End of training
Browse files- README.md +2 -1
- all_results.json +6 -6
- train_results.json +6 -6
- trainer_state.json +853 -874
- training_loss.png +0 -0
README.md
CHANGED
|
@@ -4,6 +4,7 @@ license: apache-2.0
|
|
| 4 |
base_model: Qwen/Qwen2.5-7B-Instruct
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
|
|
|
| 7 |
- generated_from_trainer
|
| 8 |
model-index:
|
| 9 |
- name: llama3-1_8b_4o_annotated_math
|
|
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# llama3-1_8b_4o_annotated_math
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
|
|
|
| 4 |
base_model: Qwen/Qwen2.5-7B-Instruct
|
| 5 |
tags:
|
| 6 |
- llama-factory
|
| 7 |
+
- full
|
| 8 |
- generated_from_trainer
|
| 9 |
model-index:
|
| 10 |
- name: llama3-1_8b_4o_annotated_math
|
|
|
|
| 16 |
|
| 17 |
# llama3-1_8b_4o_annotated_math
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on the mlfoundations-dev/4o_annotated_math dataset.
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.986175115207373,
|
| 3 |
+
"total_flos": 32031101919232.0,
|
| 4 |
+
"train_loss": 0.28372171659160544,
|
| 5 |
+
"train_runtime": 1010.9103,
|
| 6 |
+
"train_samples_per_second": 20.539,
|
| 7 |
+
"train_steps_per_second": 0.214
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 2.986175115207373,
|
| 3 |
+
"total_flos": 32031101919232.0,
|
| 4 |
+
"train_loss": 0.28372171659160544,
|
| 5 |
+
"train_runtime": 1010.9103,
|
| 6 |
+
"train_samples_per_second": 20.539,
|
| 7 |
+
"train_steps_per_second": 0.214
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1558 +1,1537 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
-
"epoch": 0.
|
| 13 |
-
"grad_norm": 3.
|
| 14 |
"learning_rate": 4.5454545454545457e-07,
|
| 15 |
-
"loss": 0.
|
| 16 |
"step": 1
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch": 0.
|
| 20 |
-
"grad_norm": 4.
|
| 21 |
"learning_rate": 9.090909090909091e-07,
|
| 22 |
-
"loss": 0.
|
| 23 |
"step": 2
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm": 3.
|
| 28 |
"learning_rate": 1.3636363636363636e-06,
|
| 29 |
-
"loss": 0.
|
| 30 |
"step": 3
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm": 3.
|
| 35 |
"learning_rate": 1.8181818181818183e-06,
|
| 36 |
-
"loss": 0.
|
| 37 |
"step": 4
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"grad_norm": 3.
|
| 42 |
"learning_rate": 2.2727272727272728e-06,
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 5
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"grad_norm": 2.
|
| 49 |
"learning_rate": 2.7272727272727272e-06,
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 6
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm": 2.
|
| 56 |
"learning_rate": 3.181818181818182e-06,
|
| 57 |
-
"loss": 0.
|
| 58 |
"step": 7
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm": 2.
|
| 63 |
"learning_rate": 3.6363636363636366e-06,
|
| 64 |
-
"loss": 0.
|
| 65 |
"step": 8
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm": 1.
|
| 70 |
"learning_rate": 4.0909090909090915e-06,
|
| 71 |
-
"loss": 0.
|
| 72 |
"step": 9
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm": 1.
|
| 77 |
"learning_rate": 4.5454545454545455e-06,
|
| 78 |
-
"loss": 0.
|
| 79 |
"step": 10
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm": 1.
|
| 84 |
"learning_rate": 5e-06,
|
| 85 |
-
"loss": 0.
|
| 86 |
"step": 11
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm": 1.
|
| 91 |
"learning_rate": 5.4545454545454545e-06,
|
| 92 |
-
"loss": 0.
|
| 93 |
"step": 12
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm": 1.
|
| 98 |
"learning_rate": 5.90909090909091e-06,
|
| 99 |
-
"loss": 0.
|
| 100 |
"step": 13
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm": 1.
|
| 105 |
"learning_rate": 6.363636363636364e-06,
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 14
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm": 1.
|
| 112 |
"learning_rate": 6.818181818181818e-06,
|
| 113 |
-
"loss": 0.
|
| 114 |
"step": 15
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm": 1.
|
| 119 |
"learning_rate": 7.272727272727273e-06,
|
| 120 |
-
"loss": 0.
|
| 121 |
"step": 16
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"grad_norm": 1.
|
| 126 |
"learning_rate": 7.727272727272727e-06,
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 17
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm": 1.
|
| 133 |
"learning_rate": 8.181818181818183e-06,
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 18
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"grad_norm": 0.
|
| 140 |
"learning_rate": 8.636363636363637e-06,
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 19
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch": 0.
|
| 146 |
-
"grad_norm": 0.
|
| 147 |
"learning_rate": 9.090909090909091e-06,
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 20
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"grad_norm": 0.
|
| 154 |
"learning_rate": 9.545454545454547e-06,
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 21
|
| 157 |
},
|
| 158 |
{
|
| 159 |
-
"epoch": 0.
|
| 160 |
-
"grad_norm": 0.
|
| 161 |
"learning_rate": 1e-05,
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 22
|
| 164 |
},
|
| 165 |
{
|
| 166 |
-
"epoch": 0.
|
| 167 |
-
"grad_norm":
|
| 168 |
-
"learning_rate": 9.
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 23
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm": 0.
|
| 175 |
-
"learning_rate": 9.
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 24
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 0.
|
| 181 |
-
"grad_norm": 0.
|
| 182 |
-
"learning_rate": 9.
|
| 183 |
-
"loss": 0.
|
| 184 |
"step": 25
|
| 185 |
},
|
| 186 |
{
|
| 187 |
-
"epoch": 0.
|
| 188 |
-
"grad_norm": 0.
|
| 189 |
-
"learning_rate": 9.
|
| 190 |
-
"loss": 0.
|
| 191 |
"step": 26
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch": 0.
|
| 195 |
-
"grad_norm": 0.
|
| 196 |
-
"learning_rate": 9.
|
| 197 |
-
"loss": 0.
|
| 198 |
"step": 27
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch": 0.
|
| 202 |
-
"grad_norm": 0.
|
| 203 |
-
"learning_rate": 9.
|
| 204 |
-
"loss": 0.
|
| 205 |
"step": 28
|
| 206 |
},
|
| 207 |
{
|
| 208 |
-
"epoch": 0.
|
| 209 |
-
"grad_norm": 0.
|
| 210 |
-
"learning_rate": 9.
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 29
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch": 0.
|
| 216 |
-
"grad_norm": 0.
|
| 217 |
-
"learning_rate": 9.
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 30
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"grad_norm": 0.
|
| 224 |
-
"learning_rate": 9.
|
| 225 |
-
"loss": 0.
|
| 226 |
"step": 31
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch": 0.
|
| 230 |
-
"grad_norm": 0.
|
| 231 |
-
"learning_rate": 9.
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 32
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch": 0.
|
| 237 |
-
"grad_norm": 0.
|
| 238 |
-
"learning_rate": 9.
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 33
|
| 241 |
},
|
| 242 |
{
|
| 243 |
-
"epoch": 0.
|
| 244 |
-
"grad_norm": 0.
|
| 245 |
-
"learning_rate": 9.
|
| 246 |
-
"loss": 0.
|
| 247 |
"step": 34
|
| 248 |
},
|
| 249 |
{
|
| 250 |
-
"epoch": 0.
|
| 251 |
-
"grad_norm": 0.
|
| 252 |
-
"learning_rate": 9.
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 35
|
| 255 |
},
|
| 256 |
{
|
| 257 |
-
"epoch": 0.
|
| 258 |
-
"grad_norm": 0.
|
| 259 |
-
"learning_rate": 9.
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 36
|
| 262 |
},
|
| 263 |
{
|
| 264 |
-
"epoch": 0.
|
| 265 |
-
"grad_norm": 0.
|
| 266 |
-
"learning_rate": 9.
|
| 267 |
-
"loss": 0.
|
| 268 |
"step": 37
|
| 269 |
},
|
| 270 |
{
|
| 271 |
-
"epoch": 0.
|
| 272 |
-
"grad_norm": 0.
|
| 273 |
-
"learning_rate": 9.
|
| 274 |
-
"loss": 0.
|
| 275 |
"step": 38
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch": 0.
|
| 279 |
-
"grad_norm": 0.
|
| 280 |
-
"learning_rate": 9.
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 39
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch": 0.
|
| 286 |
-
"grad_norm": 0.
|
| 287 |
-
"learning_rate": 9.
|
| 288 |
-
"loss": 0.
|
| 289 |
"step": 40
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"epoch": 0.
|
| 293 |
-
"grad_norm": 0.
|
| 294 |
-
"learning_rate": 9.
|
| 295 |
"loss": 0.3305,
|
| 296 |
"step": 41
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"epoch": 0.
|
| 300 |
-
"grad_norm": 0.
|
| 301 |
-
"learning_rate": 9.
|
| 302 |
-
"loss": 0.
|
| 303 |
"step": 42
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"epoch": 0.
|
| 307 |
-
"grad_norm": 0.
|
| 308 |
-
"learning_rate": 9.
|
| 309 |
-
"loss": 0.
|
| 310 |
"step": 43
|
| 311 |
},
|
| 312 |
{
|
| 313 |
-
"epoch": 0.
|
| 314 |
-
"grad_norm": 0.
|
| 315 |
-
"learning_rate": 9.
|
| 316 |
-
"loss": 0.
|
| 317 |
"step": 44
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch": 0.
|
| 321 |
-
"grad_norm": 0.
|
| 322 |
-
"learning_rate": 9.
|
| 323 |
-
"loss": 0.
|
| 324 |
"step": 45
|
| 325 |
},
|
| 326 |
{
|
| 327 |
-
"epoch": 0.
|
| 328 |
-
"grad_norm": 0.
|
| 329 |
-
"learning_rate": 9.
|
| 330 |
-
"loss": 0.
|
| 331 |
"step": 46
|
| 332 |
},
|
| 333 |
{
|
| 334 |
-
"epoch": 0.
|
| 335 |
-
"grad_norm": 0.
|
| 336 |
-
"learning_rate": 9.
|
| 337 |
-
"loss": 0.
|
| 338 |
"step": 47
|
| 339 |
},
|
| 340 |
{
|
| 341 |
-
"epoch": 0.
|
| 342 |
-
"grad_norm": 0.
|
| 343 |
-
"learning_rate": 9.
|
| 344 |
-
"loss": 0.
|
| 345 |
"step": 48
|
| 346 |
},
|
| 347 |
{
|
| 348 |
-
"epoch": 0.
|
| 349 |
-
"grad_norm": 0.
|
| 350 |
-
"learning_rate": 9.
|
| 351 |
-
"loss": 0.
|
| 352 |
"step": 49
|
| 353 |
},
|
| 354 |
{
|
| 355 |
-
"epoch": 0.
|
| 356 |
-
"grad_norm": 0.
|
| 357 |
-
"learning_rate": 9.
|
| 358 |
-
"loss": 0.
|
| 359 |
"step": 50
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch": 0.
|
| 363 |
-
"grad_norm": 0.
|
| 364 |
-
"learning_rate": 9.
|
| 365 |
-
"loss": 0.
|
| 366 |
"step": 51
|
| 367 |
},
|
| 368 |
{
|
| 369 |
-
"epoch": 0.
|
| 370 |
-
"grad_norm": 0.
|
| 371 |
-
"learning_rate": 9.
|
| 372 |
-
"loss": 0.
|
| 373 |
"step": 52
|
| 374 |
},
|
| 375 |
{
|
| 376 |
-
"epoch": 0.
|
| 377 |
-
"grad_norm": 0.
|
| 378 |
-
"learning_rate": 9.
|
| 379 |
-
"loss": 0.
|
| 380 |
"step": 53
|
| 381 |
},
|
| 382 |
{
|
| 383 |
-
"epoch": 0.
|
| 384 |
-
"grad_norm": 0.
|
| 385 |
-
"learning_rate": 9.
|
| 386 |
-
"loss": 0.
|
| 387 |
"step": 54
|
| 388 |
},
|
| 389 |
{
|
| 390 |
-
"epoch": 0.
|
| 391 |
-
"grad_norm": 0.
|
| 392 |
-
"learning_rate": 9.
|
| 393 |
-
"loss": 0.
|
| 394 |
"step": 55
|
| 395 |
},
|
| 396 |
{
|
| 397 |
-
"epoch": 0.
|
| 398 |
-
"grad_norm": 0.
|
| 399 |
-
"learning_rate": 9.
|
| 400 |
-
"loss": 0.
|
| 401 |
"step": 56
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch": 0.
|
| 405 |
-
"grad_norm": 0.
|
| 406 |
-
"learning_rate": 9.
|
| 407 |
-
"loss": 0.
|
| 408 |
"step": 57
|
| 409 |
},
|
| 410 |
{
|
| 411 |
-
"epoch": 0.
|
| 412 |
-
"grad_norm": 0.
|
| 413 |
-
"learning_rate": 9.
|
| 414 |
-
"loss": 0.
|
| 415 |
"step": 58
|
| 416 |
},
|
| 417 |
{
|
| 418 |
-
"epoch": 0.
|
| 419 |
-
"grad_norm": 0.
|
| 420 |
-
"learning_rate": 9.
|
| 421 |
-
"loss": 0.
|
| 422 |
"step": 59
|
| 423 |
},
|
| 424 |
{
|
| 425 |
-
"epoch": 0.
|
| 426 |
-
"grad_norm": 0.
|
| 427 |
-
"learning_rate": 9.
|
| 428 |
-
"loss": 0.
|
| 429 |
"step": 60
|
| 430 |
},
|
| 431 |
{
|
| 432 |
-
"epoch": 0.
|
| 433 |
-
"grad_norm": 0.
|
| 434 |
-
"learning_rate": 9.
|
| 435 |
-
"loss": 0.
|
| 436 |
"step": 61
|
| 437 |
},
|
| 438 |
{
|
| 439 |
-
"epoch": 0.
|
| 440 |
-
"grad_norm": 0.
|
| 441 |
-
"learning_rate":
|
| 442 |
-
"loss": 0.
|
| 443 |
"step": 62
|
| 444 |
},
|
| 445 |
{
|
| 446 |
-
"epoch": 0.
|
| 447 |
-
"grad_norm": 0.
|
| 448 |
-
"learning_rate": 8.
|
| 449 |
-
"loss": 0.
|
| 450 |
"step": 63
|
| 451 |
},
|
| 452 |
{
|
| 453 |
-
"epoch": 0.
|
| 454 |
-
"grad_norm": 0.
|
| 455 |
-
"learning_rate": 8.
|
| 456 |
-
"loss": 0.
|
| 457 |
"step": 64
|
| 458 |
},
|
| 459 |
{
|
| 460 |
-
"epoch": 0.
|
| 461 |
-
"grad_norm": 0.
|
| 462 |
-
"learning_rate": 8.
|
| 463 |
-
"loss": 0.
|
| 464 |
"step": 65
|
| 465 |
},
|
| 466 |
{
|
| 467 |
-
"epoch": 0.
|
| 468 |
-
"grad_norm": 0.
|
| 469 |
-
"learning_rate": 8.
|
| 470 |
-
"loss": 0.
|
| 471 |
"step": 66
|
| 472 |
},
|
| 473 |
{
|
| 474 |
-
"epoch": 0.
|
| 475 |
-
"grad_norm": 0.
|
| 476 |
-
"learning_rate": 8.
|
| 477 |
-
"loss": 0.
|
| 478 |
"step": 67
|
| 479 |
},
|
| 480 |
{
|
| 481 |
-
"epoch": 0.
