Full run push
Browse files- trainer_state.json +210 -700
trainer_state.json
CHANGED
|
@@ -4,866 +4,376 @@
|
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
"epoch": 1.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
-
"epoch": 0.
|
| 14 |
-
"grad_norm": 1.
|
| 15 |
-
"learning_rate":
|
| 16 |
-
"loss": 0.
|
| 17 |
"step": 500
|
| 18 |
},
|
| 19 |
{
|
| 20 |
-
"epoch": 0.
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 0.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 1000
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"epoch": 0.
|
| 28 |
-
"grad_norm": 1.
|
| 29 |
-
"learning_rate": 0.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 1500
|
| 32 |
},
|
| 33 |
{
|
| 34 |
-
"epoch": 0.
|
| 35 |
-
"grad_norm": 1.
|
| 36 |
-
"learning_rate": 0.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 2000
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 0.
|
| 42 |
-
"grad_norm": 1.
|
| 43 |
-
"learning_rate": 0.
|
| 44 |
-
"loss": 0.
|
| 45 |
"step": 2500
|
| 46 |
},
|
| 47 |
{
|
| 48 |
-
"epoch": 0.
|
| 49 |
-
"grad_norm":
|
| 50 |
-
"learning_rate": 0.
|
| 51 |
-
"loss": 0.
|
| 52 |
"step": 3000
|
| 53 |
},
|
| 54 |
{
|
| 55 |
-
"epoch": 0.
|
| 56 |
-
"grad_norm":
|
| 57 |
-
"learning_rate": 0.
|
| 58 |
-
"loss": 0.
|
| 59 |
"step": 3500
|
| 60 |
},
|
| 61 |
{
|
| 62 |
-
"epoch": 0.
|
| 63 |
-
"grad_norm":
|
| 64 |
-
"learning_rate": 0.
|
| 65 |
-
"loss": 0.
|
| 66 |
"step": 4000
|
| 67 |
},
|
| 68 |
{
|
| 69 |
-
"epoch": 0.
|
| 70 |
-
"grad_norm":
|
| 71 |
-
"learning_rate": 0.
|
| 72 |
-
"loss": 0.
|
| 73 |
"step": 4500
|
| 74 |
},
|
| 75 |
{
|
| 76 |
-
"epoch": 0.
|
| 77 |
-
"grad_norm":
|
| 78 |
-
"learning_rate": 0.
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 5000
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch": 0.
|
| 84 |
-
"grad_norm":
|
| 85 |
-
"learning_rate": 0.
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 5500
|
| 88 |
},
|
| 89 |
{
|
| 90 |
-
"epoch": 0.
|
| 91 |
-
"grad_norm":
|
| 92 |
-
"learning_rate": 0.
|
| 93 |
-
"loss": 0.
|
| 94 |
"step": 6000
|
| 95 |
},
|
| 96 |
{
|
| 97 |
-
"epoch": 0.
|
| 98 |
-
"grad_norm":
|
| 99 |
-
"learning_rate": 0.
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 6500
|
| 102 |
},
|
| 103 |
{
|
| 104 |
-
"epoch": 0.
|
| 105 |
-
"grad_norm": 0.
|
| 106 |
-
"learning_rate": 0.
|
| 107 |
-
"loss": 0.
|
| 108 |
"step": 7000
|
| 109 |
},
|
| 110 |
{
|
| 111 |
-
"epoch": 0.
|
| 112 |
-
"grad_norm":
|
| 113 |
-
"learning_rate": 0.
|
| 114 |
-
"loss": 0.
|
| 115 |
"step": 7500
|
| 116 |
},
|
| 117 |
{
|
| 118 |
-
"epoch": 0.
|
| 119 |
-
"grad_norm":
|
| 120 |
-
"learning_rate": 0.
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 8000
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch": 0.
|
| 126 |
-
"grad_norm":
|
| 127 |
-
"learning_rate": 0.
|
| 128 |
-
"loss": 0.
|
| 129 |
"step": 8500
|
| 130 |
},
|
| 131 |
{
|
| 132 |
-
"epoch": 0.
|
| 133 |
-
"grad_norm":
|
| 134 |
-
"learning_rate": 0.
|
| 135 |
-
"loss": 0.
|
| 136 |
"step": 9000
|
| 137 |
},
|
| 138 |
{
|
| 139 |
-
"epoch": 0.
|
| 140 |
-
"grad_norm":
|
| 141 |
-
"learning_rate": 0.
|
| 142 |
-
"loss": 0.
|
| 143 |
"step": 9500
|
| 144 |
},
|
| 145 |
{
|
| 146 |
-
"epoch": 0.
|
| 147 |
-
"grad_norm": 1.
|
| 148 |
-
"learning_rate": 0.
|
| 149 |
-
"loss": 0.
|
| 150 |
"step": 10000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
-
"epoch": 0.
|
| 154 |
-
"grad_norm": 0.
|
| 155 |
-
"learning_rate": 0.
|
| 156 |
-
"loss": 0.
|
| 157 |
"step": 10500
|
| 158 |
},
|
| 159 |
{
|
| 160 |
-
"epoch": 0.
|
| 161 |
-
"grad_norm":
|
| 162 |
-
"learning_rate": 0.
