_norm": 1.463578224182129,
|
| 655 |
+
"learning_rate": 8.157894736842106e-06,
|
| 656 |
+
"loss": 0.1056,
|
| 657 |
"step": 2250
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 1.042621448212649,
|
| 661 |
+
"grad_norm": 1.2228025197982788,
|
| 662 |
+
"learning_rate": 8.131578947368421e-06,
|
| 663 |
+
"loss": 0.0989,
|
| 664 |
"step": 2275
|
| 665 |
},
|
| 666 |
{
|
| 667 |
"epoch": 1.0540788267644363,
|
| 668 |
+
"grad_norm": 1.299790620803833,
|
| 669 |
+
"learning_rate": 8.105263157894736e-06,
|
| 670 |
+
"loss": 0.0986,
|
| 671 |
"step": 2300
|
| 672 |
},
|
| 673 |
{
|
| 674 |
"epoch": 1.0655362053162236,
|
| 675 |
+
"grad_norm": 1.5152562856674194,
|
| 676 |
+
"learning_rate": 8.078947368421053e-06,
|
| 677 |
+
"loss": 0.1098,
|
| 678 |
"step": 2325
|
| 679 |
},
|
| 680 |
{
|
| 681 |
"epoch": 1.076993583868011,
|
| 682 |
+
"grad_norm": 1.2980128526687622,
|
| 683 |
+
"learning_rate": 8.052631578947368e-06,
|
| 684 |
+
"loss": 0.1122,
|
| 685 |
"step": 2350
|
| 686 |
},
|
| 687 |
{
|
| 688 |
"epoch": 1.0884509624197984,
|
| 689 |
+
"grad_norm": 1.4537500143051147,
|
| 690 |
+
"learning_rate": 8.026315789473685e-06,
|
| 691 |
+
"loss": 0.1119,
|
| 692 |
"step": 2375
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 1.0999083409715857,
|
| 696 |
+
"grad_norm": 1.2441452741622925,
|
| 697 |
+
"learning_rate": 8.000000000000001e-06,
|
| 698 |
+
"loss": 0.1188,
|
| 699 |
"step": 2400
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 1.111365719523373,
|
| 703 |
+
"grad_norm": 1.695495367050171,
|
| 704 |
+
"learning_rate": 7.973684210526316e-06,
|
| 705 |
+
"loss": 0.1034,
|
| 706 |
"step": 2425
|
| 707 |
},
|
| 708 |
{
|
| 709 |
"epoch": 1.1228230980751603,
|
| 710 |
+
"grad_norm": 1.8364770412445068,
|
| 711 |
+
"learning_rate": 7.947368421052633e-06,
|
| 712 |
+
"loss": 0.1145,
|
| 713 |
"step": 2450
|
| 714 |
},
|
| 715 |
{
|
| 716 |
"epoch": 1.1342804766269476,
|
| 717 |
+
"grad_norm": 1.4711062908172607,
|
| 718 |
+
"learning_rate": 7.921052631578948e-06,
|
| 719 |
+
"loss": 0.117,
|
| 720 |
"step": 2475
|
| 721 |
},
|
| 722 |
{
|
| 723 |
"epoch": 1.1457378551787352,
|
| 724 |
+
"grad_norm": 1.4098833799362183,
|
| 725 |
+
"learning_rate": 7.894736842105265e-06,
|
| 726 |
+
"loss": 0.1082,
|
| 727 |
"step": 2500
|
| 728 |
},
|
| 729 |
{
|
| 730 |
"epoch": 1.1571952337305225,
|
| 731 |
+
"grad_norm": 1.5191541910171509,
|
| 732 |
+
"learning_rate": 7.86842105263158e-06,
|
| 733 |
+
"loss": 0.1089,
|
| 734 |
"step": 2525
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 1.1686526122823098,
|
| 738 |
+
"grad_norm": 1.5274536609649658,
|
| 739 |
+
"learning_rate": 7.842105263157895e-06,
|
| 740 |
+
"loss": 0.1009,
|
| 741 |
"step": 2550
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 1.180109990834097,
|
| 745 |
+
"grad_norm": 1.2877765893936157,
|
| 746 |
+
"learning_rate": 7.815789473684211e-06,
|
| 747 |
+
"loss": 0.1119,
|
| 748 |
"step": 2575
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 1.1915673693858846,
|
| 752 |
+
"grad_norm": 1.1055575609207153,
|
| 753 |
+
"learning_rate": 7.789473684210526e-06,
|
| 754 |
+
"loss": 0.1053,
|
| 755 |
"step": 2600
|
| 756 |
},
|
| 757 |
{
|
| 758 |
"epoch": 1.203024747937672,
|
| 759 |
+
"grad_norm": 1.4493708610534668,
|
| 760 |
+
"learning_rate": 7.763157894736843e-06,
|
| 761 |
+
"loss": 0.1152,
|
| 762 |
"step": 2625
|
| 763 |
},
|
| 764 |
{
|
| 765 |
"epoch": 1.2144821264894592,
|
| 766 |
+
"grad_norm": 1.4660484790802002,
|
| 767 |
+
"learning_rate": 7.736842105263158e-06,
|
| 768 |
+
"loss": 0.1114,
|
| 769 |
"step": 2650
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 1.2259395050412465,
|
| 773 |
+
"grad_norm": 1.2968230247497559,
|
| 774 |
+
"learning_rate": 7.710526315789474e-06,
|
| 775 |
+
"loss": 0.1141,
|
| 776 |
"step": 2675
|
| 777 |
},
|
| 778 |
{
|
| 779 |
"epoch": 1.2373968835930338,
|
| 780 |
+
"grad_norm": 1.7599893808364868,
|
| 781 |
+
"learning_rate": 7.68421052631579e-06,
|
| 782 |
+
"loss": 0.1231,
|
| 783 |
"step": 2700
|
| 784 |
},
|
| 785 |
{
|
| 786 |
"epoch": 1.2488542621448213,
|
| 787 |
+
"grad_norm": 1.1863300800323486,
|
| 788 |
+
"learning_rate": 7.657894736842106e-06,
|
| 789 |
+
"loss": 0.1141,
|
| 790 |
"step": 2725
|
| 791 |
},
|
| 792 |
{
|
| 793 |
"epoch": 1.2603116406966086,
|
| 794 |
+
"grad_norm": 1.7122820615768433,
|
| 795 |
+
"learning_rate": 7.631578947368423e-06,
|
| 796 |
+
"loss": 0.1158,
|
| 797 |
"step": 2750
|
| 798 |
},
|
| 799 |
{
|
| 800 |
"epoch": 1.271769019248396,
|
| 801 |
+
"grad_norm": 1.3090041875839233,
|
| 802 |
+
"learning_rate": 7.605263157894738e-06,
|
| 803 |
+
"loss": 0.1077,
|
| 804 |
"step": 2775
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 1.2832263978001834,
|
| 808 |
+
"grad_norm": 2.139279365539551,
|
| 809 |
+
"learning_rate": 7.578947368421054e-06,
|
| 810 |
+
"loss": 0.1202,
|
| 811 |
"step": 2800
|
| 812 |
},
|
| 813 |
{
|
| 814 |
"epoch": 1.2946837763519707,
|
| 815 |
+
"grad_norm": 1.7823379039764404,
|
| 816 |
+
"learning_rate": 7.552631578947369e-06,
|
| 817 |
+
"loss": 0.1138,
|
| 818 |
"step": 2825
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 1.306141154903758,
|
| 822 |
+
"grad_norm": 1.7470473051071167,
|
| 823 |
+
"learning_rate": 7.526315789473685e-06,
|
| 824 |
+
"loss": 0.1176,
|
| 825 |
"step": 2850
|
| 826 |
},
|
| 827 |
{
|
| 828 |
"epoch": 1.3175985334555453,
|
| 829 |
+
"grad_norm": 1.5477886199951172,
|
| 830 |
+
"learning_rate": 7.500000000000001e-06,
|
| 831 |
+
"loss": 0.1132,
|
| 832 |
"step": 2875
|
| 833 |
},
|
| 834 |
{
|
| 835 |
"epoch": 1.3290559120073326,
|
| 836 |
+
"grad_norm": 1.5441957712173462,
|
| 837 |
+
"learning_rate": 7.473684210526316e-06,
|
| 838 |
+
"loss": 0.1103,
|
| 839 |
"step": 2900
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 1.34051329055912,
|
| 843 |
+
"grad_norm": 1.373345136642456,
|
| 844 |
+
"learning_rate": 7.447368421052632e-06,
|
| 845 |
+
"loss": 0.1025,
|
| 846 |
"step": 2925
|
| 847 |
},
|
| 848 |
{
|
| 849 |
"epoch": 1.3519706691109075,
|
| 850 |
+
"grad_norm": 1.475250005722046,
|
| 851 |
+
"learning_rate": 7.421052631578948e-06,
|
| 852 |
+
"loss": 0.1087,
|
| 853 |
"step": 2950
|
| 854 |
},
|
| 855 |
{
|
| 856 |
"epoch": 1.3634280476626948,
|
| 857 |
+
"grad_norm": 1.5169392824172974,
|
| 858 |
+
"learning_rate": 7.3947368421052635e-06,
|
| 859 |
+
"loss": 0.1074,
|
| 860 |
"step": 2975
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 1.374885426214482,
|
| 864 |
+
"grad_norm": 1.5435885190963745,
|
| 865 |
+
"learning_rate": 7.368421052631579e-06,
|
| 866 |
+
"loss": 0.1027,
|
| 867 |
"step": 3000
|
| 868 |
},
|
| 869 |
{
|
| 870 |
"epoch": 1.374885426214482,
|
| 871 |
+
"eval_loss": 0.2679824233055115,
|
| 872 |
+
"eval_runtime": 2105.0226,
|
| 873 |
+
"eval_samples_per_second": 2.556,
|
| 874 |
+
"eval_steps_per_second": 0.16,
|
| 875 |
+
"eval_wer": 0.2021352737683407,
|
| 876 |
"step": 3000
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 1.3863428047662696,
|
| 880 |
+
"grad_norm": 1.634520173072815,
|
| 881 |
+
"learning_rate": 7.342105263157895e-06,
|
| 882 |
+
"loss": 0.1068,
|
| 883 |
"step": 3025
|
| 884 |
},
|
| 885 |
{
|
| 886 |
"epoch": 1.3978001833180569,
|
| 887 |
+
"grad_norm": 1.453640103340149,
|
| 888 |
+
"learning_rate": 7.315789473684212e-06,
|
| 889 |
+
"loss": 0.1072,
|
| 890 |
"step": 3050
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 1.4092575618698442,
|
| 894 |
+
"grad_norm": 1.805000901222229,
|
| 895 |
+
"learning_rate": 7.289473684210528e-06,
|
| 896 |
+
"loss": 0.1127,
|
| 897 |
"step": 3075
|
| 898 |
},
|
| 899 |
{
|
| 900 |
"epoch": 1.4207149404216315,
|
| 901 |
+
"grad_norm": 1.4315859079360962,
|
| 902 |
+
"learning_rate": 7.263157894736843e-06,
|
| 903 |
+
"loss": 0.1037,
|
| 904 |
"step": 3100
|
| 905 |
},
|
| 906 |
{
|
| 907 |
"epoch": 1.4321723189734188,
|
| 908 |
+
"grad_norm": 1.9786596298217773,
|
| 909 |
+
"learning_rate": 7.236842105263158e-06,
|
| 910 |
+
"loss": 0.1022,
|
| 911 |
"step": 3125
|
| 912 |
},
|
| 913 |
{
|
| 914 |
"epoch": 1.4436296975252063,
|
| 915 |
+
"grad_norm": 2.0479092597961426,
|
| 916 |
+
"learning_rate": 7.210526315789474e-06,
|
| 917 |
+
"loss": 0.1108,
|
| 918 |
"step": 3150
|
| 919 |
},
|
| 920 |
{
|
| 921 |
"epoch": 1.4550870760769936,
|
| 922 |
+
"grad_norm": 1.1353752613067627,
|
| 923 |
+
"learning_rate": 7.18421052631579e-06,
|
| 924 |
+
"loss": 0.0928,
|
| 925 |
"step": 3175
|
| 926 |
},
|
| 927 |
{
|
| 928 |
"epoch": 1.466544454628781,
|
| 929 |
+
"grad_norm": 1.322737216949463,
|
| 930 |
+
"learning_rate": 7.157894736842106e-06,
|
| 931 |
+
"loss": 0.1122,
|
| 932 |
"step": 3200
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 1.4780018331805682,
|
| 936 |
+
"grad_norm": 1.4939590692520142,
|
| 937 |
+
"learning_rate": 7.131578947368422e-06,
|
| 938 |
+
"loss": 0.1102,
|
| 939 |
"step": 3225
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 1.4894592117323557,
|
| 943 |
+
"grad_norm": 1.6188520193099976,
|
| 944 |
+
"learning_rate": 7.1052631578947375e-06,
|
| 945 |
+
"loss": 0.1025,
|
| 946 |
"step": 3250
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 1.500916590284143,
|
| 950 |