|
| 482 |
-
"grad_norm": 0.
|
| 483 |
-
"learning_rate": 8.
|
| 484 |
-
"loss": 0.
|
| 485 |
"step": 68
|
| 486 |
},
|
| 487 |
{
|
| 488 |
-
"epoch": 0.
|
| 489 |
-
"grad_norm": 0.
|
| 490 |
-
"learning_rate": 8.
|
| 491 |
-
"loss": 0.
|
| 492 |
"step": 69
|
| 493 |
},
|
| 494 |
{
|
| 495 |
-
"epoch": 0.
|
| 496 |
-
"grad_norm": 0.
|
| 497 |
-
"learning_rate": 8.
|
| 498 |
-
"loss": 0.
|
| 499 |
"step": 70
|
| 500 |
},
|
| 501 |
{
|
| 502 |
-
"epoch": 0.
|
| 503 |
-
"grad_norm": 0.
|
| 504 |
-
"learning_rate": 8.
|
| 505 |
-
"loss": 0.
|
| 506 |
"step": 71
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch": 0.
|
| 510 |
-
"grad_norm": 0.
|
| 511 |
-
"learning_rate": 8.
|
| 512 |
-
"loss": 0.
|
| 513 |
"step": 72
|
| 514 |
},
|
| 515 |
{
|
| 516 |
-
"epoch": 1.
|
| 517 |
-
"grad_norm":
|
| 518 |
-
"learning_rate": 8.
|
| 519 |
-
"loss": 0.
|
| 520 |
"step": 73
|
| 521 |
},
|
| 522 |
{
|
| 523 |
-
"epoch": 1.
|
| 524 |
-
"grad_norm": 0.
|
| 525 |
-
"learning_rate": 8.
|
| 526 |
-
"loss": 0.
|
| 527 |
"step": 74
|
| 528 |
},
|
| 529 |
{
|
| 530 |
-
"epoch": 1.
|
| 531 |
-
"grad_norm": 0.
|
| 532 |
-
"learning_rate": 8.
|
| 533 |
-
"loss": 0.
|
| 534 |
"step": 75
|
| 535 |
},
|
| 536 |
{
|
| 537 |
-
"epoch": 1.
|
| 538 |
-
"grad_norm": 0.
|
| 539 |
-
"learning_rate": 8.
|
| 540 |
-
"loss": 0.
|
| 541 |
"step": 76
|
| 542 |
},
|
| 543 |
{
|
| 544 |
-
"epoch": 1.
|
| 545 |
-
"grad_norm": 0.
|
| 546 |
-
"learning_rate": 8.
|
| 547 |
-
"loss": 0.
|
| 548 |
"step": 77
|
| 549 |
},
|
| 550 |
{
|
| 551 |
-
"epoch": 1.
|
| 552 |
-
"grad_norm": 0.
|
| 553 |
-
"learning_rate": 8.
|
| 554 |
-
"loss": 0.
|
| 555 |
"step": 78
|
| 556 |
},
|
| 557 |
{
|
| 558 |
-
"epoch": 1.
|
| 559 |
-
"grad_norm": 0.
|
| 560 |
-
"learning_rate": 8.
|
| 561 |
-
"loss": 0.
|
| 562 |
"step": 79
|
| 563 |
},
|
| 564 |
{
|
| 565 |
-
"epoch": 1.
|
| 566 |
-
"grad_norm": 0.
|
| 567 |
-
"learning_rate":
|
| 568 |
-
"loss": 0.
|
| 569 |
"step": 80
|
| 570 |
},
|
| 571 |
{
|
| 572 |
-
"epoch": 1.
|
| 573 |
-
"grad_norm": 0.
|
| 574 |
-
"learning_rate": 7.
|
| 575 |
-
"loss": 0.
|
| 576 |
"step": 81
|
| 577 |
},
|
| 578 |
{
|
| 579 |
-
"epoch": 1.
|
| 580 |
-
"grad_norm": 0.
|
| 581 |
-
"learning_rate": 7.
|
| 582 |
-
"loss": 0.
|
| 583 |
"step": 82
|
| 584 |
},
|
| 585 |
{
|
| 586 |
-
"epoch": 1.
|
| 587 |
-
"grad_norm": 0.
|
| 588 |
-
"learning_rate": 7.
|
| 589 |
-
"loss": 0.
|
| 590 |
"step": 83
|
| 591 |
},
|
| 592 |
{
|
| 593 |
-
"epoch": 1.
|
| 594 |
-
"grad_norm": 0.
|
| 595 |
-
"learning_rate": 7.
|
| 596 |
-
"loss": 0.
|
| 597 |
"step": 84
|
| 598 |
},
|
| 599 |
{
|
| 600 |
-
"epoch": 1.
|
| 601 |
-
"grad_norm": 0.
|
| 602 |
-
"learning_rate": 7.
|
| 603 |
-
"loss": 0.
|
| 604 |
"step": 85
|
| 605 |
},
|
| 606 |
{
|
| 607 |
-
"epoch": 1.
|
| 608 |
-
"grad_norm": 0.
|
| 609 |
-
"learning_rate": 7.
|
| 610 |
-
"loss": 0.
|
| 611 |
"step": 86
|
| 612 |
},
|
| 613 |
{
|
| 614 |
-
"epoch": 1.
|
| 615 |
-
"grad_norm": 0.
|
| 616 |
-
"learning_rate": 7.
|
| 617 |
-
"loss": 0.
|
| 618 |
"step": 87
|
| 619 |
},
|
| 620 |
{
|
| 621 |
-
"epoch": 1.
|
| 622 |
-
"grad_norm": 0.
|
| 623 |
-
"learning_rate": 7.
|
| 624 |
-
"loss": 0.
|
| 625 |
"step": 88
|
| 626 |
},
|
| 627 |
{
|
| 628 |
-
"epoch": 1.
|
| 629 |
-
"grad_norm": 0.
|
| 630 |
-
"learning_rate": 7.
|
| 631 |
-
"loss": 0.
|
| 632 |
"step": 89
|
| 633 |
},
|
| 634 |
{
|
| 635 |
-
"epoch": 1.
|
| 636 |
-
"grad_norm": 0.
|
| 637 |
-
"learning_rate": 7.
|
| 638 |
-
"loss": 0.
|
| 639 |
"step": 90
|
| 640 |
},
|
| 641 |
{
|
| 642 |
-
"epoch": 1.
|
| 643 |
-
"grad_norm": 0.
|
| 644 |
-
"learning_rate": 7.
|
| 645 |
-
"loss": 0.
|
| 646 |
"step": 91
|
| 647 |
},
|
| 648 |
{
|
| 649 |
-
"epoch": 1.
|
| 650 |
-
"grad_norm": 0.
|
| 651 |
-
"learning_rate": 7.
|
| 652 |
-
"loss": 0.
|
| 653 |
"step": 92
|
| 654 |
},
|
| 655 |
{
|
| 656 |
-
"epoch": 1.
|
| 657 |
-
"grad_norm": 0.
|
| 658 |
-
"learning_rate": 7.
|
| 659 |
-
"loss": 0.
|
| 660 |
"step": 93
|
| 661 |
},
|
| 662 |
{
|
| 663 |
-
"epoch": 1.
|
| 664 |
-
"grad_norm": 0.
|
| 665 |
-
"learning_rate":
|
| 666 |
-
"loss": 0.
|
| 667 |
"step": 94
|
| 668 |
},
|
| 669 |
{
|
| 670 |
-
"epoch": 1.
|
| 671 |
-
"grad_norm": 0.
|
| 672 |
-
"learning_rate": 6.
|
| 673 |
-
"loss": 0.
|
| 674 |
"step": 95
|
| 675 |
},
|
| 676 |
{
|
| 677 |
-
"epoch": 1.
|
| 678 |
-
"grad_norm": 0.
|
| 679 |
-
"learning_rate": 6.
|
| 680 |
-
"loss": 0.
|
| 681 |
"step": 96
|
| 682 |
},
|
| 683 |
{
|
| 684 |
-
"epoch": 1.
|
| 685 |
-
"grad_norm": 0.
|
| 686 |
-
"learning_rate": 6.
|
| 687 |
-
"loss": 0.
|
| 688 |
"step": 97
|
| 689 |
},
|
| 690 |
{
|
| 691 |
-
"epoch": 1.
|
| 692 |
-
"grad_norm": 0.
|
| 693 |
-
"learning_rate": 6.
|
| 694 |
-
"loss": 0.
|
| 695 |
"step": 98
|
| 696 |
},
|
| 697 |
{
|
| 698 |
-
"epoch": 1.
|
| 699 |
-
"grad_norm": 0.
|
| 700 |
-
"learning_rate": 6.
|
| 701 |
-
"loss": 0.
|
| 702 |
"step": 99
|
| 703 |
},
|
| 704 |
{
|
| 705 |
-
"epoch": 1.
|
| 706 |
-
"grad_norm": 0.
|
| 707 |
-
"learning_rate": 6.
|
| 708 |
-
"loss": 0.
|
| 709 |
"step": 100
|
| 710 |
},
|
| 711 |
{
|
| 712 |
-
"epoch": 1.
|
| 713 |
-
"grad_norm": 0.
|
| 714 |
-
"learning_rate": 6.
|
| 715 |
-
"loss": 0.
|
| 716 |
"step": 101
|
| 717 |
},
|
| 718 |
{
|
| 719 |
-
"epoch": 1.
|
| 720 |
-
"grad_norm": 0.
|
| 721 |
-
"learning_rate": 6.
|
| 722 |
-
"loss": 0.
|
| 723 |
"step": 102
|
| 724 |
},
|
| 725 |
{
|
| 726 |
-
"epoch": 1.
|
| 727 |
-
"grad_norm": 0.
|
| 728 |
-
"learning_rate": 6.
|
| 729 |
-
"loss": 0.
|
| 730 |
"step": 103
|
| 731 |
},
|
| 732 |
{
|
| 733 |
-
"epoch": 1.
|
| 734 |
-
"grad_norm": 0.
|
| 735 |
-
"learning_rate": 6.
|
| 736 |
-
"loss": 0.
|
| 737 |
"step": 104
|
| 738 |
},
|
| 739 |
{
|
| 740 |
-
"epoch": 1.
|
| 741 |
-
"grad_norm": 0.
|
| 742 |
-
"learning_rate": 6.
|
| 743 |
-
"loss": 0.
|
| 744 |
"step": 105
|
| 745 |
},
|
| 746 |
{
|
| 747 |
-
"epoch": 1.
|
| 748 |
-
"grad_norm": 0.
|
| 749 |
-
"learning_rate": 6.
|
| 750 |
-
"loss": 0.
|
| 751 |
"step": 106
|
| 752 |
},
|
| 753 |
{
|
| 754 |
-
"epoch": 1.
|
| 755 |
-
"grad_norm": 0.
|
| 756 |
-
"learning_rate":
|
| 757 |
-
"loss": 0.
|
| 758 |
"step": 107
|
| 759 |
},
|
| 760 |
{
|
| 761 |
-
"epoch": 1.
|
| 762 |
-
"grad_norm": 0.
|
| 763 |
-
"learning_rate": 5.
|
| 764 |
-
"loss": 0.
|
| 765 |
"step": 108
|
| 766 |
},
|
| 767 |
{
|
| 768 |
-
"epoch": 1.
|
| 769 |
-
"grad_norm": 0.
|
| 770 |
-
"learning_rate": 5.
|
| 771 |
-
"loss": 0.
|
| 772 |
"step": 109
|
| 773 |
},
|
| 774 |
{
|
| 775 |
-
"epoch": 1.
|
| 776 |
-
"grad_norm": 0.
|
| 777 |
-
"learning_rate": 5.
|
| 778 |
-
"loss": 0.
|
| 779 |
"step": 110
|
| 780 |
},
|
| 781 |
{
|
| 782 |
-
"epoch": 1.
|
| 783 |
-
"grad_norm": 0.
|
| 784 |
-
"learning_rate": 5.
|
| 785 |
-
"loss": 0.
|
| 786 |
"step": 111
|
| 787 |
},
|
| 788 |
{
|
| 789 |
-
"epoch": 1.
|
| 790 |
-
"grad_norm": 0.
|
| 791 |
-
"learning_rate": 5.
|
| 792 |
-
"loss": 0.
|
| 793 |
"step": 112
|
| 794 |
},
|
| 795 |
{
|
| 796 |
-
"epoch": 1.
|
| 797 |
-
"grad_norm": 0.
|
| 798 |
-
"learning_rate": 5.
|
| 799 |
-
"loss": 0.
|
| 800 |
"step": 113
|
| 801 |
},
|
| 802 |
{
|
| 803 |
-
"epoch": 1.
|
| 804 |
-
"grad_norm": 0.
|
| 805 |
-
"learning_rate": 5.
|
| 806 |
-
"loss": 0.
|
| 807 |
"step": 114
|
| 808 |
},
|
| 809 |
{
|
| 810 |
-
"epoch": 1.
|
| 811 |
-
"grad_norm": 0.
|
| 812 |
-
"learning_rate": 5.
|
| 813 |
-
"loss": 0.
|
| 814 |
"step": 115
|
| 815 |
},
|
| 816 |
{
|
| 817 |
-
"epoch": 1.
|
| 818 |
-
"grad_norm": 0.
|
| 819 |
-
"learning_rate": 5.
|
| 820 |
-
"loss": 0.
|
| 821 |
"step": 116
|
| 822 |
},
|
| 823 |
{
|
| 824 |
-
"epoch": 1.
|
| 825 |
-
"grad_norm": 0.
|
| 826 |
-
"learning_rate": 5.
|
| 827 |
-
"loss": 0.
|
| 828 |
"step": 117
|
| 829 |
},
|
| 830 |
{
|
| 831 |
-
"epoch": 1.
|
| 832 |
-
"grad_norm": 0.
|
| 833 |
-
"learning_rate": 5.
|
| 834 |
-
"loss": 0.
|
| 835 |
"step": 118
|
| 836 |
},
|
| 837 |
{
|
| 838 |
-
"epoch": 1.
|
| 839 |
-
"grad_norm": 0.
|
| 840 |
-
"learning_rate":
|
| 841 |
-
"loss": 0.
|
| 842 |
"step": 119
|
| 843 |
},
|
| 844 |
{
|
| 845 |
-
"epoch": 1.
|
| 846 |
-
"grad_norm": 0.
|
| 847 |
-
"learning_rate":
|
| 848 |
-
"loss": 0.
|
| 849 |
"step": 120
|
| 850 |
},
|
| 851 |
{
|
| 852 |
-
"epoch": 1.
|
| 853 |
-
"grad_norm": 0.
|
| 854 |
-
"learning_rate": 4.
|
| 855 |
-
"loss": 0.
|
| 856 |
"step": 121
|
| 857 |
},
|
| 858 |
{
|
| 859 |
-
"epoch": 1.
|
| 860 |
-
"grad_norm": 0.
|
| 861 |
-
"learning_rate": 4.
|
| 862 |
-
"loss": 0.
|
| 863 |
"step": 122
|
| 864 |
},
|
| 865 |
{
|
| 866 |
-
"epoch": 1.
|
| 867 |
-
"grad_norm": 0.
|
| 868 |
-
"learning_rate": 4.
|
| 869 |
-
"loss": 0.
|
| 870 |
"step": 123
|
| 871 |
},
|
| 872 |
{
|
| 873 |
-
"epoch": 1.
|
| 874 |
-
"grad_norm": 0.
|
| 875 |
-
"learning_rate": 4.
|
| 876 |
-
"loss": 0.
|
| 877 |
"step": 124
|
| 878 |
},
|
| 879 |
{
|
| 880 |
-
"epoch": 1.
|
| 881 |
-
"grad_norm": 0.
|
| 882 |
-
"learning_rate": 4.
|
| 883 |
-
"loss": 0.