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 11000
|
| 165 |
},
|
| 166 |
{
|
| 167 |
-
"epoch": 0.
|
| 168 |
-
"grad_norm":
|
| 169 |
-
"learning_rate": 0.
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 11500
|
| 172 |
},
|
| 173 |
{
|
| 174 |
-
"epoch": 0.
|
| 175 |
-
"grad_norm":
|
| 176 |
-
"learning_rate": 0.
|
| 177 |
-
"loss": 0.
|
| 178 |
"step": 12000
|
| 179 |
},
|
| 180 |
{
|
| 181 |
-
"epoch": 0.
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
-
"learning_rate": 0.
|
| 184 |
-
"loss": 0.
|
| 185 |
"step": 12500
|
| 186 |
},
|
| 187 |
{
|
| 188 |
-
"epoch": 0.
|
| 189 |
-
"grad_norm":
|
| 190 |
-
"learning_rate": 0.
|
| 191 |
-
"loss": 0.
|
| 192 |
"step": 13000
|
| 193 |
},
|
| 194 |
{
|
| 195 |
-
"epoch": 0.
|
| 196 |
-
"grad_norm":
|
| 197 |
-
"learning_rate": 0.
|
| 198 |
-
"loss": 0.
|
| 199 |
"step": 13500
|
| 200 |
},
|
| 201 |
{
|
| 202 |
-
"epoch": 0.
|
| 203 |
-
"grad_norm": 0.
|
| 204 |
-
"learning_rate": 0.
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 14000
|
| 207 |
},
|
| 208 |
{
|
| 209 |
-
"epoch": 0.
|
| 210 |
-
"grad_norm": 0.
|
| 211 |
-
"learning_rate": 0.
|
| 212 |
-
"loss": 0.
|
| 213 |
"step": 14500
|
| 214 |
},
|
| 215 |
{
|
| 216 |
-
"epoch": 0.
|
| 217 |
-
"grad_norm":
|
| 218 |
-
"learning_rate": 0.
|
| 219 |
-
"loss": 0.
|
| 220 |
"step": 15000
|
| 221 |
},
|
| 222 |
{
|
| 223 |
-
"epoch": 0.
|
| 224 |
-
"grad_norm":
|
| 225 |
-
"learning_rate":
|
| 226 |
-
"loss": 0.
|
| 227 |
"step": 15500
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"epoch": 0.
|
| 231 |
-
"grad_norm":
|
| 232 |
-
"learning_rate":
|
| 233 |
-
"loss": 0.
|
| 234 |
"step": 16000
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"grad_norm":
|
| 239 |
-
"learning_rate":
|
| 240 |
-
"loss": 0.
|
| 241 |
"step": 16500
|
| 242 |
},
|
| 243 |
{
|
| 244 |
-
"epoch": 0.
|
| 245 |
-
"grad_norm":
|
| 246 |
-
"learning_rate":
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 17000
|
| 249 |
},
|
| 250 |
{
|
| 251 |
-
"epoch": 0.
|
| 252 |
-
"grad_norm": 0.
|
| 253 |
-
"learning_rate":
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 17500
|
| 256 |
},
|
| 257 |
{
|
| 258 |
-
"epoch": 0.
|
| 259 |
-
"grad_norm":
|
| 260 |
-
"learning_rate":
|
| 261 |
-
"loss": 0.
|
| 262 |
"step": 18000
|
| 263 |
},
|
| 264 |
{
|
| 265 |
-
"epoch": 0.
|
| 266 |
-
"grad_norm":
|
| 267 |
-
"learning_rate":
|
| 268 |
-
"loss": 0.
|
| 269 |
"step": 18500
|
| 270 |
},
|
| 271 |
{
|
| 272 |
-
"epoch": 0.
|
| 273 |
-
"grad_norm": 0.
|
| 274 |
-
"learning_rate":
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 19000
|
| 277 |
},
|
| 278 |
{
|
| 279 |
-
"epoch": 0.
|
| 280 |
-
"grad_norm":
|
| 281 |
-
"learning_rate":
|
| 282 |
-
"loss": 0.
|
| 283 |
"step": 19500
|
| 284 |
},
|
| 285 |
{
|
| 286 |
-
"epoch": 0.
|
| 287 |
-
"grad_norm": 1.
|
| 288 |
-
"learning_rate":
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 20000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
-
"epoch": 0.
|
| 294 |
-
"grad_norm": 0.
|
| 295 |
-
"learning_rate":
|
| 296 |
-
"loss": 0.
|
| 297 |
"step": 20500
|
| 298 |
},
|
| 299 |
{
|
| 300 |
-
"epoch": 0.
|
| 301 |
-
"grad_norm":
|
| 302 |
-
"learning_rate":
|
| 303 |
-
"loss": 0.