+
"grad_norm": 1.3144360780715942,
|
| 951 |
+
"learning_rate": 7.078947368421053e-06,
|
| 952 |
+
"loss": 0.1043,
|
| 953 |
"step": 3275
|
| 954 |
},
|
| 955 |
{
|
| 956 |
"epoch": 1.5123739688359303,
|
| 957 |
+
"grad_norm": 1.448574423789978,
|
| 958 |
+
"learning_rate": 7.052631578947369e-06,
|
| 959 |
+
"loss": 0.1073,
|
| 960 |
"step": 3300
|
| 961 |
},
|
| 962 |
{
|
| 963 |
"epoch": 1.5238313473877176,
|
| 964 |
+
"grad_norm": 1.7066571712493896,
|
| 965 |
+
"learning_rate": 7.026315789473684e-06,
|
| 966 |
+
"loss": 0.1126,
|
| 967 |
"step": 3325
|
| 968 |
},
|
| 969 |
{
|
| 970 |
"epoch": 1.535288725939505,
|
| 971 |
+
"grad_norm": 1.7523844242095947,
|
| 972 |
+
"learning_rate": 7e-06,
|
| 973 |
+
"loss": 0.1184,
|
| 974 |
"step": 3350
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 1.5467461044912922,
|
| 978 |
+
"grad_norm": 1.8683290481567383,
|
| 979 |
+
"learning_rate": 6.973684210526316e-06,
|
| 980 |
+
"loss": 0.1155,
|
| 981 |
"step": 3375
|
| 982 |
},
|
| 983 |
{
|
| 984 |
"epoch": 1.5582034830430798,
|
| 985 |
+
"grad_norm": 1.5641534328460693,
|
| 986 |
+
"learning_rate": 6.947368421052632e-06,
|
| 987 |
+
"loss": 0.104,
|
| 988 |
"step": 3400
|
| 989 |
},
|
| 990 |
{
|
| 991 |
"epoch": 1.569660861594867,
|
| 992 |
+
"grad_norm": 1.2918671369552612,
|
| 993 |
+
"learning_rate": 6.921052631578948e-06,
|
| 994 |
+
"loss": 0.0947,
|
| 995 |
"step": 3425
|
| 996 |
},
|
| 997 |
{
|
| 998 |
"epoch": 1.5811182401466546,
|
| 999 |
+
"grad_norm": 1.385241150856018,
|
| 1000 |
+
"learning_rate": 6.894736842105264e-06,
|
| 1001 |
+
"loss": 0.1047,
|
| 1002 |
"step": 3450
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
"epoch": 1.5925756186984419,
|
| 1006 |
+
"grad_norm": 1.273760437965393,
|
| 1007 |
+
"learning_rate": 6.86842105263158e-06,
|
| 1008 |
+
"loss": 0.116,
|
| 1009 |
"step": 3475
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 1.6040329972502292,
|
| 1013 |
+
"grad_norm": 1.3886901140213013,
|
| 1014 |
+
"learning_rate": 6.842105263157896e-06,
|
| 1015 |
+
"loss": 0.0973,
|
| 1016 |
"step": 3500
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 1.6154903758020165,
|
| 1020 |
+
"grad_norm": 1.5383498668670654,
|
| 1021 |
+
"learning_rate": 6.8157894736842115e-06,
|
| 1022 |
+
"loss": 0.0997,
|
| 1023 |
"step": 3525
|
| 1024 |
},
|
| 1025 |
{
|
| 1026 |
"epoch": 1.6269477543538038,
|
| 1027 |
+
"grad_norm": 1.4934067726135254,
|
| 1028 |
+
"learning_rate": 6.789473684210527e-06,
|
| 1029 |
+
"loss": 0.1026,
|
| 1030 |
"step": 3550
|
| 1031 |
},
|
| 1032 |
{
|
| 1033 |
"epoch": 1.638405132905591,
|
| 1034 |
+
"grad_norm": 1.6199575662612915,
|
| 1035 |
+
"learning_rate": 6.763157894736842e-06,
|
| 1036 |
+
"loss": 0.0946,
|
| 1037 |
"step": 3575
|
| 1038 |
},
|
| 1039 |
{
|
| 1040 |
"epoch": 1.6498625114573784,
|
| 1041 |
+
"grad_norm": 1.6967884302139282,
|
| 1042 |
+
"learning_rate": 6.736842105263158e-06,
|
| 1043 |
+
"loss": 0.1095,
|
| 1044 |
"step": 3600
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 1.661319890009166,
|
| 1048 |
+
"grad_norm": 1.9040991067886353,
|
| 1049 |
+
"learning_rate": 6.710526315789474e-06,
|
| 1050 |
+
"loss": 0.111,
|
| 1051 |
"step": 3625
|
| 1052 |
},
|
| 1053 |
{
|
| 1054 |
"epoch": 1.6727772685609532,
|
| 1055 |
+
"grad_norm": 1.9209736585617065,
|
| 1056 |
+
"learning_rate": 6.68421052631579e-06,
|
| 1057 |
+
"loss": 0.1094,
|
| 1058 |
"step": 3650
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 1.6842346471127407,
|
| 1062 |
+
"grad_norm": 1.5375550985336304,
|
| 1063 |
+
"learning_rate": 6.6578947368421055e-06,
|
| 1064 |
+
"loss": 0.1022,
|
| 1065 |
"step": 3675
|
| 1066 |
},
|
| 1067 |
{
|
| 1068 |
"epoch": 1.695692025664528,
|
| 1069 |
+
"grad_norm": 1.663802981376648,
|
| 1070 |
+
"learning_rate": 6.631578947368421e-06,
|
| 1071 |
+
"loss": 0.1019,
|
| 1072 |
"step": 3700
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 1.7071494042163153,
|
| 1076 |
+
"grad_norm": 1.6022824048995972,
|
| 1077 |
+
"learning_rate": 6.605263157894738e-06,
|
| 1078 |
+
"loss": 0.1109,
|
| 1079 |
"step": 3725
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 1.7186067827681026,
|
| 1083 |
+
"grad_norm": 1.4675346612930298,
|
| 1084 |
+
"learning_rate": 6.578947368421054e-06,
|
| 1085 |
+
"loss": 0.0909,
|
| 1086 |
"step": 3750
|
| 1087 |
},
|
| 1088 |
{
|
| 1089 |
"epoch": 1.73006416131989,
|
| 1090 |
+
"grad_norm": 1.2123456001281738,
|
| 1091 |
+
"learning_rate": 6.55263157894737e-06,
|
| 1092 |
+
"loss": 0.0938,
|
| 1093 |
"step": 3775
|
| 1094 |
},
|
| 1095 |
{
|
| 1096 |
"epoch": 1.7415215398716772,
|
| 1097 |
+
"grad_norm": 1.5863817930221558,
|
| 1098 |
+
"learning_rate": 6.526315789473685e-06,
|
| 1099 |
+
"loss": 0.0976,
|
| 1100 |
"step": 3800
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 1.7529789184234648,
|
| 1104 |
+
"grad_norm": 1.4508000612258911,
|
| 1105 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 1106 |
+
"loss": 0.1045,
|
| 1107 |
"step": 3825
|
| 1108 |
},
|
| 1109 |
{
|
| 1110 |
"epoch": 1.764436296975252,
|
| 1111 |
+
"grad_norm": 1.1027394533157349,
|
| 1112 |
+
"learning_rate": 6.473684210526316e-06,
|
| 1113 |
+
"loss": 0.1013,
|
| 1114 |
"step": 3850
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
"epoch": 1.7758936755270394,
|
| 1118 |
+
"grad_norm": 1.564916968345642,
|
| 1119 |
+
"learning_rate": 6.447368421052632e-06,
|
| 1120 |
+
"loss": 0.1048,
|
| 1121 |
"step": 3875
|
| 1122 |
},
|
| 1123 |
{
|
| 1124 |
"epoch": 1.7873510540788269,
|
| 1125 |
+
"grad_norm": 1.8783968687057495,
|
| 1126 |
+
"learning_rate": 6.421052631578948e-06,
|
| 1127 |
+
"loss": 0.1054,
|
| 1128 |
"step": 3900
|
| 1129 |
},
|
| 1130 |
{
|
| 1131 |
"epoch": 1.7988084326306142,
|
| 1132 |
+
"grad_norm": 1.552715539932251,
|
| 1133 |
+
"learning_rate": 6.394736842105264e-06,
|
| 1134 |
+
"loss": 0.1037,
|
| 1135 |
"step": 3925
|
| 1136 |
},
|
| 1137 |
{
|
| 1138 |
"epoch": 1.8102658111824015,
|
| 1139 |
+
"grad_norm": 1.2927756309509277,
|
| 1140 |
+
"learning_rate": 6.3684210526315795e-06,
|
| 1141 |
+
"loss": 0.108,
|
| 1142 |
"step": 3950
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 1.8217231897341888,
|
| 1146 |
+
"grad_norm": 1.1923288106918335,
|
| 1147 |
+
"learning_rate": 6.342105263157895e-06,
|
| 1148 |
+
"loss": 0.1029,
|
| 1149 |
"step": 3975
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 1.833180568285976,
|
| 1153 |
+
"grad_norm": 1.3957582712173462,
|
| 1154 |
+
"learning_rate": 6.31578947368421e-06,
|
| 1155 |
+
"loss": 0.0974,
|
| 1156 |
"step": 4000
|
| 1157 |
},
|
| 1158 |
{
|
| 1159 |
"epoch": 1.833180568285976,
|
| 1160 |
+
"eval_loss": 0.2684590220451355,
|
| 1161 |
+
"eval_runtime": 2099.1884,
|
| 1162 |
+
"eval_samples_per_second": 2.563,
|
| 1163 |
+
"eval_steps_per_second": 0.161,
|
| 1164 |
+
"eval_wer": 0.19740347528728777,
|
| 1165 |
"step": 4000
|
| 1166 |
},
|
| 1167 |
{
|
| 1168 |
"epoch": 1.8446379468377634,
|
| 1169 |
+
"grad_norm": 1.8228957653045654,
|
| 1170 |
+
"learning_rate": 6.289473684210526e-06,
|
| 1171 |
+
"loss": 0.1049,
|
| 1172 |
"step": 4025
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 1.856095325389551,
|
| 1176 |
+
"grad_norm": 1.336451768875122,
|
| 1177 |
+
"learning_rate": 6.263157894736842e-06,
|
| 1178 |
+
"loss": 0.105,
|
| 1179 |
"step": 4050
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 1.8675527039413382,
|
| 1183 |
+
"grad_norm": 1.4996263980865479,
|
| 1184 |
+
"learning_rate": 6.236842105263159e-06,
|
| 1185 |
+
"loss": 0.0967,
|
| 1186 |
"step": 4075
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
"epoch": 1.8790100824931257,
|
| 1190 |
+
"grad_norm": 0.9457367062568665,
|
| 1191 |
+
"learning_rate": 6.2105263157894745e-06,
|
| 1192 |
+
"loss": 0.0968,
|
| 1193 |
"step": 4100
|
| 1194 |
},
|
| 1195 |
{
|
| 1196 |
"epoch": 1.890467461044913,
|
| 1197 |
+
"grad_norm": 1.4064022302627563,
|
| 1198 |
+
"learning_rate": 6.18421052631579e-06,
|
| 1199 |
+
"loss": 0.1,
|
| 1200 |
"step": 4125
|
| 1201 |
},
|
| 1202 |
{
|
| 1203 |
"epoch": 1.9019248395967003,
|
| 1204 |
+
"grad_norm": 1.5435906648635864,
|
| 1205 |
+
"learning_rate": 6.157894736842106e-06,
|
| 1206 |
+
"loss": 0.104,
|
| 1207 |
"step": 4150
|
| 1208 |
},
|
| 1209 |
{
|
| 1210 |
"epoch": 1.9133822181484876,
|
| 1211 |
+
"grad_norm": 1.279659390449524,
|
| 1212 |
+
"learning_rate": 6.131578947368422e-06,
|
| 1213 |
+
"loss": 0.1054,
|
| 1214 |
"step": 4175
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 1.924839596700275,
|
| 1218 |
+
"grad_norm": 1.4175007343292236,
|
| 1219 |
+
"learning_rate": 6.105263157894738e-06,
|
| 1220 |
+
"loss": 0.0944,
|
| 1221 |
"step": 4200
|
| 1222 |
},
|
| 1223 |
{
|
| 1224 |
"epoch": 1.9362969752520622,
|
| 1225 |
+
"grad_norm": 1.416519284248352,
|
| 1226 |
+
"learning_rate": 6.0789473684210535e-06,
|
| 1227 |
+
"loss": 0.0989,
|
| 1228 |
"step": 4225
|
| 1229 |
},
|
| 1230 |
{
|
| 1231 |
"epoch": 1.9477543538038495,
|
| 1232 |
+
"grad_norm": 1.407576322555542,
|
| 1233 |
+
"learning_rate": 6.0526315789473685e-06,
|
| 1234 |
+
"loss": 0.1051,
|
| 1235 |
"step": 4250
|
| 1236 |
},
|
| 1237 |
{
|
| 1238 |
"epoch": 1.959211732355637,
|