|
| 884 |
"step": 125
|
| 885 |
},
|
| 886 |
{
|
| 887 |
-
"epoch": 1.
|
| 888 |
-
"grad_norm": 0.
|
| 889 |
-
"learning_rate": 4.
|
| 890 |
-
"loss": 0.
|
| 891 |
"step": 126
|
| 892 |
},
|
| 893 |
{
|
| 894 |
-
"epoch": 1.
|
| 895 |
-
"grad_norm": 0.
|
| 896 |
-
"learning_rate": 4.
|
| 897 |
-
"loss": 0.
|
| 898 |
"step": 127
|
| 899 |
},
|
| 900 |
{
|
| 901 |
-
"epoch": 1.
|
| 902 |
-
"grad_norm": 0.
|
| 903 |
-
"learning_rate": 4.
|
| 904 |
-
"loss": 0.
|
| 905 |
"step": 128
|
| 906 |
},
|
| 907 |
{
|
| 908 |
-
"epoch": 1.
|
| 909 |
-
"grad_norm": 0.
|
| 910 |
-
"learning_rate": 4.
|
| 911 |
-
"loss": 0.
|
| 912 |
"step": 129
|
| 913 |
},
|
| 914 |
{
|
| 915 |
-
"epoch": 1.
|
| 916 |
-
"grad_norm": 0.
|
| 917 |
-
"learning_rate": 4.
|
| 918 |
-
"loss": 0.
|
| 919 |
"step": 130
|
| 920 |
},
|
| 921 |
{
|
| 922 |
-
"epoch": 1.
|
| 923 |
-
"grad_norm": 0.
|
| 924 |
-
"learning_rate": 4.
|
| 925 |
-
"loss": 0.
|
| 926 |
"step": 131
|
| 927 |
},
|
| 928 |
{
|
| 929 |
-
"epoch": 1.
|
| 930 |
-
"grad_norm": 0.
|
| 931 |
-
"learning_rate":
|
| 932 |
-
"loss": 0.
|
| 933 |
"step": 132
|
| 934 |
},
|
| 935 |
{
|
| 936 |
-
"epoch": 1.
|
| 937 |
-
"grad_norm": 0.
|
| 938 |
-
"learning_rate":
|
| 939 |
-
"loss": 0.
|
| 940 |
"step": 133
|
| 941 |
},
|
| 942 |
{
|
| 943 |
-
"epoch": 1.
|
| 944 |
-
"grad_norm": 0.
|
| 945 |
-
"learning_rate": 3.
|
| 946 |
-
"loss": 0.
|
| 947 |
"step": 134
|
| 948 |
},
|
| 949 |
{
|
| 950 |
-
"epoch": 1.
|
| 951 |
-
"grad_norm": 0.
|
| 952 |
-
"learning_rate": 3.
|
| 953 |
-
"loss": 0.
|
| 954 |
"step": 135
|
| 955 |
},
|
| 956 |
{
|
| 957 |
-
"epoch": 1.
|
| 958 |
-
"grad_norm": 0.
|
| 959 |
-
"learning_rate": 3.
|
| 960 |
-
"loss": 0.
|
| 961 |
"step": 136
|
| 962 |
},
|
| 963 |
{
|
| 964 |
-
"epoch": 1.
|
| 965 |
-
"grad_norm": 0.
|
| 966 |
-
"learning_rate": 3.
|
| 967 |
-
"loss": 0.
|
| 968 |
"step": 137
|
| 969 |
},
|
| 970 |
{
|
| 971 |
-
"epoch": 1.
|
| 972 |
-
"grad_norm": 0.
|
| 973 |
-
"learning_rate": 3.
|
| 974 |
-
"loss": 0.
|
| 975 |
"step": 138
|
| 976 |
},
|
| 977 |
{
|
| 978 |
-
"epoch": 1.
|
| 979 |
-
"grad_norm": 0.
|
| 980 |
-
"learning_rate": 3.
|
| 981 |
-
"loss": 0.
|
| 982 |
"step": 139
|
| 983 |
},
|
| 984 |
{
|
| 985 |
-
"epoch": 1.
|
| 986 |
-
"grad_norm": 0.
|
| 987 |
-
"learning_rate": 3.
|
| 988 |
-
"loss": 0.
|
| 989 |
"step": 140
|
| 990 |
},
|
| 991 |
{
|
| 992 |
-
"epoch": 1.
|
| 993 |
-
"grad_norm": 0.
|
| 994 |
-
"learning_rate": 3.
|
| 995 |
-
"loss": 0.
|
| 996 |
"step": 141
|
| 997 |
},
|
| 998 |
{
|
| 999 |
-
"epoch": 1.
|
| 1000 |
-
"grad_norm": 0.
|
| 1001 |
-
"learning_rate": 3.
|
| 1002 |
-
"loss": 0.
|
| 1003 |
"step": 142
|
| 1004 |
},
|
| 1005 |
{
|
| 1006 |
-
"epoch": 1.
|
| 1007 |
-
"grad_norm": 0.
|
| 1008 |
-
"learning_rate": 3.
|
| 1009 |
-
"loss": 0.
|
| 1010 |
"step": 143
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
-
"epoch": 1.
|
| 1014 |
-
"grad_norm": 0.
|
| 1015 |
-
"learning_rate": 3.
|
| 1016 |
-
"loss": 0.
|
| 1017 |
"step": 144
|
| 1018 |
},
|
| 1019 |
{
|
| 1020 |
-
"epoch":
|
| 1021 |
-
"grad_norm": 0.
|
| 1022 |
-
"learning_rate":
|
| 1023 |
-
"loss": 0.
|
| 1024 |
"step": 145
|
| 1025 |
},
|
| 1026 |
{
|
| 1027 |
-
"epoch": 2.
|
| 1028 |
-
"grad_norm": 0.
|
| 1029 |
-
"learning_rate":
|
| 1030 |
-
"loss": 0.
|
| 1031 |
"step": 146
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
-
"epoch": 2.
|
| 1035 |
-
"grad_norm": 0.
|
| 1036 |
-
"learning_rate": 2.
|
| 1037 |
-
"loss": 0.
|
| 1038 |
"step": 147
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
-
"epoch": 2.
|
| 1042 |
-
"grad_norm": 0.
|
| 1043 |
-
"learning_rate": 2.
|
| 1044 |
-
"loss": 0.
|
| 1045 |
"step": 148
|
| 1046 |
},
|
| 1047 |
{
|
| 1048 |
-
"epoch": 2.
|
| 1049 |
-
"grad_norm": 0.
|
| 1050 |
-
"learning_rate": 2.
|
| 1051 |
-
"loss": 0.
|
| 1052 |
"step": 149
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
-
"epoch": 2.
|
| 1056 |
-
"grad_norm": 0.
|
| 1057 |
-
"learning_rate": 2.
|
| 1058 |
-
"loss": 0.
|
| 1059 |
"step": 150
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
-
"epoch": 2.
|
| 1063 |
-
"grad_norm": 0.
|
| 1064 |
-
"learning_rate": 2.
|
| 1065 |
-
"loss": 0.
|
| 1066 |
"step": 151
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
-
"epoch": 2.
|
| 1070 |
-
"grad_norm": 0.
|
| 1071 |
-
"learning_rate": 2.
|
| 1072 |
-
"loss": 0.
|
| 1073 |
"step": 152
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
-
"epoch": 2.
|
| 1077 |
-
"grad_norm": 0.
|
| 1078 |
-
"learning_rate": 2.
|
| 1079 |
-
"loss": 0.
|
| 1080 |
"step": 153
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
-
"epoch": 2.
|
| 1084 |
-
"grad_norm": 0.
|
| 1085 |
-
"learning_rate": 2.
|
| 1086 |
-
"loss": 0.
|
| 1087 |
"step": 154
|
| 1088 |
},
|
| 1089 |
{
|
| 1090 |
-
"epoch": 2.
|
| 1091 |
-
"grad_norm": 0.
|
| 1092 |
-
"learning_rate": 2.
|
| 1093 |
-
"loss": 0.
|
| 1094 |
"step": 155
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
-
"epoch": 2.
|
| 1098 |
-
"grad_norm": 0.
|
| 1099 |
-
"learning_rate": 2.
|
| 1100 |
-
"loss": 0.
|
| 1101 |
"step": 156
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
-
"epoch": 2.
|
| 1105 |
-
"grad_norm": 0.
|
| 1106 |
-
"learning_rate": 2.
|
| 1107 |
-
"loss": 0.
|
| 1108 |
"step": 157
|
| 1109 |
},
|
| 1110 |
{
|
| 1111 |
-
"epoch": 2.
|
| 1112 |
-
"grad_norm": 0.
|
| 1113 |
-
"learning_rate": 2.
|
| 1114 |
-
"loss": 0.
|
| 1115 |
"step": 158
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
-
"epoch": 2.
|
| 1119 |
-
"grad_norm": 0.
|
| 1120 |
-
"learning_rate":
|
| 1121 |
-
"loss": 0.
|
| 1122 |
"step": 159
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
-
"epoch": 2.
|
| 1126 |
-
"grad_norm": 0.
|
| 1127 |
-
"learning_rate":
|
| 1128 |
-
"loss": 0.
|
| 1129 |
"step": 160
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
-
"epoch": 2.
|
| 1133 |
-
"grad_norm": 0.
|
| 1134 |
-
"learning_rate": 1.
|
| 1135 |
-
"loss": 0.
|
| 1136 |
"step": 161
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
-
"epoch": 2.
|
| 1140 |
-
"grad_norm": 0.
|
| 1141 |
-
"learning_rate": 1.
|
| 1142 |
-
"loss": 0.
|
| 1143 |
"step": 162
|
| 1144 |
},
|
| 1145 |
{
|
| 1146 |
-
"epoch": 2.
|
| 1147 |
-
"grad_norm": 0.
|
| 1148 |
-
"learning_rate": 1.
|
| 1149 |
-
"loss": 0.
|
| 1150 |
"step": 163
|
| 1151 |
},
|
| 1152 |
{
|
| 1153 |
-
"epoch": 2.
|
| 1154 |
-
"grad_norm": 0.
|
| 1155 |
-
"learning_rate": 1.
|
| 1156 |
-
"loss": 0.
|
| 1157 |
"step": 164
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
-
"epoch": 2.
|
| 1161 |
-
"grad_norm": 0.
|
| 1162 |
-
"learning_rate": 1.
|
| 1163 |
-
"loss": 0.
|
| 1164 |
"step": 165
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
-
"epoch": 2.
|
| 1168 |
-
"grad_norm": 0.
|
| 1169 |
-
"learning_rate": 1.
|
| 1170 |
-
"loss": 0.
|
| 1171 |
"step": 166
|
| 1172 |
},
|
| 1173 |
{
|
| 1174 |
-
"epoch": 2.
|
| 1175 |
-
"grad_norm": 0.
|
| 1176 |
-
"learning_rate": 1.
|
| 1177 |
-
"loss": 0.
|
| 1178 |
"step": 167
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
-
"epoch": 2.
|
| 1182 |
-
"grad_norm": 0.
|
| 1183 |
-
"learning_rate": 1.
|
| 1184 |
-
"loss": 0.
|
| 1185 |
"step": 168
|
| 1186 |
},
|
| 1187 |
{
|
| 1188 |
-
"epoch": 2.
|
| 1189 |
-
"grad_norm": 0.
|
| 1190 |
-
"learning_rate": 1.
|
| 1191 |
-
"loss": 0.
|
| 1192 |
"step": 169
|
| 1193 |
},
|
| 1194 |
{
|
| 1195 |
-
"epoch": 2.
|
| 1196 |
-
"grad_norm": 0.
|
| 1197 |
-
"learning_rate": 1.
|
| 1198 |
-
"loss": 0.
|
| 1199 |
"step": 170
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
-
"epoch": 2.
|
| 1203 |
-
"grad_norm": 0.
|
| 1204 |
-
"learning_rate": 1.
|
| 1205 |
-
"loss": 0.
|
| 1206 |
"step": 171
|
| 1207 |
},
|
| 1208 |
{
|
| 1209 |
-
"epoch": 2.
|
| 1210 |
-
"grad_norm": 0.
|
| 1211 |
-
"learning_rate": 1.
|
| 1212 |
-
"loss": 0.
|
| 1213 |
"step": 172
|
| 1214 |
},
|
| 1215 |
{
|
| 1216 |
-
"epoch": 2.
|
| 1217 |
-
"grad_norm": 0.
|
| 1218 |
-
"learning_rate": 1.
|
| 1219 |
-
"loss": 0.
|
| 1220 |
"step": 173
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
-
"epoch": 2.
|
| 1224 |
-
"grad_norm": 0.
|
| 1225 |
-
"learning_rate": 1.
|
| 1226 |
-
"loss": 0.
|
| 1227 |
"step": 174
|
| 1228 |
},
|
| 1229 |
{
|
| 1230 |
-
"epoch": 2.
|
| 1231 |
-
"grad_norm": 0.
|
| 1232 |
-
"learning_rate": 1.
|
| 1233 |
-
"loss": 0.
|
| 1234 |
"step": 175
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
-
"epoch": 2.
|
| 1238 |
-
"grad_norm": 0.
|
| 1239 |
-
"learning_rate": 1.
|
| 1240 |
-
"loss": 0.
|
| 1241 |
"step": 176
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
-
"epoch": 2.
|
| 1245 |
-
"grad_norm": 0.
|
| 1246 |
-
"learning_rate":
|
| 1247 |
-
"loss": 0.
|
| 1248 |
"step": 177
|
| 1249 |
},
|
| 1250 |
{
|
| 1251 |
-
"epoch": 2.
|
| 1252 |
-
"grad_norm": 0.
|
| 1253 |
-
"learning_rate":
|
| 1254 |
-
"loss": 0.
|
| 1255 |
"step": 178
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
-
"epoch": 2.
|
| 1259 |
-
"grad_norm": 0.
|
| 1260 |
-
"learning_rate":
|
| 1261 |
-
"loss": 0.
|
| 1262 |
"step": 179
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
-
"epoch": 2.
|
| 1266 |
-
"grad_norm": 0.
|
| 1267 |
-
"learning_rate":
|
| 1268 |
-
"loss": 0.
|
| 1269 |
"step": 180
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
-
"epoch": 2.
|
| 1273 |
-
"grad_norm": 0.
|
| 1274 |
-
"learning_rate":
|
| 1275 |
-
"loss": 0.
|
| 1276 |
"step": 181
|
| 1277 |
},
|
| 1278 |
{
|
| 1279 |
-
"epoch": 2.
|
| 1280 |
-
"grad_norm": 0.
|
| 1281 |
-
"learning_rate":
|
| 1282 |
-
"loss": 0.
|
| 1283 |
"step": 182
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
-
"epoch": 2.
|
| 1287 |
-
"grad_norm": 0.
|
| 1288 |
-
"learning_rate":
|
| 1289 |
-
"loss": 0.
|
| 1290 |
"step": 183
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
-
"epoch": 2.
|
| 1294 |
-
"grad_norm": 0.
|
| 1295 |
-
"learning_rate":
|
| 1296 |
-
"loss": 0.
|
| 1297 |
"step": 184
|
| 1298 |
},
|
| 1299 |
{
|
| 1300 |
-
"epoch": 2.
|
| 1301 |
-
"grad_norm": 0.
|
| 1302 |
-
"learning_rate":
|
| 1303 |
-
"loss": 0.
|
| 1304 |
"step": 185
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
-
"epoch": 2.
|
| 1308 |
-
"grad_norm": 0.
|
| 1309 |
-
"learning_rate":
|
| 1310 |
-
"loss": 0.
|
| 1311 |
"step": 186
|
| 1312 |
},
|
| 1313 |
{
|
| 1314 |
-
"epoch": 2.