|
| 304 |
"step": 21000
|
| 305 |
},
|
| 306 |
{
|
| 307 |
-
"epoch": 0.
|
| 308 |
-
"grad_norm":
|
| 309 |
-
"learning_rate":
|
| 310 |
-
"loss": 0.
|
| 311 |
"step": 21500
|
| 312 |
},
|
| 313 |
{
|
| 314 |
-
"epoch": 0.
|
| 315 |
-
"grad_norm":
|
| 316 |
-
"learning_rate":
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 22000
|
| 319 |
},
|
| 320 |
{
|
| 321 |
-
"epoch": 0.
|
| 322 |
-
"grad_norm":
|
| 323 |
-
"learning_rate":
|
| 324 |
-
"loss": 0.
|
| 325 |
"step": 22500
|
| 326 |
},
|
| 327 |
{
|
| 328 |
-
"epoch": 0.
|
| 329 |
-
"grad_norm":
|
| 330 |
-
"learning_rate":
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 23000
|
| 333 |
},
|
| 334 |
{
|
| 335 |
-
"epoch": 0.
|
| 336 |
-
"grad_norm":
|
| 337 |
-
"learning_rate":
|
| 338 |
-
"loss": 0.
|
| 339 |
"step": 23500
|
| 340 |
},
|
| 341 |
{
|
| 342 |
-
"epoch": 0.
|
| 343 |
-
"grad_norm": 0.
|
| 344 |
-
"learning_rate":
|
| 345 |
-
"loss": 0.
|
| 346 |
"step": 24000
|
| 347 |
},
|
| 348 |
{
|
| 349 |
-
"epoch": 0.
|
| 350 |
-
"grad_norm":
|
| 351 |
-
"learning_rate":
|
| 352 |
-
"loss": 0.
|
| 353 |
"step": 24500
|
| 354 |
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.4166666666666667,
|
| 357 |
-
"grad_norm": 0.9933512210845947,
|
| 358 |
-
"learning_rate": 0.0001969758525719492,
|
| 359 |
-
"loss": 0.2592,
|
| 360 |
-
"step": 25000
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.425,
|
| 364 |
-
"grad_norm": 0.8219375610351562,
|
| 365 |
-
"learning_rate": 0.00019311441837478816,
|
| 366 |
-
"loss": 0.272,
|
| 367 |
-
"step": 25500
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.43333333333333335,
|
| 371 |
-
"grad_norm": 1.2546143531799316,
|
| 372 |
-
"learning_rate": 0.0001892215798171928,
|
| 373 |
-
"loss": 0.2567,
|
| 374 |
-
"step": 26000
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.44166666666666665,
|
| 378 |
-
"grad_norm": 1.4476374387741089,
|
| 379 |
-
"learning_rate": 0.00018530017242634363,
|
| 380 |
-
"loss": 0.2588,
|
| 381 |
-
"step": 26500
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.45,
|
| 385 |
-
"grad_norm": 1.356984257698059,
|
| 386 |
-
"learning_rate": 0.0001813530525388389,
|
| 387 |
-
"loss": 0.2556,
|
| 388 |
-
"step": 27000
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.4583333333333333,
|
| 392 |
-
"grad_norm": 0.0,
|
| 393 |
-
"learning_rate": 0.00017738309522015073,
|
| 394 |
-
"loss": 0.2477,
|
| 395 |
-
"step": 27500
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.4666666666666667,
|
| 399 |
-
"grad_norm": 1.395007848739624,
|
| 400 |
-
"learning_rate": 0.00017339319217043995,
|
| 401 |
-
"loss": 0.2498,
|
| 402 |
-
"step": 28000
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.475,
|
| 406 |
-
"grad_norm": 0.7972742319107056,
|
| 407 |
-
"learning_rate": 0.00016938624961825358,
|
| 408 |
-
"loss": 0.2473,
|
| 409 |
-
"step": 28500
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 0.48333333333333334,
|
| 413 |
-
"grad_norm": 0.8678386807441711,
|
| 414 |
-
"learning_rate": 0.0001653651862036404,
|
| 415 |
-
"loss": 0.2534,
|
| 416 |
-
"step": 29000
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.49166666666666664,
|
| 420 |
-
"grad_norm": 1.0032129287719727,
|
| 421 |
-
"learning_rate": 0.00016133293085222585,
|
| 422 |
-
"loss": 0.2458,
|
| 423 |
-
"step": 29500
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 0.5,
|
| 427 |
-
"grad_norm": 0.9749704599380493,
|
| 428 |
-
"learning_rate": 0.00015729242064179487,
|
| 429 |
-
"loss": 0.2516,
|
| 430 |
-
"step": 30000
|
| 431 |
-
},
|
| 432 |
-
{
|
| 433 |