| 1239 |
+
"grad_norm": 1.37060546875,
|
| 1240 |
+
"learning_rate": 6.026315789473684e-06,
|
| 1241 |
+
"loss": 0.0902,
|
| 1242 |
"step": 4275
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
"epoch": 1.9706691109074244,
|
| 1246 |
+
"grad_norm": 1.400992751121521,
|
| 1247 |
+
"learning_rate": 6e-06,
|
| 1248 |
+
"loss": 0.1145,
|
| 1249 |
"step": 4300
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 1.9821264894592119,
|
| 1253 |
+
"grad_norm": 1.3607090711593628,
|
| 1254 |
+
"learning_rate": 5.973684210526316e-06,
|
| 1255 |
+
"loss": 0.1066,
|
| 1256 |
"step": 4325
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 1.9935838680109992,
|
| 1260 |
+
"grad_norm": 1.6090093851089478,
|
| 1261 |
+
"learning_rate": 5.947368421052632e-06,
|
| 1262 |
+
"loss": 0.1114,
|
| 1263 |
"step": 4350
|
| 1264 |
},
|
| 1265 |
{
|
| 1266 |
"epoch": 2.0050412465627865,
|
| 1267 |
+
"grad_norm": 0.8756614923477173,
|
| 1268 |
+
"learning_rate": 5.921052631578948e-06,
|
| 1269 |
+
"loss": 0.0788,
|
| 1270 |
"step": 4375
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 2.016498625114574,
|
| 1274 |
+
"grad_norm": 0.8444830179214478,
|
| 1275 |
+
"learning_rate": 5.8947368421052634e-06,
|
| 1276 |
+
"loss": 0.0528,
|
| 1277 |
"step": 4400
|
| 1278 |
},
|
| 1279 |
{
|
| 1280 |
"epoch": 2.027956003666361,
|
| 1281 |
+
"grad_norm": 1.0788055658340454,
|
| 1282 |
+
"learning_rate": 5.86842105263158e-06,
|
| 1283 |
+
"loss": 0.0584,
|
| 1284 |
"step": 4425
|
| 1285 |
},
|
| 1286 |
{
|
| 1287 |
"epoch": 2.0394133822181484,
|
| 1288 |
+
"grad_norm": 1.3058183193206787,
|
| 1289 |
+
"learning_rate": 5.842105263157896e-06,
|
| 1290 |
+
"loss": 0.0557,
|
| 1291 |
"step": 4450
|
| 1292 |
},
|
| 1293 |
{
|
| 1294 |
"epoch": 2.0508707607699357,
|
| 1295 |
+
"grad_norm": 2.1910767555236816,
|
| 1296 |
+
"learning_rate": 5.815789473684212e-06,
|
| 1297 |
+
"loss": 0.0539,
|
| 1298 |
"step": 4475
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 2.062328139321723,
|
| 1302 |
+
"grad_norm": 1.0271133184432983,
|
| 1303 |
+
"learning_rate": 5.789473684210527e-06,
|
| 1304 |
+
"loss": 0.0515,
|
| 1305 |
"step": 4500
|
| 1306 |
},
|
| 1307 |
{
|
| 1308 |
"epoch": 2.0737855178735107,
|
| 1309 |
+
"grad_norm": 1.1689499616622925,
|
| 1310 |
+
"learning_rate": 5.7631578947368425e-06,
|
| 1311 |
+
"loss": 0.048,
|
| 1312 |
"step": 4525
|
| 1313 |
},
|
| 1314 |
{
|
| 1315 |
"epoch": 2.085242896425298,
|
| 1316 |
+
"grad_norm": 1.124924659729004,
|
| 1317 |
+
"learning_rate": 5.736842105263158e-06,
|
| 1318 |
+
"loss": 0.0573,
|
| 1319 |
"step": 4550
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 2.0967002749770853,
|
| 1323 |
+
"grad_norm": 0.9833737015724182,
|
| 1324 |
+
"learning_rate": 5.710526315789474e-06,
|
| 1325 |
+
"loss": 0.0576,
|
| 1326 |
"step": 4575
|
| 1327 |
},
|
| 1328 |
{
|
| 1329 |
"epoch": 2.1081576535288726,
|
| 1330 |
+
"grad_norm": 1.5147674083709717,
|
| 1331 |
+
"learning_rate": 5.68421052631579e-06,
|
| 1332 |
+
"loss": 0.0545,
|
| 1333 |
"step": 4600
|
| 1334 |
},
|
| 1335 |
{
|
| 1336 |
"epoch": 2.11961503208066,
|
| 1337 |
+
"grad_norm": 1.2202064990997314,
|
| 1338 |
+
"learning_rate": 5.657894736842106e-06,
|
| 1339 |
+
"loss": 0.0526,
|
| 1340 |
"step": 4625
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 2.1310724106324472,
|
| 1344 |
+
"grad_norm": 1.3528815507888794,
|
| 1345 |
+
"learning_rate": 5.631578947368422e-06,
|
| 1346 |
+
"loss": 0.0473,
|
| 1347 |
"step": 4650
|
| 1348 |
},
|
| 1349 |
{
|
| 1350 |
"epoch": 2.1425297891842345,
|
| 1351 |
+
"grad_norm": 1.4190665483474731,
|
| 1352 |
+
"learning_rate": 5.605263157894737e-06,
|
| 1353 |
+
"loss": 0.0539,
|
| 1354 |
"step": 4675
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
"epoch": 2.153987167736022,
|
| 1358 |
+
"grad_norm": 0.8159627318382263,
|
| 1359 |
+
"learning_rate": 5.578947368421052e-06,
|
| 1360 |
+
"loss": 0.0471,
|
| 1361 |
"step": 4700
|
| 1362 |
},
|
| 1363 |
{
|
| 1364 |
"epoch": 2.1654445462878096,
|
| 1365 |
+
"grad_norm": 1.2614290714263916,
|
| 1366 |
+
"learning_rate": 5.552631578947368e-06,
|
| 1367 |
+
"loss": 0.0521,
|
| 1368 |
"step": 4725
|
| 1369 |
},
|
| 1370 |
{
|
| 1371 |
"epoch": 2.176901924839597,
|
| 1372 |
+
"grad_norm": 1.099128007888794,
|
| 1373 |
+
"learning_rate": 5.526315789473685e-06,
|
| 1374 |
+
"loss": 0.052,
|
| 1375 |
"step": 4750
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 2.188359303391384,
|
| 1379 |
+
"grad_norm": 1.1397240161895752,
|
| 1380 |
+
"learning_rate": 5.500000000000001e-06,
|
| 1381 |
+
"loss": 0.0559,
|
| 1382 |
"step": 4775
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 2.1998166819431715,
|
| 1386 |
+
"grad_norm": 0.8417573571205139,
|
| 1387 |
+
"learning_rate": 5.4736842105263165e-06,
|
| 1388 |
+
"loss": 0.0492,
|
| 1389 |
"step": 4800
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 2.211274060494959,
|
| 1393 |
+
"grad_norm": 1.072788119316101,
|
| 1394 |
+
"learning_rate": 5.447368421052632e-06,
|
| 1395 |
+
"loss": 0.0466,
|
| 1396 |
"step": 4825
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 2.222731439046746,
|
| 1400 |
+
"grad_norm": 1.0280921459197998,
|
| 1401 |
+
"learning_rate": 5.421052631578948e-06,
|
| 1402 |
+
"loss": 0.0482,
|
| 1403 |
"step": 4850
|
| 1404 |
},
|
| 1405 |
{
|
| 1406 |
"epoch": 2.2341888175985334,
|
| 1407 |
+
"grad_norm": 0.9380873441696167,
|
| 1408 |
+
"learning_rate": 5.394736842105264e-06,
|
| 1409 |
+
"loss": 0.0575,
|
| 1410 |
"step": 4875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
"epoch": 2.2456461961503207,
|
| 1414 |
+
"grad_norm": 0.9564135074615479,
|
| 1415 |
+
"learning_rate": 5.36842105263158e-06,
|
| 1416 |
+
"loss": 0.0555,
|
| 1417 |
"step": 4900
|
| 1418 |
},
|
| 1419 |
{
|
| 1420 |
"epoch": 2.257103574702108,
|
| 1421 |
+
"grad_norm": 1.1567769050598145,
|
| 1422 |
+
"learning_rate": 5.342105263157895e-06,
|
| 1423 |
+
"loss": 0.048,
|
| 1424 |
"step": 4925
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 2.2685609532538953,
|
| 1428 |
+
"grad_norm": 1.1080228090286255,
|
| 1429 |
+
"learning_rate": 5.315789473684211e-06,
|
| 1430 |
+
"loss": 0.0533,
|
| 1431 |
"step": 4950
|
| 1432 |
},
|
| 1433 |
{
|
| 1434 |
"epoch": 2.280018331805683,
|
| 1435 |
+
"grad_norm": 1.2571756839752197,
|
| 1436 |
+
"learning_rate": 5.289473684210526e-06,
|
| 1437 |
+
"loss": 0.0542,
|
| 1438 |
"step": 4975
|
| 1439 |
},
|
| 1440 |
{
|
| 1441 |
"epoch": 2.2914757103574703,
|
| 1442 |
+
"grad_norm": 1.1968417167663574,
|
| 1443 |
+
"learning_rate": 5.263157894736842e-06,
|
| 1444 |
+
"loss": 0.0487,
|
| 1445 |
"step": 5000
|
| 1446 |
},
|
| 1447 |
{
|
| 1448 |
"epoch": 2.2914757103574703,
|
| 1449 |
+
"eval_loss": 0.2829594910144806,
|
| 1450 |
+
"eval_runtime": 2091.7962,
|
| 1451 |
+
"eval_samples_per_second": 2.572,
|
| 1452 |
+
"eval_steps_per_second": 0.161,
|
| 1453 |
+
"eval_wer": 0.20074356833273688,
|
| 1454 |
"step": 5000
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
+
"epoch": 2.3029330889092576,
|
| 1458 |
+
"grad_norm": 1.089119791984558,
|
| 1459 |
+
"learning_rate": 5.236842105263158e-06,
|
| 1460 |
+
"loss": 0.0514,
|
| 1461 |
+
"step": 5025
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 2.314390467461045,
|
| 1465 |
+
"grad_norm": 0.9352846741676331,
|
| 1466 |
+
"learning_rate": 5.210526315789474e-06,
|
| 1467 |
+
"loss": 0.0499,
|
| 1468 |
+
"step": 5050
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 2.3258478460128322,
|
| 1472 |
+
"grad_norm": 1.3354887962341309,
|
| 1473 |
+
"learning_rate": 5.18421052631579e-06,
|
| 1474 |
+
"loss": 0.0491,
|
| 1475 |
+
"step": 5075
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 2.3373052245646195,
|
| 1479 |
+
"grad_norm": 1.4073594808578491,
|
| 1480 |
+
"learning_rate": 5.157894736842106e-06,
|
| 1481 |
+
"loss": 0.0495,
|
| 1482 |
+
"step": 5100
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 2.348762603116407,
|
| 1486 |
+
"grad_norm": 1.3891620635986328,
|
| 1487 |
+
"learning_rate": 5.131578947368422e-06,
|
| 1488 |
+
"loss": 0.0528,
|
| 1489 |
+
"step": 5125
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 2.360219981668194,
|
| 1493 |
+
"grad_norm": 1.224021315574646,
|
| 1494 |
+
"learning_rate": 5.105263157894738e-06,
|
| 1495 |
+
"loss": 0.0521,
|
| 1496 |
+
"step": 5150
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 2.371677360219982,
|
| 1500 |
+
"grad_norm": 1.1707936525344849,
|
| 1501 |
+
"learning_rate": 5.078947368421053e-06,
|
| 1502 |
+
"loss": 0.0514,
|
| 1503 |
+
"step": 5175
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 2.383134738771769,
|
| 1507 |
+
"grad_norm": 1.0324437618255615,
|
| 1508 |
+
"learning_rate": 5.052631578947369e-06,
|
| 1509 |
+
"loss": 0.0521,
|
| 1510 |
+
"step": 5200
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 2.3945921173235565,
|
| 1514 |
+
"grad_norm": 1.1729110479354858,
|
| 1515 |
+
"learning_rate": 5.026315789473685e-06,
|
| 1516 |
+
"loss": 0.0498,
|
| 1517 |
+
"step": 5225
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.406049495875344,
|
| 1521 |
+
"grad_norm": 0.8670728206634521,
|
| 1522 |
+
"learning_rate": 5e-06,
|
| 1523 |
+