|
| 1315 |
-
"grad_norm": 0.
|
| 1316 |
-
"learning_rate":
|
| 1317 |
-
"loss": 0.
|
| 1318 |
"step": 187
|
| 1319 |
},
|
| 1320 |
{
|
| 1321 |
-
"epoch": 2.
|
| 1322 |
-
"grad_norm": 0.
|
| 1323 |
-
"learning_rate": 5.
|
| 1324 |
-
"loss": 0.
|
| 1325 |
"step": 188
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
-
"epoch": 2.
|
| 1329 |
-
"grad_norm": 0.
|
| 1330 |
-
"learning_rate":
|
| 1331 |
-
"loss": 0.
|
| 1332 |
"step": 189
|
| 1333 |
},
|
| 1334 |
{
|
| 1335 |
-
"epoch": 2.
|
| 1336 |
-
"grad_norm": 0.
|
| 1337 |
-
"learning_rate":
|
| 1338 |
-
"loss": 0.
|
| 1339 |
"step": 190
|
| 1340 |
},
|
| 1341 |
{
|
| 1342 |
-
"epoch": 2.
|
| 1343 |
-
"grad_norm": 0.
|
| 1344 |
-
"learning_rate": 4.
|
| 1345 |
-
"loss": 0.
|
| 1346 |
"step": 191
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
-
"epoch": 2.
|
| 1350 |
-
"grad_norm": 0.
|
| 1351 |
-
"learning_rate":
|
| 1352 |
-
"loss": 0.
|
| 1353 |
"step": 192
|
| 1354 |
},
|
| 1355 |
{
|
| 1356 |
-
"epoch": 2.
|
| 1357 |
-
"grad_norm": 0.
|
| 1358 |
-
"learning_rate":
|
| 1359 |
-
"loss": 0.
|
| 1360 |
"step": 193
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
-
"epoch": 2.
|
| 1364 |
-
"grad_norm": 0.
|
| 1365 |
-
"learning_rate": 3.
|
| 1366 |
-
"loss": 0.
|
| 1367 |
"step": 194
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
-
"epoch": 2.
|
| 1371 |
-
"grad_norm": 0.
|
| 1372 |
-
"learning_rate":
|
| 1373 |
-
"loss": 0.
|
| 1374 |
"step": 195
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
-
"epoch": 2.
|
| 1378 |
-
"grad_norm": 0.
|
| 1379 |
-
"learning_rate":
|
| 1380 |
-
"loss": 0.
|
| 1381 |
"step": 196
|
| 1382 |
},
|
| 1383 |
{
|
| 1384 |
-
"epoch": 2.
|
| 1385 |
-
"grad_norm": 0.
|
| 1386 |
-
"learning_rate":
|
| 1387 |
-
"loss": 0.
|
| 1388 |
"step": 197
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
-
"epoch": 2.
|
| 1392 |
-
"grad_norm": 0.
|
| 1393 |
-
"learning_rate": 2.
|
| 1394 |
-
"loss": 0.
|
| 1395 |
"step": 198
|
| 1396 |
},
|
| 1397 |
{
|
| 1398 |
-
"epoch": 2.
|
| 1399 |
-
"grad_norm": 0.
|
| 1400 |
-
"learning_rate":
|
| 1401 |
-
"loss": 0.
|
| 1402 |
"step": 199
|
| 1403 |
},
|
| 1404 |
{
|
| 1405 |
-
"epoch": 2.
|
| 1406 |
-
"grad_norm": 0.
|
| 1407 |
-
"learning_rate":
|
| 1408 |
-
"loss": 0.
|
| 1409 |
"step": 200
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
-
"epoch": 2.
|
| 1413 |
-
"grad_norm": 0.
|
| 1414 |
-
"learning_rate":
|
| 1415 |
-
"loss": 0.
|
| 1416 |
"step": 201
|
| 1417 |
},
|
| 1418 |
{
|
| 1419 |
-
"epoch": 2.
|
| 1420 |
-
"grad_norm": 0.
|
| 1421 |
-
"learning_rate": 1.
|
| 1422 |
-
"loss": 0.
|
| 1423 |
"step": 202
|
| 1424 |
},
|
| 1425 |
{
|
| 1426 |
-
"epoch": 2.
|
| 1427 |
-
"grad_norm": 0.
|
| 1428 |
-
"learning_rate": 1.
|
| 1429 |
-
"loss": 0.
|
| 1430 |
"step": 203
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
-
"epoch": 2.
|
| 1434 |
-
"grad_norm": 0.
|
| 1435 |
-
"learning_rate":
|
| 1436 |
-
"loss": 0.
|
| 1437 |
"step": 204
|
| 1438 |
},
|
| 1439 |
{
|
| 1440 |
-
"epoch": 2.
|
| 1441 |
-
"grad_norm": 0.
|
| 1442 |
-
"learning_rate":
|
| 1443 |
-
"loss": 0.
|
| 1444 |
"step": 205
|
| 1445 |
},
|
| 1446 |
{
|
| 1447 |
-
"epoch": 2.
|
| 1448 |
-
"grad_norm": 0.
|
| 1449 |
-
"learning_rate":
|
| 1450 |
-
"loss": 0.
|
| 1451 |
"step": 206
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
-
"epoch": 2.
|
| 1455 |
-
"grad_norm": 0.
|
| 1456 |
-
"learning_rate":
|
| 1457 |
-
"loss": 0.
|
| 1458 |
"step": 207
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
-
"epoch": 2.
|
| 1462 |
-
"grad_norm": 0.
|
| 1463 |
-
"learning_rate":
|
| 1464 |
-
"loss": 0.
|
| 1465 |
"step": 208
|
| 1466 |
},
|
| 1467 |
{
|
| 1468 |
-
"epoch": 2.
|
| 1469 |
-
"grad_norm": 0.
|
| 1470 |
-
"learning_rate":
|
| 1471 |
-
"loss": 0.
|
| 1472 |
"step": 209
|
| 1473 |
},
|
| 1474 |
{
|
| 1475 |
-
"epoch": 2.
|
| 1476 |
-
"grad_norm": 0.
|
| 1477 |
-
"learning_rate":
|
| 1478 |
-
"loss": 0.
|
| 1479 |
"step": 210
|
| 1480 |
},
|
| 1481 |
{
|
| 1482 |
-
"epoch": 2.
|
| 1483 |
-
"grad_norm": 0.
|
| 1484 |
-
"learning_rate":
|
| 1485 |
-
"loss": 0.
|
| 1486 |
"step": 211
|
| 1487 |
},
|
| 1488 |
{
|
| 1489 |
-
"epoch": 2.
|
| 1490 |
-
"grad_norm": 0.
|
| 1491 |
-
"learning_rate":
|
| 1492 |
-
"loss": 0.
|
| 1493 |
"step": 212
|
| 1494 |
},
|
| 1495 |
{
|
| 1496 |
-
"epoch": 2.
|
| 1497 |
-
"grad_norm": 0.
|
| 1498 |
-
"learning_rate":
|
| 1499 |
-
"loss": 0.
|
| 1500 |
"step": 213
|
| 1501 |
},
|
| 1502 |
{
|
| 1503 |
-
"epoch": 2.
|
| 1504 |
-
"grad_norm": 0.
|
| 1505 |
-
"learning_rate":
|
| 1506 |
-
"loss": 0.
|
| 1507 |
"step": 214
|
| 1508 |
},
|
| 1509 |
{
|
| 1510 |
-
"epoch": 2.
|
| 1511 |
-
"grad_norm": 0.
|
| 1512 |
-
"learning_rate":
|
| 1513 |
-
"loss": 0.
|
| 1514 |
"step": 215
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
-
"epoch": 2.
|
| 1518 |
-
"grad_norm": 0.
|
| 1519 |
-
"learning_rate": 5.720934490907604e-09,
|
| 1520 |
-
"loss": 0.2142,
|
| 1521 |
-
"step": 216
|
| 1522 |
-
},
|
| 1523 |
-
{
|
| 1524 |
-
"epoch": 2.9726027397260273,
|
| 1525 |
-
"grad_norm": 0.48284047194050117,
|
| 1526 |
-
"learning_rate": 2.542906999836725e-09,
|
| 1527 |
-
"loss": 0.2298,
|
| 1528 |
-
"step": 217
|
| 1529 |
-
},
|
| 1530 |
-
{
|
| 1531 |
-
"epoch": 2.9863013698630136,
|
| 1532 |
-
"grad_norm": 0.5051550042171722,
|
| 1533 |
-
"learning_rate": 6.357671699486201e-10,
|
| 1534 |
-
"loss": 0.2056,
|
| 1535 |
-
"step": 218
|
| 1536 |
-
},
|
| 1537 |
-
{
|
| 1538 |
-
"epoch": 3.0,
|
| 1539 |
-
"grad_norm": 0.500169473170659,
|
| 1540 |
"learning_rate": 0.0,
|
| 1541 |
-
"loss": 0.
|
| 1542 |
-
"step":
|
| 1543 |
},
|
| 1544 |
{
|
| 1545 |
-
"epoch":
|
| 1546 |
-
"step":
|
| 1547 |
-
"total_flos":
|
| 1548 |
-
"train_loss": 0.
|
| 1549 |
-
"train_runtime":
|
| 1550 |
-
"train_samples_per_second":
|
| 1551 |
-
"train_steps_per_second": 0.
|
| 1552 |
}
|
| 1553 |
],
|
| 1554 |
"logging_steps": 1,
|
| 1555 |
-
"max_steps":
|
| 1556 |
"num_input_tokens_seen": 0,
|
| 1557 |
"num_train_epochs": 3,
|
| 1558 |
"save_steps": 500,
|
|
@@ -1568,8 +1547,8 @@
|
|
| 1568 |
"attributes": {}
|
| 1569 |
}
|
| 1570 |
},
|
| 1571 |
-
"total_flos":
|
| 1572 |
-
"train_batch_size":
|
| 1573 |
"trial_name": null,
|
| 1574 |
"trial_params": null
|
| 1575 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.986175115207373,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 216,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.013824884792626729,
|
| 13 |
+
"grad_norm": 3.8321747022193033,
|
| 14 |
"learning_rate": 4.5454545454545457e-07,
|
| 15 |
+
"loss": 0.4904,
|
| 16 |
"step": 1
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.027649769585253458,
|
| 20 |
+
"grad_norm": 4.166445470011081,
|
| 21 |
"learning_rate": 9.090909090909091e-07,
|
| 22 |
+
"loss": 0.4586,
|
| 23 |
"step": 2
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.041474654377880185,
|
| 27 |
+
"grad_norm": 3.8454000093459273,
|
| 28 |
"learning_rate": 1.3636363636363636e-06,
|
| 29 |
+
"loss": 0.4491,
|
| 30 |
"step": 3
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.055299539170506916,
|
| 34 |
+
"grad_norm": 3.9389330204801944,
|
| 35 |
"learning_rate": 1.8181818181818183e-06,
|
| 36 |
+
"loss": 0.495,
|
| 37 |
"step": 4
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.06912442396313365,
|
| 41 |
+
"grad_norm": 3.503496575080232,
|
| 42 |
"learning_rate": 2.2727272727272728e-06,
|
| 43 |
+
"loss": 0.4518,
|
| 44 |
"step": 5
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.08294930875576037,
|
| 48 |
+
"grad_norm": 2.5778889382942296,
|
| 49 |
"learning_rate": 2.7272727272727272e-06,
|
| 50 |
+
"loss": 0.4245,
|
| 51 |
"step": 6
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.0967741935483871,
|
| 55 |
+
"grad_norm": 2.6973870749604654,
|
| 56 |
"learning_rate": 3.181818181818182e-06,
|
| 57 |
+
"loss": 0.4485,
|
| 58 |
"step": 7
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.11059907834101383,
|
| 62 |
+
"grad_norm": 2.087801488195196,
|
| 63 |
"learning_rate": 3.6363636363636366e-06,
|
| 64 |
+
"loss": 0.4126,
|
| 65 |
"step": 8
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.12442396313364056,
|
| 69 |
+
"grad_norm": 1.937230458954793,
|
| 70 |
"learning_rate": 4.0909090909090915e-06,
|
| 71 |
+
"loss": 0.3782,
|
| 72 |
"step": 9
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.1382488479262673,
|
| 76 |
+
"grad_norm": 1.4844954637777827,
|
| 77 |
"learning_rate": 4.5454545454545455e-06,
|
| 78 |
+
"loss": 0.4114,
|
| 79 |
"step": 10
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.15207373271889402,
|
| 83 |
+
"grad_norm": 1.6047087443414494,
|
| 84 |
"learning_rate": 5e-06,
|
| 85 |
+
"loss": 0.4018,
|
| 86 |
"step": 11
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.16589861751152074,
|
| 90 |
+
"grad_norm": 1.2519442814402522,
|
| 91 |
"learning_rate": 5.4545454545454545e-06,
|
| 92 |
+
"loss": 0.3739,
|
| 93 |
"step": 12
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.17972350230414746,
|
| 97 |
+
"grad_norm": 1.11650299203384,
|
| 98 |
"learning_rate": 5.90909090909091e-06,
|
| 99 |
+
"loss": 0.3516,
|
| 100 |
"step": 13
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.1935483870967742,
|
| 104 |
+
"grad_norm": 1.3867124167107647,
|
| 105 |
"learning_rate": 6.363636363636364e-06,
|
| 106 |
+
"loss": 0.3873,
|
| 107 |
"step": 14
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.2073732718894009,
|
| 111 |
+
"grad_norm": 1.6171625319611096,
|
| 112 |
"learning_rate": 6.818181818181818e-06,
|
| 113 |
+
"loss": 0.3635,
|
| 114 |
"step": 15
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.22119815668202766,
|
| 118 |
+
"grad_norm": 1.4344025164842837,
|
| 119 |
"learning_rate": 7.272727272727273e-06,
|
| 120 |
+
"loss": 0.3538,
|
| 121 |
"step": 16
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.2350230414746544,
|
| 125 |
+
"grad_norm": 1.2769447355987351,
|
| 126 |
"learning_rate": 7.727272727272727e-06,
|
| 127 |
+
"loss": 0.3322,
|
| 128 |
"step": 17
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.2488479262672811,
|
| 132 |
+
"grad_norm": 1.1544799589527472,
|
| 133 |
"learning_rate": 8.181818181818183e-06,
|
| 134 |
+
"loss": 0.3175,
|
| 135 |
"step": 18
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.2626728110599078,
|
| 139 |
+
"grad_norm": 0.9340143537679371,
|
| 140 |
"learning_rate": 8.636363636363637e-06,
|
| 141 |
+
"loss": 0.2943,
|
| 142 |
"step": 19
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 0.2764976958525346,
|
| 146 |
+
"grad_norm": 0.8350563937099799,
|
| 147 |
"learning_rate": 9.090909090909091e-06,
|
| 148 |
+
"loss": 0.3269,
|
| 149 |
"step": 20
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.2903225806451613,
|
| 153 |
+
"grad_norm": 0.9427953311385877,
|
| 154 |
"learning_rate": 9.545454545454547e-06,
|
| 155 |
+
"loss": 0.3665,
|
| 156 |
"step": 21
|
| 157 |
},
|
| 158 |
{
|
| 159 |
+
"epoch": 0.30414746543778803,
|