-
"epoch": 0.5083333333333333,
|
| 434 |
-
"grad_norm": 1.6964690685272217,
|
| 435 |
-
"learning_rate": 0.0001532465986629368,
|
| 436 |
-
"loss": 0.2452,
|
| 437 |
-
"step": 30500
|
| 438 |
-
},
|
| 439 |
-
{
|
| 440 |
-
"epoch": 0.5166666666666667,
|
| 441 |
-
"grad_norm": 1.2840642929077148,
|
| 442 |
-
"learning_rate": 0.0001491984118753108,
|
| 443 |
-
"loss": 0.247,
|
| 444 |
-
"step": 31000
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.525,
|
| 448 |
-
"grad_norm": 1.1751660108566284,
|
| 449 |
-
"learning_rate": 0.00014515080896109272,
|
| 450 |
-
"loss": 0.2561,
|
| 451 |
-
"step": 31500
|
| 452 |
-
},
|
| 453 |
-
{
|
| 454 |
-
"epoch": 0.5333333333333333,
|
| 455 |
-
"grad_norm": 0.4990275204181671,
|
| 456 |
-
"learning_rate": 0.00014110673817716756,
|
| 457 |
-
"loss": 0.2429,
|
| 458 |
-
"step": 32000
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.5416666666666666,
|
| 462 |
-
"grad_norm": 1.0772459506988525,
|
| 463 |
-
"learning_rate": 0.0001370691452076314,
|
| 464 |
-
"loss": 0.2622,
|
| 465 |
-
"step": 32500
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"epoch": 0.55,
|
| 469 |
-
"grad_norm": 1.0913983583450317,
|
| 470 |
-
"learning_rate": 0.00013304097101816766,
|
| 471 |
-
"loss": 0.2406,
|
| 472 |
-
"step": 33000
|
| 473 |
-
},
|
| 474 |
-
{
|
| 475 |
-
"epoch": 0.5583333333333333,
|
| 476 |
-
"grad_norm": 0.7937314510345459,
|
| 477 |
-
"learning_rate": 0.0001290251497138601,
|
| 478 |
-
"loss": 0.2317,
|
| 479 |
-
"step": 33500
|
| 480 |
-
},
|
| 481 |
-
{
|
| 482 |
-
"epoch": 0.5666666666666667,
|
| 483 |
-
"grad_norm": 0.866894006729126,
|
| 484 |
-
"learning_rate": 0.0001250246064020032,
|
| 485 |
-
"loss": 0.2381,
|
| 486 |
-
"step": 34000
|
| 487 |
-
},
|
| 488 |
-
{
|
| 489 |
-
"epoch": 0.575,
|
| 490 |
-
"grad_norm": 0.904768705368042,
|
| 491 |
-
"learning_rate": 0.000121042255061466,
|
| 492 |
-
"loss": 0.2416,
|
| 493 |
-
"step": 34500
|
| 494 |
-
},
|
| 495 |
-
{
|
| 496 |
-
"epoch": 0.5833333333333334,
|
| 497 |
-
"grad_norm": 2.5710256099700928,
|
| 498 |
-
"learning_rate": 0.00011708099642016254,
|
| 499 |
-
"loss": 0.2501,
|
| 500 |
-
"step": 35000
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.5916666666666667,
|
| 504 |
-
"grad_norm": 0.9502021074295044,
|
| 505 |
-
"learning_rate": 0.00011314371584217354,
|
| 506 |
-
"loss": 0.2462,
|
| 507 |
-
"step": 35500
|
| 508 |
-
},
|
| 509 |
-
{
|
| 510 |
-
"epoch": 0.6,
|
| 511 |
-
"grad_norm": 1.1951791048049927,
|
| 512 |
-
"learning_rate": 0.00010923328122605982,
|
| 513 |
-
"loss": 0.2437,
|
| 514 |
-
"step": 36000
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 0.6083333333333333,
|
| 518 |
-
"grad_norm": 1.0552036762237549,
|
| 519 |
-
"learning_rate": 0.00010535254091589667,
|
| 520 |
-
"loss": 0.2426,
|
| 521 |
-
"step": 36500
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 0.6166666666666667,
|
| 525 |
-
"grad_norm": 1.1212034225463867,
|
| 526 |
-
"learning_rate": 0.00010150432162655218,
|
| 527 |
-
"loss": 0.2454,
|
| 528 |
-
"step": 37000
|
| 529 |
-
},
|
| 530 |
-
{
|
| 531 |
-
"epoch": 0.625,
|
| 532 |
-
"grad_norm": 0.7416337132453918,
|
| 533 |
-
"learning_rate": 9.769142638472033e-05,
|
| 534 |
-
"loss": 0.2401,
|
| 535 |
-
"step": 37500
|
| 536 |
-
},
|
| 537 |
-
{
|
| 538 |
-
"epoch": 0.6333333333333333,
|
| 539 |
-
"grad_norm": 0.6174165606498718,
|
| 540 |
-
"learning_rate": 9.391663248720767e-05,
|
| 541 |
-
"loss": 0.2462,
|
| 542 |
-
"step": 38000
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.6416666666666667,
|
| 546 |
-