"loss": 0.0493,
|
| 1524 |
+
"step": 5250
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 2.417506874427131,
|
| 1528 |
+
"grad_norm": 1.3371397256851196,
|
| 1529 |
+
"learning_rate": 4.973684210526316e-06,
|
| 1530 |
+
"loss": 0.0506,
|
| 1531 |
+
"step": 5275
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 2.4289642529789184,
|
| 1535 |
+
"grad_norm": 1.197558045387268,
|
| 1536 |
+
"learning_rate": 4.947368421052632e-06,
|
| 1537 |
+
"loss": 0.0526,
|
| 1538 |
+
"step": 5300
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 2.4404216315307057,
|
| 1542 |
+
"grad_norm": 0.9833498597145081,
|
| 1543 |
+
"learning_rate": 4.921052631578948e-06,
|
| 1544 |
+
"loss": 0.0526,
|
| 1545 |
+
"step": 5325
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 2.451879010082493,
|
| 1549 |
+
"grad_norm": 1.0011742115020752,
|
| 1550 |
+
"learning_rate": 4.894736842105264e-06,
|
| 1551 |
+
"loss": 0.0517,
|
| 1552 |
+
"step": 5350
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"epoch": 2.4633363886342803,
|
| 1556 |
+
"grad_norm": 0.8775746822357178,
|
| 1557 |
+
"learning_rate": 4.8684210526315795e-06,
|
| 1558 |
+
"loss": 0.0557,
|
| 1559 |
+
"step": 5375
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 2.4747937671860676,
|
| 1563 |
+
"grad_norm": 1.154494285583496,
|
| 1564 |
+
"learning_rate": 4.842105263157895e-06,
|
| 1565 |
+
"loss": 0.0465,
|
| 1566 |
+
"step": 5400
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 2.4862511457378553,
|
| 1570 |
+
"grad_norm": 0.7587563991546631,
|
| 1571 |
+
"learning_rate": 4.815789473684211e-06,
|
| 1572 |
+
"loss": 0.05,
|
| 1573 |
+
"step": 5425
|
| 1574 |
+
},
|
| 1575 |
+
{
|
| 1576 |
+
"epoch": 2.4977085242896426,
|
| 1577 |
+
"grad_norm": 1.0355983972549438,
|
| 1578 |
+
"learning_rate": 4.789473684210527e-06,
|
| 1579 |
+
"loss": 0.0583,
|
| 1580 |
+
"step": 5450
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 2.50916590284143,
|
| 1584 |
+
"grad_norm": 1.1253951787948608,
|
| 1585 |
+
"learning_rate": 4.763157894736842e-06,
|
| 1586 |
+
"loss": 0.0518,
|
| 1587 |
+
"step": 5475
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 2.5206232813932172,
|
| 1591 |
+
"grad_norm": 1.2069380283355713,
|
| 1592 |
+
"learning_rate": 4.736842105263158e-06,
|
| 1593 |
+
"loss": 0.0473,
|
| 1594 |
+
"step": 5500
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 2.5320806599450045,
|
| 1598 |
+
"grad_norm": 0.9651533365249634,
|
| 1599 |
+
"learning_rate": 4.710526315789474e-06,
|
| 1600 |
+
"loss": 0.0541,
|
| 1601 |
+
"step": 5525
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 2.543538038496792,
|
| 1605 |
+
"grad_norm": 0.7548551559448242,
|
| 1606 |
+
"learning_rate": 4.68421052631579e-06,
|
| 1607 |
+
"loss": 0.0504,
|
| 1608 |
+
"step": 5550
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 2.554995417048579,
|
| 1612 |
+
"grad_norm": 1.099012851715088,
|
| 1613 |
+
"learning_rate": 4.657894736842106e-06,
|
| 1614 |
+
"loss": 0.0588,
|
| 1615 |
+
"step": 5575
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 2.566452795600367,
|
| 1619 |
+
"grad_norm": 1.1170094013214111,
|
| 1620 |
+
"learning_rate": 4.631578947368421e-06,
|
| 1621 |
+
"loss": 0.0505,
|
| 1622 |
+
"step": 5600
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 2.577910174152154,
|
| 1626 |
+
"grad_norm": 1.1232651472091675,
|
| 1627 |
+
"learning_rate": 4.605263157894737e-06,
|
| 1628 |
+
"loss": 0.0487,
|
| 1629 |
+
"step": 5625
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 2.5893675527039415,
|
| 1633 |
+
"grad_norm": 1.077101469039917,
|
| 1634 |
+
"learning_rate": 4.578947368421053e-06,
|
| 1635 |
+
"loss": 0.0482,
|
| 1636 |
+
"step": 5650
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 2.600824931255729,
|
| 1640 |
+
"grad_norm": 1.3222973346710205,
|
| 1641 |
+
"learning_rate": 4.552631578947369e-06,
|
| 1642 |
+
"loss": 0.0439,
|
| 1643 |
+
"step": 5675
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 2.612282309807516,
|
| 1647 |
+
"grad_norm": 1.2257323265075684,
|
| 1648 |
+
"learning_rate": 4.526315789473685e-06,
|
| 1649 |
+
"loss": 0.0539,
|
| 1650 |
+
"step": 5700
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 2.6237396883593034,
|
| 1654 |
+
"grad_norm": 2.47105073928833,
|
| 1655 |
+
"learning_rate": 4.5e-06,
|
| 1656 |
+
"loss": 0.0513,
|
| 1657 |
+
"step": 5725
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 2.6351970669110907,
|
| 1661 |
+
"grad_norm": 1.258948564529419,
|
| 1662 |
+
"learning_rate": 4.473684210526316e-06,
|
| 1663 |
+
"loss": 0.0498,
|
| 1664 |
+
"step": 5750
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 2.646654445462878,
|
| 1668 |
+
"grad_norm": 1.0322165489196777,
|
| 1669 |
+
"learning_rate": 4.447368421052632e-06,
|
| 1670 |
+
"loss": 0.0498,
|
| 1671 |
+
"step": 5775
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 2.6581118240146653,
|
| 1675 |
+
"grad_norm": 1.2974045276641846,
|
| 1676 |
+
"learning_rate": 4.4210526315789476e-06,
|
| 1677 |
+
"loss": 0.0519,
|
| 1678 |
+
"step": 5800
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 2.6695692025664526,
|
| 1682 |
+
"grad_norm": 0.9524481892585754,
|
| 1683 |
+
"learning_rate": 4.394736842105263e-06,
|
| 1684 |
+
"loss": 0.0565,
|
| 1685 |
+
"step": 5825
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 2.68102658111824,
|
| 1689 |
+
"grad_norm": 1.3988256454467773,
|
| 1690 |
+
"learning_rate": 4.368421052631579e-06,
|
| 1691 |
+
"loss": 0.0483,
|
| 1692 |
+
"step": 5850
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 2.6924839596700276,
|
| 1696 |
+
"grad_norm": 0.9579303860664368,
|
| 1697 |
+
"learning_rate": 4.342105263157895e-06,
|
| 1698 |
+
"loss": 0.049,
|
| 1699 |
+
"step": 5875
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 2.703941338221815,
|
| 1703 |
+
"grad_norm": 1.2459129095077515,
|
| 1704 |
+
"learning_rate": 4.315789473684211e-06,
|
| 1705 |
+
"loss": 0.0556,
|
| 1706 |
+
"step": 5900
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 2.7153987167736022,
|
| 1710 |
+
"grad_norm": 0.9496270418167114,
|
| 1711 |
+
"learning_rate": 4.289473684210527e-06,
|
| 1712 |
+
"loss": 0.0502,
|
| 1713 |
+
"step": 5925
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 2.7268560953253895,
|
| 1717 |
+
"grad_norm": 1.4243717193603516,
|
| 1718 |
+
"learning_rate": 4.2631578947368425e-06,
|
| 1719 |
+
"loss": 0.0613,
|
| 1720 |
+
"step": 5950
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 2.738313473877177,
|
| 1724 |
+
"grad_norm": 1.2027771472930908,
|
| 1725 |
+
"learning_rate": 4.236842105263158e-06,
|
| 1726 |
+
"loss": 0.0496,
|
| 1727 |
+
"step": 5975
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 2.749770852428964,
|
| 1731 |
+
"grad_norm": 1.2234293222427368,
|
| 1732 |
+
"learning_rate": 4.210526315789474e-06,
|
| 1733 |
+
"loss": 0.048,
|
| 1734 |
+
"step": 6000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 2.749770852428964,
|
| 1738 |
+
"eval_loss": 0.28926825523376465,
|
| 1739 |
+
"eval_runtime": 2085.896,
|
| 1740 |
+
"eval_samples_per_second": 2.58,
|
| 1741 |
+
"eval_steps_per_second": 0.162,
|
| 1742 |
+
"eval_wer": 0.20207562924967196,
|
| 1743 |
+
"step": 6000
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 2.761228230980752,
|
| 1747 |
+
"grad_norm": 1.0857855081558228,
|
| 1748 |
+
"learning_rate": 4.18421052631579e-06,
|
| 1749 |
+
"loss": 0.0489,
|
| 1750 |
+
"step": 6025
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 2.772685609532539,
|
| 1754 |
+
"grad_norm": 0.9659659266471863,
|
| 1755 |
+
"learning_rate": 4.157894736842106e-06,
|
| 1756 |
+
"loss": 0.043,
|
| 1757 |
+
"step": 6050
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 2.7841429880843265,
|
| 1761 |
+
"grad_norm": 0.8970678448677063,
|
| 1762 |
+
"learning_rate": 4.1315789473684216e-06,
|
| 1763 |
+
"loss": 0.0506,
|
| 1764 |
+
"step": 6075
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.7956003666361138,
|
| 1768 |
+
"grad_norm": 1.3365782499313354,
|
| 1769 |
+
"learning_rate": 4.105263157894737e-06,
|
| 1770 |
+
"loss": 0.0502,
|
| 1771 |
+
"step": 6100
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 2.807057745187901,
|
| 1775 |
+
"grad_norm": 1.443233609199524,
|
| 1776 |
+
"learning_rate": 4.078947368421053e-06,
|
| 1777 |
+
"loss": 0.0508,
|
| 1778 |
+
"step": 6125
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 2.8185151237396884,
|
| 1782 |
+
"grad_norm": 1.3378056287765503,
|
| 1783 |
+
"learning_rate": 4.052631578947368e-06,
|
| 1784 |
+
"loss": 0.0451,
|
| 1785 |
+
"step": 6150
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 2.8299725022914757,
|
| 1789 |
+
"grad_norm": 1.2966234683990479,
|
| 1790 |
+
"learning_rate": 4.026315789473684e-06,
|
| 1791 |
+
"loss": 0.0526,
|
| 1792 |
+
"step": 6175
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 2.841429880843263,
|
| 1796 |
+
"grad_norm": 1.0939254760742188,
|
| 1797 |
+
"learning_rate": 4.000000000000001e-06,
|
| 1798 |
+
"loss": 0.0526,
|
| 1799 |
+
"step": 6200
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 2.8528872593950503,
|
| 1803 |
+
"grad_norm": 0.8236600160598755,
|
| 1804 |
+