| 160 |
+
"grad_norm": 0.8571337937972116,
|
| 161 |
"learning_rate": 1e-05,
|
| 162 |
+
"loss": 0.3081,
|
| 163 |
"step": 22
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 0.31797235023041476,
|
| 167 |
+
"grad_norm": 0.9574970030840929,
|
| 168 |
+
"learning_rate": 9.999344418328161e-06,
|
| 169 |
+
"loss": 0.3216,
|
| 170 |
"step": 23
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.3317972350230415,
|
| 174 |
+
"grad_norm": 0.8240000968027297,
|
| 175 |
+
"learning_rate": 9.997377845227577e-06,
|
| 176 |
+
"loss": 0.322,
|
| 177 |
"step": 24
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 0.3456221198156682,
|
| 181 |
+
"grad_norm": 0.7442732006332352,
|
| 182 |
+
"learning_rate": 9.994100796397954e-06,
|
| 183 |
+
"loss": 0.3406,
|
| 184 |
"step": 25
|
| 185 |
},
|
| 186 |
{
|
| 187 |
+
"epoch": 0.35944700460829493,
|
| 188 |
+
"grad_norm": 0.862489472956441,
|
| 189 |
+
"learning_rate": 9.98951413118856e-06,
|
| 190 |
+
"loss": 0.342,
|
| 191 |
"step": 26
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 0.37327188940092165,
|
| 195 |
+
"grad_norm": 0.7897584822354033,
|
| 196 |
+
"learning_rate": 9.983619052372847e-06,
|
| 197 |
+
"loss": 0.3228,
|
| 198 |
"step": 27
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 0.3870967741935484,
|
| 202 |
+
"grad_norm": 0.8247735964866483,
|
| 203 |
+
"learning_rate": 9.97641710583307e-06,
|
| 204 |
+
"loss": 0.34,
|
| 205 |
"step": 28
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 0.4009216589861751,
|
| 209 |
+
"grad_norm": 0.6877859922485577,
|
| 210 |
+
"learning_rate": 9.96791018015489e-06,
|
| 211 |
+
"loss": 0.3624,
|
| 212 |
"step": 29
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 0.4147465437788018,
|
| 216 |
+
"grad_norm": 0.6854976819111457,
|
| 217 |
+
"learning_rate": 9.958100506132127e-06,
|
| 218 |
+
"loss": 0.3362,
|
| 219 |
"step": 30
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.42857142857142855,
|
| 223 |
+
"grad_norm": 0.6780848305353676,
|
| 224 |
+
"learning_rate": 9.946990656181782e-06,
|
| 225 |
+
"loss": 0.3306,
|
| 226 |
"step": 31
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.4423963133640553,
|
| 230 |
+
"grad_norm": 0.8635579814520753,
|
| 231 |
+
"learning_rate": 9.934583543669454e-06,
|
| 232 |
+
"loss": 0.333,
|
| 233 |
"step": 32
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 0.45622119815668205,
|
| 237 |
+
"grad_norm": 0.7095969477530761,
|
| 238 |
+
"learning_rate": 9.920882422145372e-06,
|
| 239 |
+
"loss": 0.3297,
|
| 240 |
"step": 33
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 0.4700460829493088,
|
| 244 |
+
"grad_norm": 0.6554896662064854,
|
| 245 |
+
"learning_rate": 9.905890884491196e-06,
|
| 246 |
+
"loss": 0.3474,
|
| 247 |
"step": 34
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 0.4838709677419355,
|
| 251 |
+
"grad_norm": 0.7233953981673558,
|
| 252 |
+
"learning_rate": 9.889612861977855e-06,
|
| 253 |
+
"loss": 0.3366,
|
| 254 |
"step": 35
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 0.4976958525345622,
|
| 258 |
+
"grad_norm": 0.7808564766414073,
|
| 259 |
+
"learning_rate": 9.872052623234632e-06,
|
| 260 |
+
"loss": 0.3247,
|
| 261 |
"step": 36
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 0.511520737327189,
|
| 265 |
+
"grad_norm": 0.7662284906743654,
|
| 266 |
+
"learning_rate": 9.853214773129796e-06,
|
| 267 |
+
"loss": 0.3035,
|
| 268 |
"step": 37
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"epoch": 0.5253456221198156,
|
| 272 |
+
"grad_norm": 0.7589864104505978,
|
| 273 |
+
"learning_rate": 9.833104251563058e-06,
|
| 274 |
+
"loss": 0.3222,
|
| 275 |
"step": 38
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 0.5391705069124424,
|
| 279 |
+
"grad_norm": 0.7364643638285358,
|
| 280 |
+
"learning_rate": 9.811726332170153e-06,
|
| 281 |
+
"loss": 0.3557,
|
| 282 |
"step": 39
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 0.5529953917050692,
|
| 286 |
+
"grad_norm": 0.7228711702399517,
|
| 287 |
+
"learning_rate": 9.789086620939936e-06,
|
| 288 |
+
"loss": 0.3392,
|
| 289 |
"step": 40
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 0.5668202764976958,
|
| 293 |
+
"grad_norm": 0.6418796946575632,
|
| 294 |
+
"learning_rate": 9.765191054744305e-06,
|
| 295 |
"loss": 0.3305,
|
| 296 |
"step": 41
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"epoch": 0.5806451612903226,
|
| 300 |
+
"grad_norm": 0.7360519892248771,
|
| 301 |
+
"learning_rate": 9.740045899781353e-06,
|
| 302 |
+
"loss": 0.3387,
|
| 303 |
"step": 42
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 0.5944700460829493,
|
| 307 |
+
"grad_norm": 0.7117878571151136,
|
| 308 |
+
"learning_rate": 9.713657749932172e-06,
|
| 309 |
+
"loss": 0.378,
|
| 310 |
"step": 43
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"epoch": 0.6082949308755761,
|
| 314 |
+
"grad_norm": 0.6562391589488417,
|
| 315 |
+
"learning_rate": 9.68603352503172e-06,
|
| 316 |
+
"loss": 0.3808,
|
| 317 |
"step": 44
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 0.6221198156682027,
|
| 321 |
+
"grad_norm": 0.7011857576729418,
|
| 322 |
+
"learning_rate": 9.657180469054213e-06,
|
| 323 |
+
"loss": 0.3561,
|
| 324 |
"step": 45
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 0.6359447004608295,
|
| 328 |
+
"grad_norm": 0.6588644804163393,
|
| 329 |
+
"learning_rate": 9.627106148213521e-06,
|
| 330 |
+
"loss": 0.3224,
|
| 331 |
"step": 46
|
| 332 |
},
|
| 333 |
{
|
| 334 |
+
"epoch": 0.6497695852534562,
|
| 335 |
+
"grad_norm": 0.6413771673007567,
|
| 336 |
+
"learning_rate": 9.595818448979061e-06,
|
| 337 |
+
"loss": 0.3502,
|
| 338 |
"step": 47
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 0.663594470046083,
|
| 342 |
+
"grad_norm": 0.6132130927693553,
|
| 343 |
+
"learning_rate": 9.563325576007702e-06,
|
| 344 |
+
"loss": 0.3172,
|
| 345 |
"step": 48
|
| 346 |
},
|
| 347 |
{
|
| 348 |
+
"epoch": 0.6774193548387096,
|
| 349 |
+
"grad_norm": 0.7562554878178558,
|
| 350 |
+
"learning_rate": 9.529636049992235e-06,
|
| 351 |
+
"loss": 0.3078,
|
| 352 |
"step": 49
|
| 353 |
},
|
| 354 |
{
|
| 355 |
+
"epoch": 0.6912442396313364,
|
| 356 |
+
"grad_norm": 0.7025632258958434,
|
| 357 |
+
"learning_rate": 9.494758705426978e-06,
|
| 358 |
+
"loss": 0.3378,
|
| 359 |
"step": 50
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 0.7050691244239631,
|
| 363 |
+
"grad_norm": 0.6580901818465655,
|
| 364 |
+
"learning_rate": 9.458702688291072e-06,
|
| 365 |
+
"loss": 0.3749,
|
| 366 |
"step": 51
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 0.7188940092165899,
|
| 370 |
+
"grad_norm": 0.837525662904706,
|
| 371 |
+
"learning_rate": 9.421477453650118e-06,
|
| 372 |
+
"loss": 0.3646,
|
| 373 |
"step": 52
|
| 374 |
},
|
| 375 |
{
|
| 376 |
+
"epoch": 0.7327188940092166,
|
| 377 |
+
"grad_norm": 0.6988946242836359,
|
| 378 |
+
"learning_rate": 9.38309276317674e-06,
|
| 379 |
+
"loss": 0.2952,
|
| 380 |
"step": 53
|
| 381 |
},
|
| 382 |
{
|
| 383 |
+
"epoch": 0.7465437788018433,
|
| 384 |
+
"grad_norm": 0.7088548576538783,
|
| 385 |
+
"learning_rate": 9.343558682590757e-06,
|
| 386 |
+
"loss": 0.3402,
|
| 387 |
"step": 54
|
| 388 |
},
|
| 389 |
{
|
| 390 |
+
"epoch": 0.7603686635944701,
|
| 391 |
+
"grad_norm": 0.6031364330217948,
|
| 392 |
+
"learning_rate": 9.302885579019626e-06,
|
| 393 |
+
"loss": 0.3157,
|
| 394 |
"step": 55
|
| 395 |
},
|
| 396 |
{
|
| 397 |
+
"epoch": 0.7741935483870968,
|
| 398 |
+
"grad_norm": 0.6869976637076074,
|
| 399 |
+
"learning_rate": 9.261084118279846e-06,
|
| 400 |
+
"loss": 0.3724,
|
| 401 |
"step": 56
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 0.7880184331797235,
|
| 405 |
+
"grad_norm": 0.6362516594505776,
|
| 406 |
+
"learning_rate": 9.218165262080024e-06,
|
| 407 |
+
"loss": 0.2915,
|
| 408 |
"step": 57
|
| 409 |
},
|
| 410 |
{
|
| 411 |
+
"epoch": 0.8018433179723502,
|
| 412 |
+
"grad_norm": 0.6133826251571223,
|
| 413 |
+
"learning_rate": 9.174140265146356e-06,
|
| 414 |
+
"loss": 0.3374,
|
| 415 |
"step": 58
|
| 416 |
},
|
| 417 |
{
|
| 418 |
+
"epoch": 0.815668202764977,
|
| 419 |
+
"grad_norm": 0.5552231441277784,
|
| 420 |
+
"learning_rate": 9.129020672271283e-06,
|
| 421 |
+
"loss": 0.3119,
|
| 422 |
"step": 59
|
| 423 |
},
|
| 424 |
{
|
| 425 |
+
"epoch": 0.8294930875576036,
|
| 426 |
+
"grad_norm": 0.5344163684333119,
|
| 427 |
+
"learning_rate": 9.082818315286054e-06,
|
| 428 |
+
"loss": 0.3139,
|
| 429 |
"step": 60
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 0.8433179723502304,
|
| 433 |
+
"grad_norm": 0.5908367481260651,
|
| 434 |
+
"learning_rate": 9.035545309958048e-06,
|
| 435 |
+
"loss": 0.3391,
|
| 436 |
"step": 61
|
| 437 |
},
|
| 438 |
{
|
| 439 |
+
"epoch": 0.8571428571428571,
|
| 440 |
+
"grad_norm": 0.6049797134002246,
|
| 441 |
+
"learning_rate": 8.987214052813605e-06,
|
| 442 |
+
"loss": 0.3385,
|
| 443 |
"step": 62
|
| 444 |
},
|
| 445 |
{
|
| 446 |
+
"epoch": 0.8709677419354839,
|
| 447 |
+
"grad_norm": 0.632151936446925,
|
| 448 |
+
"learning_rate": 8.937837217887273e-06,
|
| 449 |
+
"loss": 0.3216,
|
| 450 |
"step": 63
|
| 451 |
},
|
| 452 |
{
|
| 453 |
+
"epoch": 0.8847926267281107,
|
| 454 |
+
"grad_norm": 0.573287253380456,
|
| 455 |
+
"learning_rate": 8.887427753398249e-06,
|
| 456 |
+
"loss": 0.2931,
|
| 457 |
"step": 64
|
| 458 |
},
|
| 459 |
{
|
| 460 |
+
"epoch": 0.8986175115207373,
|
| 461 |
+
"grad_norm": 0.5852983805141764,
|
| 462 |
+
"learning_rate": 8.83599887835493e-06,
|
| 463 |
+
"loss": 0.3172,
|
| 464 |
"step": 65
|
| 465 |
},
|
| 466 |
{
|
| 467 |
+
"epoch": 0.9124423963133641,
|
| 468 |
+
"grad_norm": 0.6363174322514683,
|
| 469 |
+
"learning_rate": 8.783564079088478e-06,
|
| 470 |
+
"loss": 0.3882,
|
| 471 |
"step": 66
|
| 472 |
},
|
| 473 |
{
|
| 474 |
+
"epoch": 0.9262672811059908,
|
| 475 |
+
"grad_norm": 0.6351465289963182,
|
| 476 |
+
"learning_rate": 8.730137105716231e-06,
|
| 477 |
+
"loss": 0.2963,
|
| 478 |
"step": 67
|
| 479 |
},
|
| 480 |
{
|
| 481 |
+
"epoch": 0.9400921658986175,
|
| 482 |
+
"grad_norm": 0.5424528987366075,
|
| 483 |
+
"learning_rate": 8.675731968536004e-06,
|
| 484 |
+
"loss": 0.2824,
|
| 485 |
"step": 68
|
| 486 |
},
|
| 487 |
{
|
| 488 |
+
"epoch": 0.9539170506912442,
|
| 489 |
+
"grad_norm": 0.5850480108903799,
|
| 490 |
+
"learning_rate": 8.620362934352109e-06,
|
| 491 |
+
"loss": 0.3439,
|
| 492 |
"step": 69
|
| 493 |
},
|
| 494 |
{
|
| 495 |
+
"epoch": 0.967741935483871,
|
| 496 |
+
"grad_norm": 0.5924509545505863,
|
| 497 |
+
"learning_rate": 8.564044522734147e-06,
|
| 498 |
+
"loss": 0.345,
|
| 499 |
"step": 70
|
| 500 |
},
|
| 501 |
{
|
| 502 |
+
"epoch": 0.9815668202764977,
|
| 503 |
+
"grad_norm": 0.6356218187245817,
|
| 504 |
+
"learning_rate": 8.506791502209497e-06,
|
| 505 |
+
"loss": 0.3383,
|
| 506 |
"step": 71
|
| 507 |
},
|
| 508 |
{
|
| 509 |
+
"epoch": 0.9953917050691244,
|
| 510 |
+
"grad_norm": 0.5255727075687244,
|
| 511 |
+