"grad_norm": 1.7195667028427124,
|
| 547 |
-
"learning_rate": 9.01826894779624e-05,
|
| 548 |
-
"loss": 0.2404,
|
| 549 |
-
"step": 38500
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.65,
|
| 553 |
-
"grad_norm": 0.8892576098442078,
|
| 554 |
-
"learning_rate": 8.649231714531805e-05,
|
| 555 |
-
"loss": 0.237,
|
| 556 |
-
"step": 39000
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 0.6583333333333333,
|
| 560 |
-
"grad_norm": 0.5483108758926392,
|
| 561 |
-
"learning_rate": 8.284820354091123e-05,
|
| 562 |
-
"loss": 0.2458,
|
| 563 |
-
"step": 39500
|
| 564 |
-
},
|
| 565 |
-
{
|
| 566 |
-
"epoch": 0.6666666666666666,
|
| 567 |
-
"grad_norm": 0.9700618982315063,
|
| 568 |
-
"learning_rate": 7.925300302171602e-05,
|
| 569 |
-
"loss": 0.24,
|
| 570 |
-
"step": 40000
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"epoch": 0.675,
|
| 574 |
-
"grad_norm": 0.8393011689186096,
|
| 575 |
-
"learning_rate": 7.570933431662156e-05,
|
| 576 |
-
"loss": 0.2334,
|
| 577 |
-
"step": 40500
|
| 578 |
-
},
|
| 579 |
-
{
|
| 580 |
-
"epoch": 0.6833333333333333,
|
| 581 |
-
"grad_norm": 1.7267228364944458,
|
| 582 |
-
"learning_rate": 7.221977861896105e-05,
|
| 583 |
-
"loss": 0.2454,
|
| 584 |
-
"step": 41000
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.6916666666666667,
|
| 588 |
-
"grad_norm": 1.5230600833892822,
|
| 589 |
-
"learning_rate": 6.878687770638148e-05,
|
| 590 |
-
"loss": 0.2291,
|
| 591 |
-
"step": 41500
|
| 592 |
-
},
|
| 593 |
-
{
|
| 594 |
-
"epoch": 0.7,
|
| 595 |
-
"grad_norm": 0.8894439935684204,
|
| 596 |
-
"learning_rate": 6.541313208942333e-05,
|
| 597 |
-
"loss": 0.2304,
|
| 598 |
-
"step": 42000
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 0.7083333333333334,
|
| 602 |
-
"grad_norm": 0.6637595891952515,
|
| 603 |
-
"learning_rate": 6.210099919015931e-05,
|
| 604 |
-
"loss": 0.2256,
|
| 605 |
-
"step": 42500
|
| 606 |
-
},
|
| 607 |
-
{
|
| 608 |
-
"epoch": 0.7166666666666667,
|
| 609 |
-
"grad_norm": 1.5464119911193848,
|
| 610 |
-
"learning_rate": 5.8852891552218674e-05,
|
| 611 |
-
"loss": 0.2296,
|
| 612 |
-
"step": 43000
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 0.725,
|
| 616 |
-
"grad_norm": 1.361809492111206,
|
| 617 |
-
"learning_rate": 5.567117508350033e-05,
|
| 618 |
-
"loss": 0.225,
|
| 619 |
-
"step": 43500
|
| 620 |
-
},
|
| 621 |
-
{
|
| 622 |
-
"epoch": 0.7333333333333333,
|
| 623 |
-
"grad_norm": 1.100778341293335,
|
| 624 |
-
"learning_rate": 5.2558167332855614e-05,
|
| 625 |
-
"loss": 0.2171,
|
| 626 |
-
"step": 44000
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.7416666666666667,
|
| 630 |
-
"grad_norm": 0.8792382478713989,
|
| 631 |
-
"learning_rate": 4.951613580199569e-05,
|
| 632 |
-
"loss": 0.2253,
|
| 633 |
-
"step": 44500
|
| 634 |
-
},
|
| 635 |
-
{
|
| 636 |
-
"epoch": 0.75,
|
| 637 |
-
"grad_norm": 1.008772611618042,
|
| 638 |
-
"learning_rate": 4.65472962938525e-05,
|
| 639 |
-
"loss": 0.2185,
|
| 640 |
-
"step": 45000
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 0.7583333333333333,
|
| 644 |
-
"grad_norm": 1.769035816192627,
|
| 645 |
-
"learning_rate": 4.365381129859762e-05,
|
| 646 |
-
"loss": 0.2323,
|
| 647 |
-
"step": 45500
|
| 648 |
-
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 0.7666666666666667,
|
| 651 |
-
"grad_norm": 1.0002933740615845,
|
| 652 |
-
"learning_rate": 4.0837788418493234e-05,
|
| 653 |
-
"loss": 0.2258,
|
| 654 |
-
"step": 46000
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 0.775,
|
| 658 |
-
"grad_norm": 0.6770097613334656,
|
| 659 |