"learning_rate": 3.9736842105263165e-06,
|
| 1805 |
+
"loss": 0.0422,
|
| 1806 |
+
"step": 6225
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 2.8643446379468376,
|
| 1810 |
+
"grad_norm": 1.5178745985031128,
|
| 1811 |
+
"learning_rate": 3.947368421052632e-06,
|
| 1812 |
+
"loss": 0.0522,
|
| 1813 |
+
"step": 6250
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 2.875802016498625,
|
| 1817 |
+
"grad_norm": 1.1574722528457642,
|
| 1818 |
+
"learning_rate": 3.921052631578947e-06,
|
| 1819 |
+
"loss": 0.0468,
|
| 1820 |
+
"step": 6275
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 2.8872593950504126,
|
| 1824 |
+
"grad_norm": 1.4163748025894165,
|
| 1825 |
+
"learning_rate": 3.894736842105263e-06,
|
| 1826 |
+
"loss": 0.048,
|
| 1827 |
+
"step": 6300
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 2.8987167736022,
|
| 1831 |
+
"grad_norm": 0.8275723457336426,
|
| 1832 |
+
"learning_rate": 3.868421052631579e-06,
|
| 1833 |
+
"loss": 0.0454,
|
| 1834 |
+
"step": 6325
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.9101741521539872,
|
| 1838 |
+
"grad_norm": 1.1487932205200195,
|
| 1839 |
+
"learning_rate": 3.842105263157895e-06,
|
| 1840 |
+
"loss": 0.0484,
|
| 1841 |
+
"step": 6350
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 2.9216315307057745,
|
| 1845 |
+
"grad_norm": 1.0844917297363281,
|
| 1846 |
+
"learning_rate": 3.815789473684211e-06,
|
| 1847 |
+
"loss": 0.0478,
|
| 1848 |
+
"step": 6375
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 2.933088909257562,
|
| 1852 |
+
"grad_norm": 1.010520100593567,
|
| 1853 |
+
"learning_rate": 3.789473684210527e-06,
|
| 1854 |
+
"loss": 0.0463,
|
| 1855 |
+
"step": 6400
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 2.944546287809349,
|
| 1859 |
+
"grad_norm": 0.9863224625587463,
|
| 1860 |
+
"learning_rate": 3.7631578947368426e-06,
|
| 1861 |
+
"loss": 0.0497,
|
| 1862 |
+
"step": 6425
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 2.9560036663611364,
|
| 1866 |
+
"grad_norm": 0.5758341550827026,
|
| 1867 |
+
"learning_rate": 3.736842105263158e-06,
|
| 1868 |
+
"loss": 0.0482,
|
| 1869 |
+
"step": 6450
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 2.967461044912924,
|
| 1873 |
+
"grad_norm": 0.9042133092880249,
|
| 1874 |
+
"learning_rate": 3.710526315789474e-06,
|
| 1875 |
+
"loss": 0.0531,
|
| 1876 |
+
"step": 6475
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 2.9789184234647115,
|
| 1880 |
+
"grad_norm": 1.2652826309204102,
|
| 1881 |
+
"learning_rate": 3.6842105263157896e-06,
|
| 1882 |
+
"loss": 0.0489,
|
| 1883 |
+
"step": 6500
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 2.9903758020164988,
|
| 1887 |
+
"grad_norm": 0.9611257910728455,
|
| 1888 |
+
"learning_rate": 3.658947368421053e-06,
|
| 1889 |
+
"loss": 0.0556,
|
| 1890 |
+
"step": 6525
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 3.001833180568286,
|
| 1894 |
+
"grad_norm": 0.86037278175354,
|
| 1895 |
+
"learning_rate": 3.632631578947369e-06,
|
| 1896 |
+
"loss": 0.0449,
|
| 1897 |
+
"step": 6550
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 3.0132905591200734,
|
| 1901 |
+
"grad_norm": 0.6133009791374207,
|
| 1902 |
+
"learning_rate": 3.6063157894736846e-06,
|
| 1903 |
+
"loss": 0.0236,
|
| 1904 |
+
"step": 6575
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 3.0247479376718607,
|
| 1908 |
+
"grad_norm": 0.7855672836303711,
|
| 1909 |
+
"learning_rate": 3.58e-06,
|
| 1910 |
+
"loss": 0.0231,
|
| 1911 |
+
"step": 6600
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 3.036205316223648,
|
| 1915 |
+
"grad_norm": 0.6326671838760376,
|
| 1916 |
+
"learning_rate": 3.553684210526316e-06,
|
| 1917 |
+
"loss": 0.0221,
|
| 1918 |
+
"step": 6625
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 3.0476626947754353,
|
| 1922 |
+
"grad_norm": 0.48643797636032104,
|
| 1923 |
+
"learning_rate": 3.527368421052632e-06,
|
| 1924 |
+
"loss": 0.0229,
|
| 1925 |
+
"step": 6650
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 3.0591200733272226,
|
| 1929 |
+
"grad_norm": 0.820506751537323,
|
| 1930 |
+
"learning_rate": 3.501052631578948e-06,
|
| 1931 |
+
"loss": 0.0226,
|
| 1932 |
+
"step": 6675
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 3.07057745187901,
|
| 1936 |
+
"grad_norm": 0.6653295159339905,
|
| 1937 |
+
"learning_rate": 3.4747368421052637e-06,
|
| 1938 |
+
"loss": 0.0187,
|
| 1939 |
+
"step": 6700
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 3.0820348304307976,
|
| 1943 |
+
"grad_norm": 0.9965829253196716,
|
| 1944 |
+
"learning_rate": 3.448421052631579e-06,
|
| 1945 |
+
"loss": 0.0232,
|
| 1946 |
+
"step": 6725
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 3.093492208982585,
|
| 1950 |
+
"grad_norm": 0.7007229328155518,
|
| 1951 |
+
"learning_rate": 3.422105263157895e-06,
|
| 1952 |
+
"loss": 0.0223,
|
| 1953 |
+
"step": 6750
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 3.1049495875343722,
|
| 1957 |
+
"grad_norm": 0.6407040357589722,
|
| 1958 |
+
"learning_rate": 3.3957894736842108e-06,
|
| 1959 |
+
"loss": 0.0246,
|
| 1960 |
+
"step": 6775
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 3.1164069660861595,
|
| 1964 |
+
"grad_norm": 0.7243047952651978,
|
| 1965 |
+
"learning_rate": 3.369473684210526e-06,
|
| 1966 |
+
"loss": 0.0191,
|
| 1967 |
+
"step": 6800
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 3.127864344637947,
|
| 1971 |
+
"grad_norm": 1.1909897327423096,
|
| 1972 |
+
"learning_rate": 3.3431578947368424e-06,
|
| 1973 |
+
"loss": 0.0239,
|
| 1974 |
+
"step": 6825
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 3.139321723189734,
|
| 1978 |
+
"grad_norm": 0.46466654539108276,
|
| 1979 |
+
"learning_rate": 3.3168421052631582e-06,
|
| 1980 |
+
"loss": 0.0216,
|
| 1981 |
+
"step": 6850
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 3.1507791017415214,
|
| 1985 |
+
"grad_norm": 0.9051262736320496,
|
| 1986 |
+
"learning_rate": 3.290526315789474e-06,
|
| 1987 |
+
"loss": 0.0207,
|
| 1988 |
+
"step": 6875
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 3.1622364802933087,
|
| 1992 |
+
"grad_norm": 0.633877694606781,
|
| 1993 |
+
"learning_rate": 3.26421052631579e-06,
|
| 1994 |
+
"loss": 0.0213,
|
| 1995 |
+
"step": 6900
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 3.1736938588450965,
|
| 1999 |
+
"grad_norm": 0.9209594130516052,
|
| 2000 |
+
"learning_rate": 3.2378947368421052e-06,
|
| 2001 |
+
"loss": 0.0212,
|
| 2002 |
+
"step": 6925
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 3.1851512373968838,
|
| 2006 |
+
"grad_norm": 0.8294422626495361,
|
| 2007 |
+
"learning_rate": 3.211578947368421e-06,
|
| 2008 |
+
"loss": 0.0215,
|
| 2009 |
+
"step": 6950
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 3.196608615948671,
|
| 2013 |
+
"grad_norm": 0.5120604038238525,
|
| 2014 |
+
"learning_rate": 3.185263157894737e-06,
|
| 2015 |
+
"loss": 0.0217,
|
| 2016 |
+
"step": 6975
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 3.2080659945004584,
|
| 2020 |
+
"grad_norm": 0.9592424631118774,
|
| 2021 |
+
"learning_rate": 3.158947368421053e-06,
|
| 2022 |
+
"loss": 0.0206,
|
| 2023 |
+
"step": 7000
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 3.2080659945004584,
|
| 2027 |
+
"eval_loss": 0.30604422092437744,
|
| 2028 |
+
"eval_runtime": 2080.3707,
|
| 2029 |
+
"eval_samples_per_second": 2.587,
|
| 2030 |
+
"eval_steps_per_second": 0.162,
|
| 2031 |
+
"eval_wer": 0.19801980198019803,
|
| 2032 |
+
"step": 7000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 3.2195233730522457,
|
| 2036 |
+
"grad_norm": 0.9377903938293457,
|
| 2037 |
+
"learning_rate": 3.132631578947369e-06,
|
| 2038 |
+
"loss": 0.0223,
|
| 2039 |
+
"step": 7025
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 3.230980751604033,
|
| 2043 |
+
"grad_norm": 0.7186349630355835,
|
| 2044 |
+
"learning_rate": 3.1063157894736843e-06,
|
| 2045 |
+
"loss": 0.0215,
|
| 2046 |
+
"step": 7050
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 3.2424381301558203,
|
| 2050 |
+
"grad_norm": 0.9919578433036804,
|
| 2051 |
+
"learning_rate": 3.08e-06,
|
| 2052 |
+
"loss": 0.023,
|
| 2053 |
+
"step": 7075
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 3.2538955087076076,
|
| 2057 |
+
"grad_norm": 1.0826903581619263,
|
| 2058 |
+
"learning_rate": 3.053684210526316e-06,
|
| 2059 |
+
"loss": 0.0209,
|
| 2060 |
+
"step": 7100
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 3.265352887259395,
|
| 2064 |
+
"grad_norm": 0.5050626993179321,
|
| 2065 |
+
"learning_rate": 3.027368421052632e-06,
|
| 2066 |
+
"loss": 0.0209,
|
| 2067 |
+
"step": 7125
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 3.276810265811182,
|
| 2071 |
+
"grad_norm": 0.8884072303771973,
|
| 2072 |
+
"learning_rate": 3.001052631578948e-06,
|
| 2073 |
+
"loss": 0.0184,
|
| 2074 |
+
"step": 7150
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 3.28826764436297,
|
| 2078 |
+
"grad_norm": 0.8532277345657349,
|
| 2079 |
+
"learning_rate": 2.9747368421052634e-06,
|
| 2080 |
+
"loss": 0.0229,
|
| 2081 |
+
"step": 7175
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 3.299725022914757,
|
| 2085 |
+