"learning_rate": 8.448618886390523e-06,
|
| 512 |
+
"loss": 0.31,
|
| 513 |
"step": 72
|
| 514 |
},
|
| 515 |
{
|
| 516 |
+
"epoch": 1.0092165898617511,
|
| 517 |
+
"grad_norm": 1.075353564084177,
|
| 518 |
+
"learning_rate": 8.389541930037516e-06,
|
| 519 |
+
"loss": 0.4918,
|
| 520 |
"step": 73
|
| 521 |
},
|
| 522 |
{
|
| 523 |
+
"epoch": 1.023041474654378,
|
| 524 |
+
"grad_norm": 0.5411851622437062,
|
| 525 |
+
"learning_rate": 8.329576125058406e-06,
|
| 526 |
+
"loss": 0.284,
|
| 527 |
"step": 74
|
| 528 |
},
|
| 529 |
{
|
| 530 |
+
"epoch": 1.0368663594470047,
|
| 531 |
+
"grad_norm": 0.5438469424766922,
|
| 532 |
+
"learning_rate": 8.268737196446264e-06,
|
| 533 |
+
"loss": 0.2914,
|
| 534 |
"step": 75
|
| 535 |
},
|
| 536 |
{
|
| 537 |
+
"epoch": 1.0506912442396312,
|
| 538 |
+
"grad_norm": 0.5420670331882803,
|
| 539 |
+
"learning_rate": 8.207041098155701e-06,
|
| 540 |
+
"loss": 0.2723,
|
| 541 |
"step": 76
|
| 542 |
},
|
| 543 |
{
|
| 544 |
+
"epoch": 1.064516129032258,
|
| 545 |
+
"grad_norm": 0.6001856622683356,
|
| 546 |
+
"learning_rate": 8.144504008919224e-06,
|
| 547 |
+
"loss": 0.3119,
|
| 548 |
"step": 77
|
| 549 |
},
|
| 550 |
{
|
| 551 |
+
"epoch": 1.0783410138248848,
|
| 552 |
+
"grad_norm": 0.5657652505048809,
|
| 553 |
+
"learning_rate": 8.081142328004638e-06,
|
| 554 |
+
"loss": 0.2598,
|
| 555 |
"step": 78
|
| 556 |
},
|
| 557 |
{
|
| 558 |
+
"epoch": 1.0921658986175116,
|
| 559 |
+
"grad_norm": 0.5575714850260662,
|
| 560 |
+
"learning_rate": 8.016972670914624e-06,
|
| 561 |
+
"loss": 0.322,
|
| 562 |
"step": 79
|
| 563 |
},
|
| 564 |
{
|
| 565 |
+
"epoch": 1.1059907834101383,
|
| 566 |
+
"grad_norm": 0.6083958309172656,
|
| 567 |
+
"learning_rate": 7.952011865029614e-06,
|
| 568 |
+
"loss": 0.2509,
|
| 569 |
"step": 80
|
| 570 |
},
|
| 571 |
{
|
| 572 |
+
"epoch": 1.119815668202765,
|
| 573 |
+
"grad_norm": 0.5266548143398665,
|
| 574 |
+
"learning_rate": 7.886276945195098e-06,
|
| 575 |
+
"loss": 0.2206,
|
| 576 |
"step": 81
|
| 577 |
},
|
| 578 |
{
|
| 579 |
+
"epoch": 1.1336405529953917,
|
| 580 |
+
"grad_norm": 0.5687307229026926,
|
| 581 |
+
"learning_rate": 7.819785149254534e-06,
|
| 582 |
+
"loss": 0.2954,
|
| 583 |
"step": 82
|
| 584 |
},
|
| 585 |
{
|
| 586 |
+
"epoch": 1.1474654377880185,
|
| 587 |
+
"grad_norm": 0.5640940659138813,
|
| 588 |
+
"learning_rate": 7.752553913529019e-06,
|
| 589 |
+
"loss": 0.2628,
|
| 590 |
"step": 83
|
| 591 |
},
|
| 592 |
{
|
| 593 |
+
"epoch": 1.1612903225806452,
|
| 594 |
+
"grad_norm": 0.6680289489599975,
|
| 595 |
+
"learning_rate": 7.68460086824492e-06,
|
| 596 |
+
"loss": 0.3206,
|
| 597 |
"step": 84
|
| 598 |
},
|
| 599 |
{
|
| 600 |
+
"epoch": 1.1751152073732718,
|
| 601 |
+
"grad_norm": 0.5283769448858802,
|
| 602 |
+
"learning_rate": 7.61594383291065e-06,
|
| 603 |
+
"loss": 0.2363,
|
| 604 |
"step": 85
|
| 605 |
},
|
| 606 |
{
|
| 607 |
+
"epoch": 1.1889400921658986,
|
| 608 |
+
"grad_norm": 0.6482385383082108,
|
| 609 |
+
"learning_rate": 7.546600811643816e-06,
|
| 610 |
+
"loss": 0.3353,
|
| 611 |
"step": 86
|
| 612 |
},
|
| 613 |
{
|
| 614 |
+
"epoch": 1.2027649769585254,
|
| 615 |
+
"grad_norm": 0.5264348159834544,
|
| 616 |
+
"learning_rate": 7.476589988449939e-06,
|
| 617 |
+
"loss": 0.2618,
|
| 618 |
"step": 87
|
| 619 |
},
|
| 620 |
{
|
| 621 |
+
"epoch": 1.2165898617511521,
|
| 622 |
+
"grad_norm": 0.5512488141503382,
|
| 623 |
+
"learning_rate": 7.405929722454026e-06,
|
| 624 |
+
"loss": 0.2639,
|
| 625 |
"step": 88
|
| 626 |
},
|
| 627 |
{
|
| 628 |
+
"epoch": 1.230414746543779,
|
| 629 |
+
"grad_norm": 0.5093764966098362,
|
| 630 |
+
"learning_rate": 7.334638543086203e-06,
|
| 631 |
+
"loss": 0.2477,
|
| 632 |
"step": 89
|
| 633 |
},
|
| 634 |
{
|
| 635 |
+
"epoch": 1.2442396313364055,
|
| 636 |
+
"grad_norm": 0.5160996147207031,
|
| 637 |
+
"learning_rate": 7.262735145222696e-06,
|
| 638 |
+
"loss": 0.2437,
|
| 639 |
"step": 90
|
| 640 |
},
|
| 641 |
{
|
| 642 |
+
"epoch": 1.2580645161290323,
|
| 643 |
+
"grad_norm": 0.5507115707918743,
|
| 644 |
+
"learning_rate": 7.190238384283413e-06,
|
| 645 |
+
"loss": 0.2625,
|
| 646 |
"step": 91
|
| 647 |
},
|
| 648 |
{
|
| 649 |
+
"epoch": 1.271889400921659,
|
| 650 |
+
"grad_norm": 0.5596845582953057,
|
| 651 |
+
"learning_rate": 7.117167271287453e-06,
|
| 652 |
+
"loss": 0.2838,
|
| 653 |
"step": 92
|
| 654 |
},
|
| 655 |
{
|
| 656 |
+
"epoch": 1.2857142857142856,
|
| 657 |
+
"grad_norm": 0.5261309041679847,
|
| 658 |
+
"learning_rate": 7.043540967867782e-06,
|
| 659 |
+
"loss": 0.2596,
|
| 660 |
"step": 93
|
| 661 |
},
|
| 662 |
{
|
| 663 |
+
"epoch": 1.2995391705069124,
|
| 664 |
+
"grad_norm": 0.606787789098388,
|
| 665 |
+
"learning_rate": 6.969378781246436e-06,
|
| 666 |
+
"loss": 0.295,
|
| 667 |
"step": 94
|
| 668 |
},
|
| 669 |
{
|
| 670 |
+
"epoch": 1.3133640552995391,
|
| 671 |
+
"grad_norm": 0.5525493866624152,
|
| 672 |
+
"learning_rate": 6.894700159171535e-06,
|
| 673 |
+
"loss": 0.2478,
|
| 674 |
"step": 95
|
| 675 |
},
|
| 676 |
{
|
| 677 |
+
"epoch": 1.327188940092166,
|
| 678 |
+
"grad_norm": 0.4948582007424209,
|
| 679 |
+
"learning_rate": 6.819524684817439e-06,
|
| 680 |
+
"loss": 0.2375,
|
| 681 |
"step": 96
|
| 682 |
},
|
| 683 |
{
|
| 684 |
+
"epoch": 1.3410138248847927,
|
| 685 |
+
"grad_norm": 0.5910117531529238,
|
| 686 |
+
"learning_rate": 6.743872071649411e-06,
|
| 687 |
+
"loss": 0.293,
|
| 688 |
"step": 97
|
| 689 |
},
|
| 690 |
{
|
| 691 |
+
"epoch": 1.3548387096774195,
|
| 692 |
+
"grad_norm": 0.5642544177081138,
|
| 693 |
+
"learning_rate": 6.667762158254104e-06,
|
| 694 |
+
"loss": 0.2759,
|
| 695 |
"step": 98
|
| 696 |
},
|
| 697 |
{
|
| 698 |
+
"epoch": 1.368663594470046,
|
| 699 |
+
"grad_norm": 0.5640787140131098,
|
| 700 |
+
"learning_rate": 6.591214903137221e-06,
|
| 701 |
+
"loss": 0.2919,
|
| 702 |
"step": 99
|
| 703 |
},
|
| 704 |
{
|
| 705 |
+
"epoch": 1.3824884792626728,
|
| 706 |
+
"grad_norm": 0.627883249036726,
|
| 707 |
+
"learning_rate": 6.514250379489754e-06,
|
| 708 |
+
"loss": 0.3236,
|
| 709 |
"step": 100
|
| 710 |
},
|
| 711 |
{
|
| 712 |
+
"epoch": 1.3963133640552996,
|
| 713 |
+
"grad_norm": 0.5777799162176069,
|
| 714 |
+
"learning_rate": 6.436888769924142e-06,
|
| 715 |
+
"loss": 0.2599,
|
| 716 |
"step": 101
|
| 717 |
},
|
| 718 |
{
|
| 719 |
+
"epoch": 1.4101382488479262,
|
| 720 |
+
"grad_norm": 0.5575789159801893,
|
| 721 |
+
"learning_rate": 6.3591503611817155e-06,
|
| 722 |
+
"loss": 0.2881,
|
| 723 |
"step": 102
|
| 724 |
},
|
| 725 |
{
|
| 726 |
+
"epoch": 1.423963133640553,
|
| 727 |
+
"grad_norm": 0.5454339637824865,
|
| 728 |
+
"learning_rate": 6.281055538812861e-06,
|
| 729 |
+
"loss": 0.2528,
|
| 730 |
"step": 103
|
| 731 |
},
|
| 732 |
{
|
| 733 |
+
"epoch": 1.4377880184331797,
|
| 734 |
+
"grad_norm": 0.5785302727121883,
|
| 735 |
+
"learning_rate": 6.202624781831269e-06,
|
| 736 |
+
"loss": 0.2586,
|
| 737 |
"step": 104
|
| 738 |
},
|
| 739 |
{
|
| 740 |
+
"epoch": 1.4516129032258065,
|
| 741 |
+
"grad_norm": 0.5795551004814405,
|
| 742 |
+
"learning_rate": 6.123878657343648e-06,
|
| 743 |
+
"loss": 0.2933,
|
| 744 |
"step": 105
|
| 745 |
},
|
| 746 |
{
|
| 747 |
+
"epoch": 1.4654377880184333,
|
| 748 |
+
"grad_norm": 0.5562868873733925,
|
| 749 |
+
"learning_rate": 6.044837815156377e-06,
|
| 750 |
+
"loss": 0.2682,
|
| 751 |
"step": 106
|
| 752 |
},
|
| 753 |
{
|
| 754 |
+
"epoch": 1.4792626728110598,
|
| 755 |
+
"grad_norm": 0.5788310033089493,
|
| 756 |
+
"learning_rate": 5.965522982360441e-06,
|
| 757 |
+
"loss": 0.2574,
|
| 758 |
"step": 107
|
| 759 |
},
|
| 760 |
{
|
| 761 |
+
"epoch": 1.4930875576036866,
|
| 762 |
+
"grad_norm": 0.5239597133970942,
|
| 763 |
+
"learning_rate": 5.885954957896115e-06,
|
| 764 |
+
"loss": 0.256,
|
| 765 |
"step": 108
|
| 766 |
},
|
| 767 |
{
|
| 768 |
+
"epoch": 1.5069124423963134,
|
| 769 |
+
"grad_norm": 0.5873986776610259,
|
| 770 |
+
"learning_rate": 5.806154607098799e-06,
|
| 771 |
+
"loss": 0.3367,
|
| 772 |
"step": 109
|
| 773 |
},
|
| 774 |
{
|
| 775 |
+
"epoch": 1.52073732718894,
|
| 776 |
+
"grad_norm": 0.573952411610799,
|
| 777 |
+
"learning_rate": 5.726142856227453e-06,
|
| 778 |
+
"loss": 0.2801,
|
| 779 |
"step": 110
|
| 780 |
},
|
| 781 |
{
|
| 782 |
+
"epoch": 1.5345622119815667,
|
| 783 |
+
"grad_norm": 0.5754764716263057,
|
| 784 |
+
"learning_rate": 5.645940686977033e-06,
|
| 785 |
+
"loss": 0.2512,
|
| 786 |
"step": 111
|
| 787 |
},
|
| 788 |
{
|
| 789 |
+
"epoch": 1.5483870967741935,
|
| 790 |
+
"grad_norm": 0.5612364021450903,
|
| 791 |
+
"learning_rate": 5.5655691309764225e-06,
|
| 792 |
+
"loss": 0.2938,
|
| 793 |
"step": 112
|
| 794 |
},
|
| 795 |
{
|
| 796 |
+
"epoch": 1.5622119815668203,
|
| 797 |
+
"grad_norm": 0.5214616228011532,
|
| 798 |
+
"learning_rate": 5.485049264273241e-06,
|
| 799 |
+
"loss": 0.2598,
|
| 800 |
"step": 113
|
| 801 |
},
|
| 802 |
{
|
| 803 |
+
"epoch": 1.576036866359447,
|
| 804 |
+
"grad_norm": 0.5442190086345122,
|
| 805 |
+
"learning_rate": 5.404402201807022e-06,
|
| 806 |
+
"loss": 0.2587,
|
| 807 |
"step": 114
|
| 808 |
},
|
| 809 |
{
|
| 810 |
+
"epoch": 1.5898617511520738,
|
| 811 |
+
"grad_norm": 0.6223597688467173,
|
| 812 |
+
"learning_rate": 5.323649091872179e-06,
|
| 813 |
+
"loss": 0.2917,
|
| 814 |
"step": 115
|
| 815 |
},
|
| 816 |
{
|
| 817 |
+
"epoch": 1.6036866359447006,
|
| 818 |
+
"grad_norm": 0.5267088823599309,
|
| 819 |
+
"learning_rate": 5.242811110572243e-06,
|
| 820 |
+
"loss": 0.2357,
|
| 821 |
"step": 116
|
| 822 |
},
|
| 823 |
{
|
| 824 |
+
"epoch": 1.6175115207373272,
|
| 825 |
+
"grad_norm": 0.5310849049848287,
|
| 826 |
+
"learning_rate": 5.161909456266781e-06,
|
| 827 |
+
"loss": 0.2259,
|
| 828 |
"step": 117
|
| 829 |
},
|
| 830 |
{
|
| 831 |
+
"epoch": 1.631336405529954,
|
| 832 |
+
"grad_norm": 0.5327550485763282,
|
| 833 |
+
"learning_rate": 5.080965344012509e-06,
|
| 834 |
+
"loss": 0.2918,
|
| 835 |
"step": 118
|
| 836 |
},
|
| 837 |
{
|
| 838 |
+
"epoch": 1.6451612903225805,
|
| 839 |
+
"grad_norm": 0.48433766437116366,
|
| 840 |
+
"learning_rate": 5e-06,
|
| 841 |
+
"loss": 0.2057,
|
| 842 |
"step": 119
|
| 843 |
},
|
| 844 |
{
|
| 845 |
+
"epoch": 1.6589861751152073,
|
| 846 |
+
"grad_norm": 0.5201772106049566,
|
| 847 |
+
"learning_rate": 4.919034655987493e-06,
|
| 848 |
+
"loss": 0.2469,
|
| 849 |
"step": 120
|
| 850 |
},
|
| 851 |
{
|
| 852 |
+
"epoch": 1.672811059907834,
|
| 853 |
+
"grad_norm": 0.5728059995080512,
|
| 854 |
+
"learning_rate": 4.838090543733222e-06,
|
| 855 |
+
"loss": 0.2977,
|
| 856 |
"step": 121
|
| 857 |
},
|
| 858 |
{
|
| 859 |
+
"epoch": 1.6866359447004609,
|
| 860 |
+
"grad_norm": 0.5451785777689631,
|
| 861 |
+
"learning_rate": 4.757188889427761e-06,
|
| 862 |
+
"loss": 0.2664,
|
| 863 |
"step": 122
|
| 864 |
},
|
| 865 |
{
|
| 866 |
+
"epoch": 1.7004608294930876,
|
| 867 |
+
"grad_norm": 0.55325878163931,
|
| 868 |
+
"learning_rate": 4.6763509081278215e-06,
|
| 869 |
+
"loss": 0.2938,
|
| 870 |