-
"learning_rate": 3.810127883272383e-05,
|
| 660 |
-
"loss": 0.2237,
|
| 661 |
-
"step": 46500
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"epoch": 0.7833333333333333,
|
| 665 |
-
"grad_norm": 1.1669566631317139,
|
| 666 |
-
"learning_rate": 3.544627580332564e-05,
|
| 667 |
-
"loss": 0.2106,
|
| 668 |
-
"step": 47000
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.7916666666666666,
|
| 672 |
-
"grad_norm": 0.6947309374809265,
|
| 673 |
-
"learning_rate": 3.2874713223303216e-05,
|
| 674 |
-
"loss": 0.2198,
|
| 675 |
-
"step": 47500
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 0.8,
|
| 679 |
-
"grad_norm": 0.6739930510520935,
|
| 680 |
-
"learning_rate": 3.038846420798978e-05,
|
| 681 |
-
"loss": 0.2183,
|
| 682 |
-
"step": 48000
|
| 683 |
-
},
|
| 684 |
-
{
|
| 685 |
-
"epoch": 0.8083333333333333,
|
| 686 |
-
"grad_norm": 1.489799976348877,
|
| 687 |
-
"learning_rate": 2.7989339730678078e-05,
|
| 688 |
-
"loss": 0.2189,
|
| 689 |
-
"step": 48500
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 0.8166666666666667,
|
| 693 |
-
"grad_norm": 0.8515140414237976,
|
| 694 |
-
"learning_rate": 2.5679087303514894e-05,
|
| 695 |
-
"loss": 0.2244,
|
| 696 |
-
"step": 49000
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 0.825,
|
| 700 |
-
"grad_norm": 0.49479931592941284,
|
| 701 |
-
"learning_rate": 2.345938970462068e-05,
|
| 702 |
-
"loss": 0.2369,
|
| 703 |
-
"step": 49500
|
| 704 |
-
},
|
| 705 |
-
{
|
| 706 |
-
"epoch": 0.8333333333333334,
|
| 707 |
-
"grad_norm": 0.8055633902549744,
|
| 708 |
-
"learning_rate": 2.1331863752360973e-05,
|
| 709 |
-
"loss": 0.2217,
|
| 710 |
-
"step": 50000
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 0.8416666666666667,
|
| 714 |
-
"grad_norm": 2.17622447013855,
|
| 715 |
-
"learning_rate": 1.9298059127662975e-05,
|
| 716 |
-
"loss": 0.2087,
|
| 717 |
-
"step": 50500
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.85,
|
| 721 |
-
"grad_norm": 1.8723386526107788,
|
| 722 |
-
"learning_rate": 1.7359457245234156e-05,
|
| 723 |
-
"loss": 0.223,
|
| 724 |
-
"step": 51000
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.8583333333333333,
|
| 728 |
-
"grad_norm": 1.4284367561340332,
|
| 729 |
-
"learning_rate": 1.5517470174506246e-05,
|
| 730 |
-
"loss": 0.2201,
|
| 731 |
-
"step": 51500
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.8666666666666667,
|
| 735 |
-
"grad_norm": 1.5737581253051758,
|
| 736 |
-
"learning_rate": 1.3773439611089832e-05,
|
| 737 |
-
"loss": 0.224,
|
| 738 |
-
"step": 52000
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 0.875,
|
| 742 |
-
"grad_norm": 1.9173991680145264,
|
| 743 |
-
"learning_rate": 1.212863589948883e-05,
|
| 744 |
-
"loss": 0.2316,
|
| 745 |
-
"step": 52500
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 0.8833333333333333,
|
| 749 |
-
"grad_norm": 0.9126625657081604,
|
| 750 |
-
"learning_rate": 1.058425710778692e-05,
|
| 751 |
-
"loss": 0.2099,
|
| 752 |
-
"step": 53000
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.8916666666666667,
|
| 756 |
-
"grad_norm": 0.9519773721694946,
|
| 757 |
-
"learning_rate": 9.141428154979886e-06,
|
| 758 |
-
"loss": 0.2231,
|
| 759 |
-
"step": 53500
|
| 760 |
-
},
|
| 761 |
-
{
|
| 762 |
-
"epoch": 0.9,
|
| 763 |
-
"grad_norm": 1.7559552192687988,
|
| 764 |
-
"learning_rate": 7.801199991589335e-06,
|
| 765 |
-
"loss": 0.2203,
|
| 766 |
-
"step": 54000
|
| 767 |
-
},
|
| 768 |
-
{
|
| 769 |
-
"epoch": 0.9083333333333333,
|
| 770 |
-
"grad_norm": 1.3599870204925537,
|
| 771 |
-
"learning_rate": 6.564548834154759e-06,