"grad_norm": 0.6487672924995422,
|
| 2086 |
+
"learning_rate": 2.9484210526315792e-06,
|
| 2087 |
+
"loss": 0.0235,
|
| 2088 |
+
"step": 7200
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 3.3111824014665445,
|
| 2092 |
+
"grad_norm": 0.8241958022117615,
|
| 2093 |
+
"learning_rate": 2.922105263157895e-06,
|
| 2094 |
+
"loss": 0.0193,
|
| 2095 |
+
"step": 7225
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 3.322639780018332,
|
| 2099 |
+
"grad_norm": 1.106886625289917,
|
| 2100 |
+
"learning_rate": 2.895789473684211e-06,
|
| 2101 |
+
"loss": 0.02,
|
| 2102 |
+
"step": 7250
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 3.334097158570119,
|
| 2106 |
+
"grad_norm": 0.7387537360191345,
|
| 2107 |
+
"learning_rate": 2.8694736842105263e-06,
|
| 2108 |
+
"loss": 0.0227,
|
| 2109 |
+
"step": 7275
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 3.3455545371219064,
|
| 2113 |
+
"grad_norm": 1.1280341148376465,
|
| 2114 |
+
"learning_rate": 2.843157894736842e-06,
|
| 2115 |
+
"loss": 0.0226,
|
| 2116 |
+
"step": 7300
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 3.3570119156736937,
|
| 2120 |
+
"grad_norm": 0.9412670731544495,
|
| 2121 |
+
"learning_rate": 2.8168421052631583e-06,
|
| 2122 |
+
"loss": 0.023,
|
| 2123 |
+
"step": 7325
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 3.368469294225481,
|
| 2127 |
+
"grad_norm": 0.7090961337089539,
|
| 2128 |
+
"learning_rate": 2.790526315789474e-06,
|
| 2129 |
+
"loss": 0.0232,
|
| 2130 |
+
"step": 7350
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 3.3799266727772688,
|
| 2134 |
+
"grad_norm": 0.828831672668457,
|
| 2135 |
+
"learning_rate": 2.76421052631579e-06,
|
| 2136 |
+
"loss": 0.0196,
|
| 2137 |
+
"step": 7375
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 3.391384051329056,
|
| 2141 |
+
"grad_norm": 0.6595268845558167,
|
| 2142 |
+
"learning_rate": 2.7378947368421054e-06,
|
| 2143 |
+
"loss": 0.0183,
|
| 2144 |
+
"step": 7400
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 3.4028414298808434,
|
| 2148 |
+
"grad_norm": 0.6520357131958008,
|
| 2149 |
+
"learning_rate": 2.711578947368421e-06,
|
| 2150 |
+
"loss": 0.021,
|
| 2151 |
+
"step": 7425
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 3.4142988084326307,
|
| 2155 |
+
"grad_norm": 0.5938483476638794,
|
| 2156 |
+
"learning_rate": 2.685263157894737e-06,
|
| 2157 |
+
"loss": 0.0223,
|
| 2158 |
+
"step": 7450
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 3.425756186984418,
|
| 2162 |
+
"grad_norm": 0.6200276017189026,
|
| 2163 |
+
"learning_rate": 2.6589473684210524e-06,
|
| 2164 |
+
"loss": 0.0215,
|
| 2165 |
+
"step": 7475
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 3.4372135655362053,
|
| 2169 |
+
"grad_norm": 1.1649701595306396,
|
| 2170 |
+
"learning_rate": 2.632631578947369e-06,
|
| 2171 |
+
"loss": 0.0262,
|
| 2172 |
+
"step": 7500
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 3.4486709440879926,
|
| 2176 |
+
"grad_norm": 0.8650830388069153,
|
| 2177 |
+
"learning_rate": 2.6063157894736845e-06,
|
| 2178 |
+
"loss": 0.0233,
|
| 2179 |
+
"step": 7525
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 3.46012832263978,
|
| 2183 |
+
"grad_norm": 0.6462095379829407,
|
| 2184 |
+
"learning_rate": 2.5800000000000003e-06,
|
| 2185 |
+
"loss": 0.0235,
|
| 2186 |
+
"step": 7550
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 3.471585701191567,
|
| 2190 |
+
"grad_norm": 0.8197653293609619,
|
| 2191 |
+
"learning_rate": 2.553684210526316e-06,
|
| 2192 |
+
"loss": 0.0236,
|
| 2193 |
+
"step": 7575
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 3.4830430797433545,
|
| 2197 |
+
"grad_norm": 0.6480600833892822,
|
| 2198 |
+
"learning_rate": 2.5273684210526315e-06,
|
| 2199 |
+
"loss": 0.0196,
|
| 2200 |
+
"step": 7600
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 3.494500458295142,
|
| 2204 |
+
"grad_norm": 0.8653817772865295,
|
| 2205 |
+
"learning_rate": 2.5010526315789473e-06,
|
| 2206 |
+
"loss": 0.023,
|
| 2207 |
+
"step": 7625
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 3.5059578368469295,
|
| 2211 |
+
"grad_norm": 0.9173413515090942,
|
| 2212 |
+
"learning_rate": 2.4747368421052636e-06,
|
| 2213 |
+
"loss": 0.0232,
|
| 2214 |
+
"step": 7650
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 3.517415215398717,
|
| 2218 |
+
"grad_norm": 0.8143410086631775,
|
| 2219 |
+
"learning_rate": 2.448421052631579e-06,
|
| 2220 |
+
"loss": 0.0229,
|
| 2221 |
+
"step": 7675
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 3.528872593950504,
|
| 2225 |
+
"grad_norm": 0.5569326281547546,
|
| 2226 |
+
"learning_rate": 2.422105263157895e-06,
|
| 2227 |
+
"loss": 0.0205,
|
| 2228 |
+
"step": 7700
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 3.5403299725022914,
|
| 2232 |
+
"grad_norm": 0.9305413365364075,
|
| 2233 |
+
"learning_rate": 2.3957894736842106e-06,
|
| 2234 |
+
"loss": 0.023,
|
| 2235 |
+
"step": 7725
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 3.5517873510540787,
|
| 2239 |
+
"grad_norm": 1.082289218902588,
|
| 2240 |
+
"learning_rate": 2.3694736842105264e-06,
|
| 2241 |
+
"loss": 0.022,
|
| 2242 |
+
"step": 7750
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 3.563244729605866,
|
| 2246 |
+
"grad_norm": 0.6017982959747314,
|
| 2247 |
+
"learning_rate": 2.3431578947368422e-06,
|
| 2248 |
+
"loss": 0.0218,
|
| 2249 |
+
"step": 7775
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 3.5747021081576538,
|
| 2253 |
+
"grad_norm": 1.0354348421096802,
|
| 2254 |
+
"learning_rate": 2.316842105263158e-06,
|
| 2255 |
+
"loss": 0.0216,
|
| 2256 |
+
"step": 7800
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.586159486709441,
|
| 2260 |
+
"grad_norm": 0.6072118282318115,
|
| 2261 |
+
"learning_rate": 2.290526315789474e-06,
|
| 2262 |
+
"loss": 0.0225,
|
| 2263 |
+
"step": 7825
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 3.5976168652612284,
|
| 2267 |
+
"grad_norm": 0.6628375053405762,
|
| 2268 |
+
"learning_rate": 2.2642105263157897e-06,
|
| 2269 |
+
"loss": 0.0191,
|
| 2270 |
+
"step": 7850
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 3.6090742438130157,
|
| 2274 |
+
"grad_norm": 0.9418803453445435,
|
| 2275 |
+
"learning_rate": 2.2378947368421055e-06,
|
| 2276 |
+
"loss": 0.0247,
|
| 2277 |
+
"step": 7875
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 3.620531622364803,
|
| 2281 |
+
"grad_norm": 1.1218111515045166,
|
| 2282 |
+
"learning_rate": 2.2115789473684213e-06,
|
| 2283 |
+
"loss": 0.0219,
|
| 2284 |
+
"step": 7900
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 3.6319890009165903,
|
| 2288 |
+
"grad_norm": 0.9946068525314331,
|
| 2289 |
+
"learning_rate": 2.185263157894737e-06,
|
| 2290 |
+
"loss": 0.0226,
|
| 2291 |
+
"step": 7925
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 3.6434463794683776,
|
| 2295 |
+
"grad_norm": 0.8803802728652954,
|
| 2296 |
+
"learning_rate": 2.1589473684210525e-06,
|
| 2297 |
+
"loss": 0.0205,
|
| 2298 |
+
"step": 7950
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 3.654903758020165,
|
| 2302 |
+
"grad_norm": 0.7733520269393921,
|
| 2303 |
+
"learning_rate": 2.1326315789473688e-06,
|
| 2304 |
+
"loss": 0.023,
|
| 2305 |
+
"step": 7975
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 3.666361136571952,
|
| 2309 |
+
"grad_norm": 0.8802329897880554,
|
| 2310 |
+
"learning_rate": 2.106315789473684e-06,
|
| 2311 |
+
"loss": 0.0225,
|
| 2312 |
+
"step": 8000
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 3.666361136571952,
|
| 2316 |
+
"eval_loss": 0.3290001451969147,
|
| 2317 |
+
"eval_runtime": 2119.6879,
|
| 2318 |
+
"eval_samples_per_second": 2.539,
|
| 2319 |
+
"eval_steps_per_second": 0.159,
|
| 2320 |
+
"eval_wer": 0.2035468607101674,
|
| 2321 |
+
"step": 8000
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 3.6778185151237395,
|
| 2325 |
+
"grad_norm": 0.8936401009559631,
|
| 2326 |
+
"learning_rate": 2.08e-06,
|
| 2327 |
+
"loss": 0.0211,
|
| 2328 |
+
"step": 8025
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 3.6892758936755268,
|
| 2332 |
+
"grad_norm": 0.6175429821014404,
|
| 2333 |
+
"learning_rate": 2.0536842105263162e-06,
|
| 2334 |
+
"loss": 0.0242,
|
| 2335 |
+
"step": 8050
|
| 2336 |
+
},
|
| 2337 |
+
{
|
| 2338 |
+
"epoch": 3.7007332722273145,
|
| 2339 |
+
"grad_norm": 0.8317164778709412,
|
| 2340 |
+
"learning_rate": 2.0273684210526316e-06,
|
| 2341 |
+
"loss": 0.0217,
|
| 2342 |
+
"step": 8075
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 3.712190650779102,
|
| 2346 |
+
"grad_norm": 0.522230863571167,
|
| 2347 |
+
"learning_rate": 2.0010526315789474e-06,
|
| 2348 |
+
"loss": 0.0183,
|
| 2349 |
+
"step": 8100
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 3.723648029330889,
|
| 2353 |
+
"grad_norm": 0.5177659392356873,
|
| 2354 |
+
"learning_rate": 1.9747368421052633e-06,
|
| 2355 |
+
"loss": 0.0232,
|
| 2356 |
+
"step": 8125
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 3.7351054078826764,
|
| 2360 |
+
"grad_norm": 1.0480682849884033,
|
| 2361 |
+
"learning_rate": 1.948421052631579e-06,
|
| 2362 |
+
"loss": 0.0212,
|
| 2363 |
+
"step": 8150