"step": 123
|
| 871 |
},
|
| 872 |
{
|
| 873 |
+
"epoch": 1.7142857142857144,
|
| 874 |
+
"grad_norm": 0.5482322683176598,
|
| 875 |
+
"learning_rate": 4.59559779819298e-06,
|
| 876 |
+
"loss": 0.2918,
|
| 877 |
"step": 124
|
| 878 |
},
|
| 879 |
{
|
| 880 |
+
"epoch": 1.728110599078341,
|
| 881 |
+
"grad_norm": 0.5317903391669024,
|
| 882 |
+
"learning_rate": 4.51495073572676e-06,
|
| 883 |
+
"loss": 0.2593,
|
| 884 |
"step": 125
|
| 885 |
},
|
| 886 |
{
|
| 887 |
+
"epoch": 1.7419354838709677,
|
| 888 |
+
"grad_norm": 0.5947363067321887,
|
| 889 |
+
"learning_rate": 4.434430869023579e-06,
|
| 890 |
+
"loss": 0.2565,
|
| 891 |
"step": 126
|
| 892 |
},
|
| 893 |
{
|
| 894 |
+
"epoch": 1.7557603686635943,
|
| 895 |
+
"grad_norm": 0.5166626884834731,
|
| 896 |
+
"learning_rate": 4.3540593130229695e-06,
|
| 897 |
+
"loss": 0.2542,
|
| 898 |
"step": 127
|
| 899 |
},
|
| 900 |
{
|
| 901 |
+
"epoch": 1.769585253456221,
|
| 902 |
+
"grad_norm": 0.5716701172322802,
|
| 903 |
+
"learning_rate": 4.27385714377255e-06,
|
| 904 |
+
"loss": 0.2809,
|
| 905 |
"step": 128
|
| 906 |
},
|
| 907 |
{
|
| 908 |
+
"epoch": 1.7834101382488479,
|
| 909 |
+
"grad_norm": 0.548457756610785,
|
| 910 |
+
"learning_rate": 4.1938453929012014e-06,
|
| 911 |
+
"loss": 0.2656,
|
| 912 |
"step": 129
|
| 913 |
},
|
| 914 |
{
|
| 915 |
+
"epoch": 1.7972350230414746,
|
| 916 |
+
"grad_norm": 0.5101224008525175,
|
| 917 |
+
"learning_rate": 4.1140450421038865e-06,
|
| 918 |
+
"loss": 0.2602,
|
| 919 |
"step": 130
|
| 920 |
},
|
| 921 |
{
|
| 922 |
+
"epoch": 1.8110599078341014,
|
| 923 |
+
"grad_norm": 0.5688581933379532,
|
| 924 |
+
"learning_rate": 4.034477017639561e-06,
|
| 925 |
+
"loss": 0.2708,
|
| 926 |
"step": 131
|
| 927 |
},
|
| 928 |
{
|
| 929 |
+
"epoch": 1.8248847926267282,
|
| 930 |
+
"grad_norm": 0.5608572985897213,
|
| 931 |
+
"learning_rate": 3.955162184843625e-06,
|
| 932 |
+
"loss": 0.2728,
|
| 933 |
"step": 132
|
| 934 |
},
|
| 935 |
{
|
| 936 |
+
"epoch": 1.838709677419355,
|
| 937 |
+
"grad_norm": 0.60913972327474,
|
| 938 |
+
"learning_rate": 3.8761213426563546e-06,
|
| 939 |
+
"loss": 0.2763,
|
| 940 |
"step": 133
|
| 941 |
},
|
| 942 |
{
|
| 943 |
+
"epoch": 1.8525345622119815,
|
| 944 |
+
"grad_norm": 0.5498604689214065,
|
| 945 |
+
"learning_rate": 3.7973752181687336e-06,
|
| 946 |
+
"loss": 0.2574,
|
| 947 |
"step": 134
|
| 948 |
},
|
| 949 |
{
|
| 950 |
+
"epoch": 1.8663594470046083,
|
| 951 |
+
"grad_norm": 0.6459792319529313,
|
| 952 |
+
"learning_rate": 3.7189444611871383e-06,
|
| 953 |
+
"loss": 0.2768,
|
| 954 |
"step": 135
|
| 955 |
},
|
| 956 |
{
|
| 957 |
+
"epoch": 1.8801843317972349,
|
| 958 |
+
"grad_norm": 0.6375472108520018,
|
| 959 |
+
"learning_rate": 3.6408496388182857e-06,
|
| 960 |
+
"loss": 0.3172,
|
| 961 |
"step": 136
|
| 962 |
},
|
| 963 |
{
|
| 964 |
+
"epoch": 1.8940092165898617,
|
| 965 |
+
"grad_norm": 0.5323139638568577,
|
| 966 |
+
"learning_rate": 3.5631112300758595e-06,
|
| 967 |
+
"loss": 0.2375,
|
| 968 |
"step": 137
|
| 969 |
},
|
| 970 |
{
|
| 971 |
+
"epoch": 1.9078341013824884,
|
| 972 |
+
"grad_norm": 0.5704026138530017,
|
| 973 |
+
"learning_rate": 3.4857496205102475e-06,
|
| 974 |
+
"loss": 0.2889,
|
| 975 |
"step": 138
|
| 976 |
},
|
| 977 |
{
|
| 978 |
+
"epoch": 1.9216589861751152,
|
| 979 |
+
"grad_norm": 0.5751626829380007,
|
| 980 |
+
"learning_rate": 3.4087850968627823e-06,
|
| 981 |
+
"loss": 0.2668,
|
| 982 |
"step": 139
|
| 983 |
},
|
| 984 |
{
|
| 985 |
+
"epoch": 1.935483870967742,
|
| 986 |
+
"grad_norm": 0.5176599089594356,
|
| 987 |
+
"learning_rate": 3.3322378417458985e-06,
|
| 988 |
+
"loss": 0.2675,
|
| 989 |
"step": 140
|
| 990 |
},
|
| 991 |
{
|
| 992 |
+
"epoch": 1.9493087557603688,
|
| 993 |
+
"grad_norm": 0.605437544433559,
|
| 994 |
+
"learning_rate": 3.2561279283505888e-06,
|
| 995 |
+
"loss": 0.2711,
|
| 996 |
"step": 141
|
| 997 |
},
|
| 998 |
{
|
| 999 |
+
"epoch": 1.9631336405529956,
|
| 1000 |
+
"grad_norm": 0.5219033376075164,
|
| 1001 |
+
"learning_rate": 3.180475315182563e-06,
|
| 1002 |
+
"loss": 0.2665,
|
| 1003 |
"step": 142
|
| 1004 |
},
|
| 1005 |
{
|
| 1006 |
+
"epoch": 1.976958525345622,
|
| 1007 |
+
"grad_norm": 0.6188916507167108,
|
| 1008 |
+
"learning_rate": 3.1052998408284664e-06,
|
| 1009 |
+
"loss": 0.2708,
|
| 1010 |
"step": 143
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
+
"epoch": 1.9907834101382489,
|
| 1014 |
+
"grad_norm": 0.5752086490165061,
|
| 1015 |
+
"learning_rate": 3.0306212187535653e-06,
|
| 1016 |
+
"loss": 0.2798,
|
| 1017 |
"step": 144
|
| 1018 |
},
|
| 1019 |
{
|
| 1020 |
+
"epoch": 2.0046082949308754,
|
| 1021 |
+
"grad_norm": 0.9737614239717218,
|
| 1022 |
+
"learning_rate": 2.9564590321322206e-06,
|
| 1023 |
+
"loss": 0.3732,
|
| 1024 |
"step": 145
|
| 1025 |
},
|
| 1026 |
{
|
| 1027 |
+
"epoch": 2.0184331797235022,
|
| 1028 |
+
"grad_norm": 0.6171348142735417,
|
| 1029 |
+
"learning_rate": 2.882832728712551e-06,
|
| 1030 |
+
"loss": 0.2203,
|
| 1031 |
"step": 146
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
+
"epoch": 2.032258064516129,
|
| 1035 |
+
"grad_norm": 0.5478831551166238,
|
| 1036 |
+
"learning_rate": 2.8097616157165886e-06,
|
| 1037 |
+
"loss": 0.196,
|
| 1038 |
"step": 147
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
+
"epoch": 2.046082949308756,
|
| 1042 |
+
"grad_norm": 0.5300984312964834,
|
| 1043 |
+
"learning_rate": 2.7372648547773063e-06,
|
| 1044 |
+
"loss": 0.2146,
|
| 1045 |
"step": 148
|
| 1046 |
},
|
| 1047 |
{
|
| 1048 |
+
"epoch": 2.0599078341013826,
|
| 1049 |
+
"grad_norm": 0.527691094200599,
|
| 1050 |
+
"learning_rate": 2.665361456913797e-06,
|
| 1051 |
+
"loss": 0.2404,
|
| 1052 |
"step": 149
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
+
"epoch": 2.0737327188940093,
|
| 1056 |
+
"grad_norm": 0.5203329902273544,
|
| 1057 |
+
"learning_rate": 2.594070277545975e-06,
|
| 1058 |
+
"loss": 0.2051,
|
| 1059 |
"step": 150
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
+
"epoch": 2.087557603686636,
|
| 1063 |
+
"grad_norm": 0.5500390166449323,
|
| 1064 |
+
"learning_rate": 2.5234100115500643e-06,
|
| 1065 |
+
"loss": 0.2648,
|
| 1066 |
"step": 151
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
+
"epoch": 2.1013824884792625,
|
| 1070 |
+
"grad_norm": 0.4893779390529743,
|
| 1071 |
+
"learning_rate": 2.4533991883561868e-06,
|
| 1072 |
+
"loss": 0.2245,
|
| 1073 |
"step": 152
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
+
"epoch": 2.1152073732718892,
|
| 1077 |
+
"grad_norm": 0.5441384122952133,
|
| 1078 |
+
"learning_rate": 2.38405616708935e-06,
|
| 1079 |
+
"loss": 0.2162,
|
| 1080 |
"step": 153
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
+
"epoch": 2.129032258064516,
|
| 1084 |
+
"grad_norm": 0.6239869837559884,
|
| 1085 |
+
"learning_rate": 2.315399131755081e-06,
|
| 1086 |
+
"loss": 0.3197,
|
| 1087 |
"step": 154
|
| 1088 |
},
|
| 1089 |
{
|
| 1090 |
+
"epoch": 2.142857142857143,
|
| 1091 |
+
"grad_norm": 0.5137901138193296,
|
| 1092 |
+
"learning_rate": 2.2474460864709825e-06,
|
| 1093 |
+
"loss": 0.2537,
|
| 1094 |
"step": 155
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
+
"epoch": 2.1566820276497696,
|
| 1098 |
+
"grad_norm": 0.5666929612913303,
|
| 1099 |
+
"learning_rate": 2.1802148507454675e-06,
|
| 1100 |
+
"loss": 0.245,
|
| 1101 |
"step": 156
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
+
"epoch": 2.1705069124423964,
|
| 1105 |
+
"grad_norm": 0.522249344639658,
|
| 1106 |
+
"learning_rate": 2.1137230548049042e-06,
|
| 1107 |
+
"loss": 0.2074,
|
| 1108 |
"step": 157
|
| 1109 |
},
|
| 1110 |
{
|
| 1111 |
+
"epoch": 2.184331797235023,
|
| 1112 |
+
"grad_norm": 0.5427924264706615,
|
| 1113 |
+
"learning_rate": 2.0479881349703885e-06,
|
| 1114 |
+
"loss": 0.1891,
|
| 1115 |
"step": 158
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
+
"epoch": 2.19815668202765,
|
| 1119 |
+
"grad_norm": 0.5395578092838311,
|
| 1120 |
+
"learning_rate": 1.983027329085377e-06,
|
| 1121 |
+
"loss": 0.2141,
|
| 1122 |
"step": 159
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
+
"epoch": 2.2119815668202767,
|
| 1126 |
+
"grad_norm": 0.6203214366115929,
|
| 1127 |
+
"learning_rate": 1.9188576719953635e-06,
|
| 1128 |
+
"loss": 0.2395,
|
| 1129 |
"step": 160
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
+
"epoch": 2.225806451612903,
|
| 1133 |
+
"grad_norm": 0.5561057678969209,
|
| 1134 |
+
"learning_rate": 1.8554959910807773e-06,
|
| 1135 |
+
"loss": 0.2325,
|
| 1136 |
"step": 161
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
+
"epoch": 2.23963133640553,
|
| 1140 |
+
"grad_norm": 0.4975297831721625,
|
| 1141 |
+
"learning_rate": 1.7929589018443016e-06,
|
| 1142 |
+
"loss": 0.2189,
|
| 1143 |
"step": 162
|
| 1144 |
},
|
| 1145 |
{
|
| 1146 |
+
"epoch": 2.2534562211981566,
|
| 1147 |
+
"grad_norm": 0.51442024085124,
|
| 1148 |
+
"learning_rate": 1.7312628035537388e-06,
|
| 1149 |
+
"loss": 0.1941,
|
| 1150 |
"step": 163
|
| 1151 |
},
|
| 1152 |
{
|
| 1153 |
+
"epoch": 2.2672811059907834,
|
| 1154 |
+
"grad_norm": 0.5689991394989393,
|
| 1155 |
+
"learning_rate": 1.6704238749415958e-06,
|
| 1156 |
+
"loss": 0.2274,
|
| 1157 |
"step": 164
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
+
"epoch": 2.28110599078341,
|
| 1161 |
+
"grad_norm": 0.5545628564372191,
|
| 1162 |
+
"learning_rate": 1.6104580699624839e-06,
|
| 1163 |
+
"loss": 0.2582,
|
| 1164 |
"step": 165
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
+
"epoch": 2.294930875576037,
|
| 1168 |
+
"grad_norm": 0.5532036253850108,
|
| 1169 |
+
"learning_rate": 1.5513811136094786e-06,
|
| 1170 |
+
"loss": 0.2518,
|
| 1171 |
"step": 166
|
| 1172 |
},
|
| 1173 |
{
|
| 1174 |
+
"epoch": 2.3087557603686637,
|
| 1175 |
+
"grad_norm": 0.4892665339324161,
|
| 1176 |
+
"learning_rate": 1.4932084977905043e-06,
|
| 1177 |
+
"loss": 0.1982,
|
| 1178 |
"step": 167
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
+
"epoch": 2.3225806451612905,
|
| 1182 |
+
"grad_norm": 0.551947647990794,
|
| 1183 |
+
"learning_rate": 1.4359554772658551e-06,
|
| 1184 |
+
"loss": 0.2607,
|
| 1185 |
"step": 168
|
| 1186 |
},
|
| 1187 |
{
|
| 1188 |
+
"epoch": 2.3364055299539173,
|
| 1189 |
+
"grad_norm": 0.5151354916631968,
|
| 1190 |
+
"learning_rate": 1.3796370656478936e-06,
|
| 1191 |
+
"loss": 0.2072,
|
| 1192 |
"step": 169
|
| 1193 |
},
|
| 1194 |
{
|
| 1195 |
+
"epoch": 2.3502304147465436,
|
| 1196 |
+
"grad_norm": 0.5685893753938921,
|
| 1197 |
+
"learning_rate": 1.3242680314639995e-06,
|
| 1198 |
+
"loss": 0.2143,
|
| 1199 |
"step": 170
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
+
"epoch": 2.3640552995391704,
|
| 1203 |
+
"grad_norm": 0.4880909328371241,
|
| 1204 |
+
"learning_rate": 1.2698628942837698e-06,
|
| 1205 |
+
"loss": 0.1718,
|
| 1206 |
"step": 171
|
| 1207 |
},
|
| 1208 |
{
|
| 1209 |
+
"epoch": 2.377880184331797,
|
| 1210 |
+
"grad_norm": 0.493208956793857,
|
| 1211 |
+