|
| 772 |
-
"loss": 0.2224,
|
| 773 |
-
"step": 54500
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 0.9166666666666666,
|
| 777 |
-
"grad_norm": 1.5323917865753174,
|
| 778 |
-
"learning_rate": 5.432375454161458e-06,
|
| 779 |
-
"loss": 0.2131,
|
| 780 |
-
"step": 55000
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 0.925,
|
| 784 |
-
"grad_norm": 0.8116886615753174,
|
| 785 |
-
"learning_rate": 4.405504521922571e-06,
|
| 786 |
-
"loss": 0.2122,
|
| 787 |
-
"step": 55500
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"epoch": 0.9333333333333333,
|
| 791 |
-
"grad_norm": 0.9112006425857544,
|
| 792 |
-
"learning_rate": 3.484684005892502e-06,
|
| 793 |
-
"loss": 0.2184,
|
| 794 |
-
"step": 56000
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.9416666666666667,
|
| 798 |
-
"grad_norm": 0.6394028663635254,
|
| 799 |
-
"learning_rate": 2.6705846278499532e-06,
|
| 800 |
-
"loss": 0.238,
|
| 801 |
-
"step": 56500
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"epoch": 0.95,
|
| 805 |
-
"grad_norm": 0.9545285701751709,
|
| 806 |
-
"learning_rate": 1.9637993743470525e-06,
|
| 807 |
-
"loss": 0.226,
|
| 808 |
-
"step": 57000
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.9583333333333334,
|
| 812 |
-
"grad_norm": 0.7243571877479553,
|
| 813 |
-
"learning_rate": 1.36484306478033e-06,
|
| 814 |
-
"loss": 0.2236,
|
| 815 |
-
"step": 57500
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"epoch": 0.9666666666666667,
|
| 819 |
-
"grad_norm": 0.8191093802452087,
|
| 820 |
-
"learning_rate": 8.741519763985627e-07,
|
| 821 |
-
"loss": 0.217,
|
| 822 |
-
"step": 58000
|
| 823 |
-
},
|
| 824 |
-
{
|
| 825 |
-
"epoch": 0.975,
|
| 826 |
-
"grad_norm": 1.4871535301208496,
|
| 827 |
-
"learning_rate": 4.920835265201595e-07,
|
| 828 |
-
"loss": 0.2176,
|
| 829 |
-
"step": 58500
|
| 830 |
-
},
|
| 831 |
-
{
|
| 832 |
-
"epoch": 0.9833333333333333,
|
| 833 |
-
"grad_norm": 0.9401077628135681,
|
| 834 |
-
"learning_rate": 2.1891601219199417e-07,
|
| 835 |
-
"loss": 0.2267,
|
| 836 |
-
"step": 59000
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.9916666666666667,
|
| 840 |
-
"grad_norm": 1.3500021696090698,
|
| 841 |
-
"learning_rate": 5.484840747892461e-08,
|
| 842 |
-
"loss": 0.2189,
|
| 843 |
-
"step": 59500
|
| 844 |
-
},
|
| 845 |
{
|
| 846 |
"epoch": 1.0,
|
| 847 |
-
"grad_norm":
|
| 848 |
-
"learning_rate":
|
| 849 |
-
"loss": 0.
|
| 850 |
-
"step":
|
| 851 |
},
|
| 852 |
{
|
| 853 |
"epoch": 1.0,
|
| 854 |
-
"step":
|
| 855 |
-
"total_flos":
|
| 856 |
-
"train_loss": 0.
|
| 857 |
-
"train_runtime":
|
| 858 |
-
"train_samples_per_second": 5.
|
| 859 |
-
"train_steps_per_second": 5.
|
| 860 |
}
|
| 861 |
],
|
| 862 |
"logging_steps": 500,
|
| 863 |
-
"max_steps":
|
| 864 |
"num_input_tokens_seen": 0,
|
| 865 |
"num_train_epochs": 1,
|
| 866 |
-
"save_steps":
|
| 867 |
"stateful_callbacks": {
|
| 868 |
"TrainerControl": {
|
| 869 |
"args": {
|
|
@@ -876,7 +386,7 @@
|
|
| 876 |
"attributes": {}
|
| 877 |
}
|
| 878 |
},
|
| 879 |
-
"total_flos":
|
| 880 |
"train_batch_size": 1,
|
| 881 |
"trial_name": null,
|
| 882 |
"trial_params": null
|
|
|
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
"epoch": 1.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 25000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"epoch": 0.02,
|
| 14 |
+
"grad_norm": 1.1523245573043823,
|
| 15 |
+
"learning_rate": 0.00019959999999999997,
|
| 16 |
+
"loss": 0.4793,
|
| 17 |
"step": 500
|
| 18 |
},
|
| 19 |
{
|
| 20 |
+
"epoch": 0.04,
|
| 21 |
+