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 3.7465627864344637,
|
| 2367 |
+
"grad_norm": 0.7136332988739014,
|
| 2368 |
+
"learning_rate": 1.922105263157895e-06,
|
| 2369 |
+
"loss": 0.0218,
|
| 2370 |
+
"step": 8175
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 3.758020164986251,
|
| 2374 |
+
"grad_norm": 1.165453314781189,
|
| 2375 |
+
"learning_rate": 1.8957894736842105e-06,
|
| 2376 |
+
"loss": 0.0189,
|
| 2377 |
+
"step": 8200
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 3.7694775435380388,
|
| 2381 |
+
"grad_norm": 0.8374521136283875,
|
| 2382 |
+
"learning_rate": 1.8694736842105265e-06,
|
| 2383 |
+
"loss": 0.0206,
|
| 2384 |
+
"step": 8225
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 3.780934922089826,
|
| 2388 |
+
"grad_norm": 0.6272546648979187,
|
| 2389 |
+
"learning_rate": 1.8431578947368424e-06,
|
| 2390 |
+
"loss": 0.0178,
|
| 2391 |
+
"step": 8250
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 3.7923923006416134,
|
| 2395 |
+
"grad_norm": 1.2483456134796143,
|
| 2396 |
+
"learning_rate": 1.816842105263158e-06,
|
| 2397 |
+
"loss": 0.023,
|
| 2398 |
+
"step": 8275
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 3.8038496791934007,
|
| 2402 |
+
"grad_norm": 0.8603928089141846,
|
| 2403 |
+
"learning_rate": 1.790526315789474e-06,
|
| 2404 |
+
"loss": 0.0219,
|
| 2405 |
+
"step": 8300
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 3.815307057745188,
|
| 2409 |
+
"grad_norm": 0.7172021269798279,
|
| 2410 |
+
"learning_rate": 1.7642105263157896e-06,
|
| 2411 |
+
"loss": 0.0205,
|
| 2412 |
+
"step": 8325
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 3.8267644362969753,
|
| 2416 |
+
"grad_norm": 0.9887197613716125,
|
| 2417 |
+
"learning_rate": 1.7378947368421054e-06,
|
| 2418 |
+
"loss": 0.02,
|
| 2419 |
+
"step": 8350
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 3.8382218148487626,
|
| 2423 |
+
"grad_norm": 1.3047144412994385,
|
| 2424 |
+
"learning_rate": 1.711578947368421e-06,
|
| 2425 |
+
"loss": 0.0194,
|
| 2426 |
+
"step": 8375
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 3.84967919340055,
|
| 2430 |
+
"grad_norm": 0.8804651498794556,
|
| 2431 |
+
"learning_rate": 1.685263157894737e-06,
|
| 2432 |
+
"loss": 0.0218,
|
| 2433 |
+
"step": 8400
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 3.861136571952337,
|
| 2437 |
+
"grad_norm": 0.8523846864700317,
|
| 2438 |
+
"learning_rate": 1.6589473684210527e-06,
|
| 2439 |
+
"loss": 0.0193,
|
| 2440 |
+
"step": 8425
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 3.8725939505041245,
|
| 2444 |
+
"grad_norm": 1.0217586755752563,
|
| 2445 |
+
"learning_rate": 1.6326315789473685e-06,
|
| 2446 |
+
"loss": 0.021,
|
| 2447 |
+
"step": 8450
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 3.8840513290559118,
|
| 2451 |
+
"grad_norm": 0.5876717567443848,
|
| 2452 |
+
"learning_rate": 1.6063157894736845e-06,
|
| 2453 |
+
"loss": 0.0213,
|
| 2454 |
+
"step": 8475
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 3.8955087076076995,
|
| 2458 |
+
"grad_norm": 0.8419985175132751,
|
| 2459 |
+
"learning_rate": 1.5800000000000001e-06,
|
| 2460 |
+
"loss": 0.0197,
|
| 2461 |
+
"step": 8500
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 3.906966086159487,
|
| 2465 |
+
"grad_norm": 1.1159563064575195,
|
| 2466 |
+
"learning_rate": 1.553684210526316e-06,
|
| 2467 |
+
"loss": 0.0221,
|
| 2468 |
+
"step": 8525
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 3.918423464711274,
|
| 2472 |
+
"grad_norm": 0.7730568051338196,
|
| 2473 |
+
"learning_rate": 1.5273684210526318e-06,
|
| 2474 |
+
"loss": 0.018,
|
| 2475 |
+
"step": 8550
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 3.9298808432630614,
|
| 2479 |
+
"grad_norm": 0.6262918710708618,
|
| 2480 |
+
"learning_rate": 1.5010526315789476e-06,
|
| 2481 |
+
"loss": 0.0206,
|
| 2482 |
+
"step": 8575
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"epoch": 3.9413382218148487,
|
| 2486 |
+
"grad_norm": 0.7903968095779419,
|
| 2487 |
+
"learning_rate": 1.4747368421052632e-06,
|
| 2488 |
+
"loss": 0.0236,
|
| 2489 |
+
"step": 8600
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 3.952795600366636,
|
| 2493 |
+
"grad_norm": 0.5999600291252136,
|
| 2494 |
+
"learning_rate": 1.448421052631579e-06,
|
| 2495 |
+
"loss": 0.0226,
|
| 2496 |
+
"step": 8625
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 3.9642529789184233,
|
| 2500 |
+
"grad_norm": 0.5939409732818604,
|
| 2501 |
+
"learning_rate": 1.422105263157895e-06,
|
| 2502 |
+
"loss": 0.0199,
|
| 2503 |
+
"step": 8650
|
| 2504 |
+
},
|
| 2505 |
+
{
|
| 2506 |
+
"epoch": 3.975710357470211,
|
| 2507 |
+
"grad_norm": 0.6602426767349243,
|
| 2508 |
+
"learning_rate": 1.3957894736842106e-06,
|
| 2509 |
+
"loss": 0.0198,
|
| 2510 |
+
"step": 8675
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 3.9871677360219984,
|
| 2514 |
+
"grad_norm": 0.6953304409980774,
|
| 2515 |
+
"learning_rate": 1.3694736842105262e-06,
|
| 2516 |
+
"loss": 0.0181,
|
| 2517 |
+
"step": 8700
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 3.9986251145737857,
|
| 2521 |
+
"grad_norm": 0.9689406156539917,
|
| 2522 |
+
"learning_rate": 1.3431578947368423e-06,
|
| 2523 |
+
"loss": 0.0216,
|
| 2524 |
+
"step": 8725
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 4.010082493125573,
|
| 2528 |
+
"grad_norm": 0.3220241367816925,
|
| 2529 |
+
"learning_rate": 1.316842105263158e-06,
|
| 2530 |
+
"loss": 0.0101,
|
| 2531 |
+
"step": 8750
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 4.02153987167736,
|
| 2535 |
+
"grad_norm": 0.28109627962112427,
|
| 2536 |
+
"learning_rate": 1.2905263157894737e-06,
|
| 2537 |
+
"loss": 0.0103,
|
| 2538 |
+
"step": 8775
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 4.032997250229148,
|
| 2542 |
+
"grad_norm": 0.43271881341934204,
|
| 2543 |
+
"learning_rate": 1.2642105263157895e-06,
|
| 2544 |
+
"loss": 0.0112,
|
| 2545 |
+
"step": 8800
|
| 2546 |
+
},
|
| 2547 |
+
{
|
| 2548 |
+
"epoch": 4.044454628780935,
|
| 2549 |
+
"grad_norm": 0.5643872022628784,
|
| 2550 |
+
"learning_rate": 1.2378947368421053e-06,
|
| 2551 |
+
"loss": 0.0103,
|
| 2552 |
+
"step": 8825
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 4.055912007332722,
|
| 2556 |
+
"grad_norm": 0.5255523324012756,
|
| 2557 |
+
"learning_rate": 1.2115789473684212e-06,
|
| 2558 |
+
"loss": 0.0094,
|
| 2559 |
+
"step": 8850
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 4.0673693858845095,
|
| 2563 |
+
"grad_norm": 0.5204137563705444,
|
| 2564 |
+
"learning_rate": 1.185263157894737e-06,
|
| 2565 |
+
"loss": 0.0107,
|
| 2566 |
+
"step": 8875
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 4.078826764436297,
|
| 2570 |
+
"grad_norm": 0.4249601364135742,
|
| 2571 |
+
"learning_rate": 1.1589473684210526e-06,
|
| 2572 |
+
"loss": 0.0101,
|
| 2573 |
+
"step": 8900
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 4.090284142988084,
|
| 2577 |
+
"grad_norm": 0.6326686143875122,
|
| 2578 |
+
"learning_rate": 1.1326315789473686e-06,
|
| 2579 |
+
"loss": 0.0091,
|
| 2580 |
+
"step": 8925
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 4.101741521539871,
|
| 2584 |
+
"grad_norm": 0.3784690499305725,
|
| 2585 |
+
"learning_rate": 1.1063157894736842e-06,
|
| 2586 |
+
"loss": 0.0091,
|
| 2587 |
+
"step": 8950
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 4.113198900091659,
|
| 2591 |
+
"grad_norm": 0.426418662071228,
|
| 2592 |
+
"learning_rate": 1.08e-06,
|
| 2593 |
+
"loss": 0.0094,
|
| 2594 |
+
"step": 8975
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 4.124656278643446,
|
| 2598 |
+
"grad_norm": 0.5608317852020264,
|
| 2599 |
+
"learning_rate": 1.0536842105263159e-06,
|
| 2600 |
+
"loss": 0.0093,
|
| 2601 |
+
"step": 9000
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 4.124656278643446,
|
| 2605 |
+
"eval_loss": 0.36179110407829285,
|
| 2606 |
+
"eval_runtime": 2107.0997,
|
| 2607 |
+
"eval_samples_per_second": 2.554,
|
| 2608 |
+
"eval_steps_per_second": 0.16,
|
| 2609 |
+
"eval_wer": 0.20167799912521372,
|
| 2610 |
+
"step": 9000
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 4.136113657195233,
|
| 2614 |
+
"grad_norm": 0.5352091789245605,
|
| 2615 |
+
"learning_rate": 1.0273684210526317e-06,
|
| 2616 |
+
"loss": 0.0104,
|
| 2617 |
+
"step": 9025
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 4.1475710357470215,
|
| 2621 |
+
"grad_norm": 0.40450939536094666,
|
| 2622 |
+
"learning_rate": 1.0010526315789475e-06,
|
| 2623 |
+
"loss": 0.0082,
|
| 2624 |
+
"step": 9050
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 4.159028414298809,
|
| 2628 |
+
"grad_norm": 0.5160329341888428,
|
| 2629 |
+
"learning_rate": 9.74736842105263e-07,
|
| 2630 |
+
"loss": 0.0083,
|
| 2631 |
+
"step": 9075
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 4.170485792850596,
|
| 2635 |
+
"grad_norm": 0.3774925768375397,
|
| 2636 |
+
"learning_rate": 9.48421052631579e-07,
|
| 2637 |
+
"loss": 0.0096,
|
| 2638 |
+
"step": 9100
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 4.181943171402383,
|
| 2642 |
+
"grad_norm": 0.4281198978424072,
|
| 2643 |
+
"learning_rate": 9.221052631578948e-07,