"learning_rate": 1.2164359209115235e-06,
|
| 1212 |
+
"loss": 0.1923,
|
| 1213 |
"step": 172
|
| 1214 |
},
|
| 1215 |
{
|
| 1216 |
+
"epoch": 2.391705069124424,
|
| 1217 |
+
"grad_norm": 0.5582991832891734,
|
| 1218 |
+
"learning_rate": 1.164001121645069e-06,
|
| 1219 |
+
"loss": 0.2524,
|
| 1220 |
"step": 173
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
+
"epoch": 2.4055299539170507,
|
| 1224 |
+
"grad_norm": 0.505450721310959,
|
| 1225 |
+
"learning_rate": 1.1125722466017547e-06,
|
| 1226 |
+
"loss": 0.2125,
|
| 1227 |
"step": 174
|
| 1228 |
},
|
| 1229 |
{
|
| 1230 |
+
"epoch": 2.4193548387096775,
|
| 1231 |
+
"grad_norm": 0.5255011925428328,
|
| 1232 |
+
"learning_rate": 1.062162782112729e-06,
|
| 1233 |
+
"loss": 0.2456,
|
| 1234 |
"step": 175
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
+
"epoch": 2.4331797235023043,
|
| 1238 |
+
"grad_norm": 0.4989323482397167,
|
| 1239 |
+
"learning_rate": 1.012785947186397e-06,
|
| 1240 |
+
"loss": 0.1834,
|
| 1241 |
"step": 176
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
+
"epoch": 2.447004608294931,
|
| 1245 |
+
"grad_norm": 0.51435324430214,
|
| 1246 |
+
"learning_rate": 9.644546900419533e-07,
|
| 1247 |
+
"loss": 0.2056,
|
| 1248 |
"step": 177
|
| 1249 |
},
|
| 1250 |
{
|
| 1251 |
+
"epoch": 2.460829493087558,
|
| 1252 |
+
"grad_norm": 0.4941625397188085,
|
| 1253 |
+
"learning_rate": 9.171816847139447e-07,
|
| 1254 |
+
"loss": 0.2044,
|
| 1255 |
"step": 178
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
+
"epoch": 2.474654377880184,
|
| 1259 |
+
"grad_norm": 0.5742862761852323,
|
| 1260 |
+
"learning_rate": 8.709793277287182e-07,
|
| 1261 |
+
"loss": 0.2334,
|
| 1262 |
"step": 179
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
+
"epoch": 2.488479262672811,
|
| 1266 |
+
"grad_norm": 0.45964093874741846,
|
| 1267 |
+
"learning_rate": 8.258597348536452e-07,
|
| 1268 |
+
"loss": 0.1885,
|
| 1269 |
"step": 180
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
+
"epoch": 2.5023041474654377,
|
| 1273 |
+
"grad_norm": 0.5736994982571209,
|
| 1274 |
+
"learning_rate": 7.818347379199781e-07,
|
| 1275 |
+
"loss": 0.2601,
|
| 1276 |
"step": 181
|
| 1277 |
},
|
| 1278 |
{
|
| 1279 |
+
"epoch": 2.5161290322580645,
|
| 1280 |
+
"grad_norm": 0.4946177722077173,
|
| 1281 |
+
"learning_rate": 7.389158817201541e-07,
|
| 1282 |
+
"loss": 0.1937,
|
| 1283 |
"step": 182
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
+
"epoch": 2.5299539170506913,
|
| 1287 |
+
"grad_norm": 0.5484333491191469,
|
| 1288 |
+
"learning_rate": 6.971144209803738e-07,
|
| 1289 |
+
"loss": 0.2487,
|
| 1290 |
"step": 183
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
+
"epoch": 2.543778801843318,
|
| 1294 |
+
"grad_norm": 0.5338015925288597,
|
| 1295 |
+
"learning_rate": 6.564413174092443e-07,
|
| 1296 |
+
"loss": 0.2134,
|
| 1297 |
"step": 184
|
| 1298 |
},
|
| 1299 |
{
|
| 1300 |
+
"epoch": 2.557603686635945,
|
| 1301 |
+
"grad_norm": 0.4987205048741818,
|
| 1302 |
+
"learning_rate": 6.16907236823262e-07,
|
| 1303 |
+
"loss": 0.2115,
|
| 1304 |
"step": 185
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
+
"epoch": 2.571428571428571,
|
| 1308 |
+
"grad_norm": 0.5116461576836463,
|
| 1309 |
+
"learning_rate": 5.785225463498828e-07,
|
| 1310 |
+
"loss": 0.2161,
|
| 1311 |
"step": 186
|
| 1312 |
},
|
| 1313 |
{
|
| 1314 |
+
"epoch": 2.5852534562211984,
|
| 1315 |
+
"grad_norm": 0.5549460042068343,
|
| 1316 |
+
"learning_rate": 5.412973117089288e-07,
|
| 1317 |
+
"loss": 0.249,
|
| 1318 |
"step": 187
|
| 1319 |
},
|
| 1320 |
{
|
| 1321 |
+
"epoch": 2.5990783410138247,
|
| 1322 |
+
"grad_norm": 0.4625486356955096,
|
| 1323 |
+
"learning_rate": 5.05241294573024e-07,
|
| 1324 |
+
"loss": 0.1794,
|
| 1325 |
"step": 188
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
+
"epoch": 2.6129032258064515,
|
| 1329 |
+
"grad_norm": 0.5601735455591889,
|
| 1330 |
+
"learning_rate": 4.7036395000776556e-07,
|
| 1331 |
+
"loss": 0.2483,
|
| 1332 |
"step": 189
|
| 1333 |
},
|
| 1334 |
{
|
| 1335 |
+
"epoch": 2.6267281105990783,
|
| 1336 |
+
"grad_norm": 0.4873480362305034,
|
| 1337 |
+
"learning_rate": 4.3667442399229985e-07,
|
| 1338 |
+
"loss": 0.1891,
|
| 1339 |
"step": 190
|
| 1340 |
},
|
| 1341 |
{
|
| 1342 |
+
"epoch": 2.640552995391705,
|
| 1343 |
+
"grad_norm": 0.5036174624248448,
|
| 1344 |
+
"learning_rate": 4.041815510209396e-07,
|
| 1345 |
+
"loss": 0.2294,
|
| 1346 |
"step": 191
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
+
"epoch": 2.654377880184332,
|
| 1350 |
+
"grad_norm": 0.5288779116171689,
|
| 1351 |
+
"learning_rate": 3.728938517864794e-07,
|
| 1352 |
+
"loss": 0.2278,
|
| 1353 |
"step": 192
|
| 1354 |
},
|
| 1355 |
{
|
| 1356 |
+
"epoch": 2.6682027649769586,
|
| 1357 |
+
"grad_norm": 0.5090457286467068,
|
| 1358 |
+
"learning_rate": 3.4281953094578877e-07,
|
| 1359 |
+
"loss": 0.2144,
|
| 1360 |
"step": 193
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
+
"epoch": 2.6820276497695854,
|
| 1364 |
+
"grad_norm": 0.5256240057085287,
|
| 1365 |
+
"learning_rate": 3.1396647496828245e-07,
|
| 1366 |
+
"loss": 0.2517,
|
| 1367 |
"step": 194
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
+
"epoch": 2.6958525345622117,
|
| 1371 |
+
"grad_norm": 0.5629758036361966,
|
| 1372 |
+
"learning_rate": 2.8634225006782867e-07,
|
| 1373 |
+
"loss": 0.2611,
|
| 1374 |
"step": 195
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
+
"epoch": 2.709677419354839,
|
| 1378 |
+
"grad_norm": 0.5114348887925635,
|
| 1379 |
+
"learning_rate": 2.599541002186479e-07,
|
| 1380 |
+
"loss": 0.2059,
|
| 1381 |
"step": 196
|
| 1382 |
},
|
| 1383 |
{
|
| 1384 |
+
"epoch": 2.7235023041474653,
|
| 1385 |
+
"grad_norm": 0.467691586903513,
|
| 1386 |
+
"learning_rate": 2.3480894525569564e-07,
|
| 1387 |
+
"loss": 0.1805,
|
| 1388 |
"step": 197
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
+
"epoch": 2.737327188940092,
|
| 1392 |
+
"grad_norm": 0.5136153625365067,
|
| 1393 |
+
"learning_rate": 2.109133790600648e-07,
|
| 1394 |
+
"loss": 0.2479,
|
| 1395 |
"step": 198
|
| 1396 |
},
|
| 1397 |
{
|
| 1398 |
+
"epoch": 2.751152073732719,
|
| 1399 |
+
"grad_norm": 0.521711267048856,
|
| 1400 |
+
"learning_rate": 1.8827366782984913e-07,
|
| 1401 |
+
"loss": 0.2388,
|
| 1402 |
"step": 199
|
| 1403 |
},
|
| 1404 |
{
|
| 1405 |
+
"epoch": 2.7649769585253456,
|
| 1406 |
+
"grad_norm": 0.5264204516952466,
|
| 1407 |
+
"learning_rate": 1.6689574843694433e-07,
|
| 1408 |
+
"loss": 0.2238,
|
| 1409 |
"step": 200
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
+
"epoch": 2.7788018433179724,
|
| 1413 |
+
"grad_norm": 0.5123667240282413,
|
| 1414 |
+
"learning_rate": 1.4678522687020414e-07,
|
| 1415 |
+
"loss": 0.1927,
|
| 1416 |
"step": 201
|
| 1417 |
},
|
| 1418 |
{
|
| 1419 |
+
"epoch": 2.792626728110599,
|
| 1420 |
+
"grad_norm": 0.5109045161700035,
|
| 1421 |
+
"learning_rate": 1.2794737676536993e-07,
|
| 1422 |
+
"loss": 0.219,
|
| 1423 |
"step": 202
|
| 1424 |
},
|
| 1425 |
{
|
| 1426 |
+
"epoch": 2.806451612903226,
|
| 1427 |
+
"grad_norm": 0.5412780752972016,
|
| 1428 |
+
"learning_rate": 1.1038713802214718e-07,
|
| 1429 |
+
"loss": 0.2258,
|
| 1430 |
"step": 203
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
+
"epoch": 2.8202764976958523,
|
| 1434 |
+
"grad_norm": 0.5001657996677221,
|
| 1435 |
+
"learning_rate": 9.410911550880474e-08,
|
| 1436 |
+
"loss": 0.2272,
|
| 1437 |
"step": 204
|
| 1438 |
},
|
| 1439 |
{
|
| 1440 |
+
"epoch": 2.8341013824884795,
|
| 1441 |
+
"grad_norm": 0.4830345984951105,
|
| 1442 |
+
"learning_rate": 7.911757785462882e-08,
|
| 1443 |
+
"loss": 0.1845,
|
| 1444 |
"step": 205
|
| 1445 |
},
|
| 1446 |
{
|
| 1447 |
+
"epoch": 2.847926267281106,
|
| 1448 |
+
"grad_norm": 0.4754821970835569,
|
| 1449 |
+
"learning_rate": 6.54164563305465e-08,
|
| 1450 |
+
"loss": 0.1772,
|
| 1451 |
"step": 206
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
+
"epoch": 2.8617511520737327,
|
| 1455 |
+
"grad_norm": 0.5180006446663228,
|
| 1456 |
+
"learning_rate": 5.3009343818219985e-08,
|
| 1457 |
+
"loss": 0.2429,
|
| 1458 |
"step": 207
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
+
"epoch": 2.8755760368663594,
|
| 1462 |
+
"grad_norm": 0.4922784862533948,
|
| 1463 |
+
"learning_rate": 4.189949386787462e-08,
|
| 1464 |
+
"loss": 0.2,
|
| 1465 |
"step": 208
|
| 1466 |
},
|
| 1467 |
{
|
| 1468 |
+
"epoch": 2.889400921658986,
|
| 1469 |
+
"grad_norm": 0.5733544961657199,
|
| 1470 |
+
"learning_rate": 3.2089819845111946e-08,
|
| 1471 |
+
"loss": 0.2812,
|
| 1472 |
"step": 209
|
| 1473 |
},
|
| 1474 |
{
|
| 1475 |
+
"epoch": 2.903225806451613,
|
| 1476 |
+
"grad_norm": 0.5752601209401671,
|
| 1477 |
+
"learning_rate": 2.358289416693027e-08,
|
| 1478 |
+
"loss": 0.2579,
|
| 1479 |
"step": 210
|
| 1480 |
},
|
| 1481 |
{
|
| 1482 |
+
"epoch": 2.9170506912442398,
|
| 1483 |
+
"grad_norm": 0.4757128270112064,
|
| 1484 |
+
"learning_rate": 1.6380947627153143e-08,
|
| 1485 |
+
"loss": 0.2217,
|
| 1486 |
"step": 211
|
| 1487 |
},
|
| 1488 |
{
|
| 1489 |
+
"epoch": 2.9308755760368665,
|
| 1490 |
+
"grad_norm": 0.5139378693377304,
|
| 1491 |
+
"learning_rate": 1.0485868811441757e-08,
|
| 1492 |
+
"loss": 0.2273,
|
| 1493 |
"step": 212
|
| 1494 |
},
|
| 1495 |
{
|
| 1496 |
+
"epoch": 2.944700460829493,
|
| 1497 |
+
"grad_norm": 0.5102846639395582,
|
| 1498 |
+
"learning_rate": 5.899203602046655e-09,
|
| 1499 |
+
"loss": 0.2158,
|
| 1500 |
"step": 213
|
| 1501 |
},
|
| 1502 |
{
|
| 1503 |
+
"epoch": 2.9585253456221197,
|
| 1504 |
+
"grad_norm": 0.5057982261906225,
|
| 1505 |
+
"learning_rate": 2.6221547724253337e-09,
|
| 1506 |
+
"loss": 0.2063,
|
| 1507 |
"step": 214
|
| 1508 |
},
|
| 1509 |
{
|
| 1510 |
+
"epoch": 2.9723502304147464,
|
| 1511 |
+
"grad_norm": 0.5645707694001406,
|
| 1512 |
+
"learning_rate": 6.555816718389896e-10,
|
| 1513 |
+
"loss": 0.2588,
|
| 1514 |
"step": 215
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
+
"epoch": 2.986175115207373,
|
| 1518 |
+
"grad_norm": 0.48505546530830695,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1519 |
"learning_rate": 0.0,
|
| 1520 |
+
"loss": 0.2179,
|
| 1521 |
+
"step": 216
|
| 1522 |
},
|
| 1523 |
{
|
| 1524 |
+
"epoch": 2.986175115207373,
|
| 1525 |
+
"step": 216,
|
| 1526 |
+
"total_flos": 32031101919232.0,
|
| 1527 |
+
"train_loss": 0.28372171659160544,
|
| 1528 |
+
"train_runtime": 1010.9103,
|
| 1529 |
+
"train_samples_per_second": 20.539,
|
| 1530 |
+
"train_steps_per_second": 0.214
|
| 1531 |
}
|
| 1532 |
],
|
| 1533 |
"logging_steps": 1,
|
| 1534 |
+
"max_steps": 216,
|
| 1535 |
"num_input_tokens_seen": 0,
|
| 1536 |
"num_train_epochs": 3,
|
| 1537 |
"save_steps": 500,
|
|
|
|
| 1547 |
"attributes": {}
|
| 1548 |
}
|
| 1549 |
},
|
| 1550 |
+
"total_flos": 32031101919232.0,
|
| 1551 |
+
"train_batch_size": 1,
|
| 1552 |
"trial_name": null,
|
| 1553 |
"trial_params": null
|
| 1554 |
}
|
training_loss.png
CHANGED
|
|