|
| 2644 |
+
"loss": 0.0099,
|
| 2645 |
+
"step": 9125
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 4.193400549954171,
|
| 2649 |
+
"grad_norm": 0.4555334746837616,
|
| 2650 |
+
"learning_rate": 8.957894736842106e-07,
|
| 2651 |
+
"loss": 0.011,
|
| 2652 |
+
"step": 9150
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 4.204857928505958,
|
| 2656 |
+
"grad_norm": 0.4731394350528717,
|
| 2657 |
+
"learning_rate": 8.694736842105264e-07,
|
| 2658 |
+
"loss": 0.0091,
|
| 2659 |
+
"step": 9175
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 4.216315307057745,
|
| 2663 |
+
"grad_norm": 0.5529298186302185,
|
| 2664 |
+
"learning_rate": 8.431578947368421e-07,
|
| 2665 |
+
"loss": 0.0101,
|
| 2666 |
+
"step": 9200
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 4.227772685609533,
|
| 2670 |
+
"grad_norm": 0.600356936454773,
|
| 2671 |
+
"learning_rate": 8.16842105263158e-07,
|
| 2672 |
+
"loss": 0.0084,
|
| 2673 |
+
"step": 9225
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 4.23923006416132,
|
| 2677 |
+
"grad_norm": 0.3486729562282562,
|
| 2678 |
+
"learning_rate": 7.905263157894738e-07,
|
| 2679 |
+
"loss": 0.0086,
|
| 2680 |
+
"step": 9250
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 4.250687442713107,
|
| 2684 |
+
"grad_norm": 0.38798272609710693,
|
| 2685 |
+
"learning_rate": 7.642105263157895e-07,
|
| 2686 |
+
"loss": 0.009,
|
| 2687 |
+
"step": 9275
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 4.2621448212648945,
|
| 2691 |
+
"grad_norm": 0.4498680830001831,
|
| 2692 |
+
"learning_rate": 7.378947368421054e-07,
|
| 2693 |
+
"loss": 0.0084,
|
| 2694 |
+
"step": 9300
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 4.273602199816682,
|
| 2698 |
+
"grad_norm": 0.4311501085758209,
|
| 2699 |
+
"learning_rate": 7.115789473684211e-07,
|
| 2700 |
+
"loss": 0.0084,
|
| 2701 |
+
"step": 9325
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 4.285059578368469,
|
| 2705 |
+
"grad_norm": 0.49663326144218445,
|
| 2706 |
+
"learning_rate": 6.852631578947369e-07,
|
| 2707 |
+
"loss": 0.0095,
|
| 2708 |
+
"step": 9350
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 4.296516956920256,
|
| 2712 |
+
"grad_norm": 0.6134785413742065,
|
| 2713 |
+
"learning_rate": 6.589473684210527e-07,
|
| 2714 |
+
"loss": 0.01,
|
| 2715 |
+
"step": 9375
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 4.307974335472044,
|
| 2719 |
+
"grad_norm": 0.4844453036785126,
|
| 2720 |
+
"learning_rate": 6.326315789473684e-07,
|
| 2721 |
+
"loss": 0.0109,
|
| 2722 |
+
"step": 9400
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 4.319431714023831,
|
| 2726 |
+
"grad_norm": 0.23624840378761292,
|
| 2727 |
+
"learning_rate": 6.063157894736842e-07,
|
| 2728 |
+
"loss": 0.008,
|
| 2729 |
+
"step": 9425
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 4.330889092575619,
|
| 2733 |
+
"grad_norm": 0.43371546268463135,
|
| 2734 |
+
"learning_rate": 5.800000000000001e-07,
|
| 2735 |
+
"loss": 0.009,
|
| 2736 |
+
"step": 9450
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 4.3423464711274065,
|
| 2740 |
+
"grad_norm": 0.3144814670085907,
|
| 2741 |
+
"learning_rate": 5.536842105263159e-07,
|
| 2742 |
+
"loss": 0.0113,
|
| 2743 |
+
"step": 9475
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 4.353803849679194,
|
| 2747 |
+
"grad_norm": 0.2468927651643753,
|
| 2748 |
+
"learning_rate": 5.273684210526316e-07,
|
| 2749 |
+
"loss": 0.0096,
|
| 2750 |
+
"step": 9500
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 4.365261228230981,
|
| 2754 |
+
"grad_norm": 0.4175880253314972,
|
| 2755 |
+
"learning_rate": 5.010526315789474e-07,
|
| 2756 |
+
"loss": 0.0101,
|
| 2757 |
+
"step": 9525
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 4.376718606782768,
|
| 2761 |
+
"grad_norm": 0.27920958399772644,
|
| 2762 |
+
"learning_rate": 4.747368421052632e-07,
|
| 2763 |
+
"loss": 0.0083,
|
| 2764 |
+
"step": 9550
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 4.388175985334556,
|
| 2768 |
+
"grad_norm": 0.4185403883457184,
|
| 2769 |
+
"learning_rate": 4.4842105263157894e-07,
|
| 2770 |
+
"loss": 0.0093,
|
| 2771 |
+
"step": 9575
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 4.399633363886343,
|
| 2775 |
+
"grad_norm": 0.35738441348075867,
|
| 2776 |
+
"learning_rate": 4.221052631578948e-07,
|
| 2777 |
+
"loss": 0.0091,
|
| 2778 |
+
"step": 9600
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 4.41109074243813,
|
| 2782 |
+
"grad_norm": 0.562145471572876,
|
| 2783 |
+
"learning_rate": 3.957894736842106e-07,
|
| 2784 |
+
"loss": 0.0099,
|
| 2785 |
+
"step": 9625
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 4.422548120989918,
|
| 2789 |
+
"grad_norm": 0.7420178055763245,
|
| 2790 |
+
"learning_rate": 3.6947368421052634e-07,
|
| 2791 |
+
"loss": 0.0096,
|
| 2792 |
+
"step": 9650
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 4.434005499541705,
|
| 2796 |
+
"grad_norm": 0.6182712912559509,
|
| 2797 |
+
"learning_rate": 3.431578947368421e-07,
|
| 2798 |
+
"loss": 0.009,
|
| 2799 |
+
"step": 9675
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 4.445462878093492,
|
| 2803 |
+
"grad_norm": 0.4601646065711975,
|
| 2804 |
+
"learning_rate": 3.1684210526315787e-07,
|
| 2805 |
+
"loss": 0.0093,
|
| 2806 |
+
"step": 9700
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 4.4569202566452795,
|
| 2810 |
+
"grad_norm": 0.4484894871711731,
|
| 2811 |
+
"learning_rate": 2.905263157894737e-07,
|
| 2812 |
+
"loss": 0.0093,
|
| 2813 |
+
"step": 9725
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 4.468377635197067,
|
| 2817 |
+
"grad_norm": 0.491921603679657,
|
| 2818 |
+
"learning_rate": 2.6421052631578946e-07,
|
| 2819 |
+
"loss": 0.0111,
|
| 2820 |
+
"step": 9750
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 4.479835013748854,
|
| 2824 |
+
"grad_norm": 0.565032422542572,
|
| 2825 |
+
"learning_rate": 2.378947368421053e-07,
|
| 2826 |
+
"loss": 0.009,
|
| 2827 |
+
"step": 9775
|
| 2828 |
+
},
|
| 2829 |
+
{
|
| 2830 |
+
"epoch": 4.491292392300641,
|
| 2831 |
+
"grad_norm": 0.39479389786720276,
|
| 2832 |
+
"learning_rate": 2.1157894736842107e-07,
|
| 2833 |
+
"loss": 0.009,
|
| 2834 |
+
"step": 9800
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 4.502749770852429,
|
| 2838 |
+
"grad_norm": 0.4282767176628113,
|
| 2839 |
+
"learning_rate": 1.8526315789473689e-07,
|
| 2840 |
+
"loss": 0.0088,
|
| 2841 |
+
"step": 9825
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 4.514207149404216,
|
| 2845 |
+
"grad_norm": 0.6674570441246033,
|
| 2846 |
+
"learning_rate": 1.5894736842105265e-07,
|
| 2847 |
+
"loss": 0.0083,
|
| 2848 |
+
"step": 9850
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 4.525664527956003,
|
| 2852 |
+
"grad_norm": 0.5396430492401123,
|
| 2853 |
+
"learning_rate": 1.3263157894736842e-07,
|
| 2854 |
+
"loss": 0.0087,
|
| 2855 |
+
"step": 9875
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 4.537121906507791,
|
| 2859 |
+
"grad_norm": 0.4620998501777649,
|
| 2860 |
+
"learning_rate": 1.0631578947368421e-07,
|
| 2861 |
+
"loss": 0.009,
|
| 2862 |
+
"step": 9900
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 4.548579285059578,
|
| 2866 |
+
"grad_norm": 0.6867234706878662,
|
| 2867 |
+
"learning_rate": 8e-08,
|
| 2868 |
+
"loss": 0.0093,
|
| 2869 |
+
"step": 9925
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 4.560036663611366,
|
| 2873 |
+
"grad_norm": 0.5100435018539429,
|
| 2874 |
+
"learning_rate": 5.3684210526315794e-08,
|
| 2875 |
+
"loss": 0.0114,
|
| 2876 |
+
"step": 9950
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 4.571494042163153,
|
| 2880 |
+
"grad_norm": 0.44454291462898254,
|
| 2881 |
+
"learning_rate": 2.736842105263158e-08,
|
| 2882 |
+
"loss": 0.0088,
|
| 2883 |
+
"step": 9975
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 4.582951420714941,
|
| 2887 |
+
"grad_norm": 0.47733160853385925,
|
| 2888 |
+
"learning_rate": 1.0526315789473685e-09,
|
| 2889 |
+
"loss": 0.0084,
|
| 2890 |
+
"step": 10000
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 4.582951420714941,
|
| 2894 |
+
"eval_loss": 0.36154302954673767,
|
| 2895 |
+
"eval_runtime": 2092.0122,
|
| 2896 |
+
"eval_samples_per_second": 2.572,
|
| 2897 |
+
"eval_steps_per_second": 0.161,
|
| 2898 |
+
"eval_wer": 0.2007833313451827,
|
| 2899 |
+
"step": 10000
|
| 2900 |
+
},
|
| 2901 |
+
{
|
| 2902 |
+
"epoch": 4.582951420714941,
|
| 2903 |
+
"step": 10000,
|
| 2904 |
+
"total_flos": 1.0870363781804851e+21,
|
| 2905 |
+
"train_loss": 0.0910786449804902,
|
| 2906 |
+
"train_runtime": 82447.7266,
|
| 2907 |
+
"train_samples_per_second": 3.881,
|
| 2908 |
+
"train_steps_per_second": 0.121
|
| 2909 |
}
|
| 2910 |
],
|
| 2911 |
"logging_steps": 25,
|
| 2912 |
+
"max_steps": 10000,
|
| 2913 |
"num_input_tokens_seen": 0,
|
| 2914 |
+
"num_train_epochs": 5,
|
| 2915 |
"save_steps": 1000,
|
| 2916 |
"stateful_callbacks": {
|
| 2917 |
"TrainerControl": {
|
|
|
|
| 2925 |
"attributes": {}
|
| 2926 |
}
|
| 2927 |
},
|
| 2928 |
+
"total_flos": 1.0870363781804851e+21,
|
| 2929 |
"train_batch_size": 16,
|
| 2930 |
"trial_name": null,
|
| 2931 |
"trial_params": null
|