"grad_norm": 1.207442283630371,
|
| 22 |
+
"learning_rate": 0.0002999219633608753,
|
| 23 |
+
"loss": 0.365,
|
| 24 |
"step": 1000
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.06,
|
| 28 |
+
"grad_norm": 1.608832836151123,
|
| 29 |
+
"learning_rate": 0.0002992943970692375,
|
| 30 |
+
"loss": 0.3408,
|
| 31 |
"step": 1500
|
| 32 |
},
|
| 33 |
{
|
| 34 |
+
"epoch": 0.08,
|
| 35 |
+
"grad_norm": 1.6748807430267334,
|
| 36 |
+
"learning_rate": 0.0002980406381928192,
|
| 37 |
+
"loss": 0.3432,
|
| 38 |
"step": 2000
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.1,
|
| 42 |
+
"grad_norm": 1.1549451351165771,
|
| 43 |
+
"learning_rate": 0.0002961659454320601,
|
| 44 |
+
"loss": 0.3337,
|
| 45 |
"step": 2500
|
| 46 |
},
|
| 47 |
{
|
| 48 |
+
"epoch": 0.12,
|
| 49 |
+
"grad_norm": 1.3943244218826294,
|
| 50 |
+
"learning_rate": 0.0002936781818999006,
|
| 51 |
+
"loss": 0.328,
|
| 52 |
"step": 3000
|
| 53 |
},
|
| 54 |
{
|
| 55 |
+
"epoch": 0.14,
|
| 56 |
+
"grad_norm": 0.8221294283866882,
|
| 57 |
+
"learning_rate": 0.0002905877821411536,
|
| 58 |
+
"loss": 0.3184,
|
| 59 |
"step": 3500
|
| 60 |
},
|
| 61 |
{
|
| 62 |
+
"epoch": 0.16,
|
| 63 |
+
"grad_norm": 0.8499279022216797,
|
| 64 |
+
"learning_rate": 0.00028690770836639715,
|
| 65 |
+
"loss": 0.312,
|
| 66 |
"step": 4000
|
| 67 |
},
|
| 68 |
{
|
| 69 |
+
"epoch": 0.18,
|
| 70 |
+
"grad_norm": 1.1005589962005615,
|
| 71 |
+
"learning_rate": 0.0002826533960839586,
|
| 72 |
+
"loss": 0.3175,
|
| 73 |
"step": 4500
|
| 74 |
},
|
| 75 |
{
|
| 76 |
+
"epoch": 0.2,
|
| 77 |
+
"grad_norm": 1.239676594734192,
|
| 78 |
+
"learning_rate": 0.0002778426893580286,
|
| 79 |
+
"loss": 0.3139,
|
| 80 |
"step": 5000
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 0.22,
|
| 84 |
+
"grad_norm": 0.5692813396453857,
|
| 85 |
+
"learning_rate": 0.00027249576596445455,
|
| 86 |
+
"loss": 0.3065,
|
| 87 |
"step": 5500
|
| 88 |
},
|
| 89 |
{
|
| 90 |
+
"epoch": 0.24,
|
| 91 |
+
"grad_norm": 0.6223649382591248,
|
| 92 |
+
"learning_rate": 0.00026663505275813633,
|
| 93 |
+
"loss": 0.3015,
|
| 94 |
"step": 6000
|
| 95 |
},
|
| 96 |
{
|
| 97 |
+
"epoch": 0.26,
|
| 98 |
+
"grad_norm": 1.6326464414596558,
|
| 99 |
+
"learning_rate": 0.0002602851316070032,
|
| 100 |
+
"loss": 0.298,
|
| 101 |
"step": 6500
|
| 102 |
},
|
| 103 |
{
|
| 104 |
+
"epoch": 0.28,
|
| 105 |
+
"grad_norm": 0.44571417570114136,
|
| 106 |
+
"learning_rate": 0.0002534726362871166,
|
| 107 |
+
"loss": 0.3043,
|
| 108 |
"step": 7000
|
| 109 |
},
|
| 110 |
{
|
| 111 |
+
"epoch": 0.3,
|
| 112 |
+
"grad_norm": 0.7266321182250977,
|
| 113 |
+
"learning_rate": 0.00024622614077135773,
|
| 114 |
+
"loss": 0.2949,
|
| 115 |
"step": 7500
|
| 116 |
},
|
| 117 |
{
|
| 118 |
+
"epoch": 0.32,
|
| 119 |
+
"grad_norm": 1.133889079093933,
|
| 120 |
+
"learning_rate": 0.00023857603938025494,
|
| 121 |
+
"loss": 0.2826,
|
| 122 |
"step": 8000
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 0.34,
|
| 126 |
+
"grad_norm": 0.7410082817077637,
|
| 127 |
+
"learning_rate": 0.00023055441929764077,
|
| 128 |
+
"loss": 0.2836,
|
| 129 |
"step": 8500
|
| 130 |
},
|
| 131 |
{
|
| 132 |
+
"epoch": 0.36,
|
| 133 |
+
"grad_norm": 0.5899648070335388,
|
| 134 |
+
"learning_rate": 0.00022219492598585185,
|
| 135 |
+
"loss": 0.2868,
|
| 136 |
"step": 9000
|
| 137 |
},
|
| 138 |
{
|
| 139 |
+
"epoch": 0.38,
|
| 140 |
+
"grad_norm": 1.3609607219696045,
|
| 141 |
+
"learning_rate": 0.00021353262206496714,
|
| 142 |
+
"loss": 0.3002,
|
| 143 |
"step": 9500
|
| 144 |
},
|
| 145 |
{
|
| 146 |
+
"epoch": 0.4,
|
| 147 |
+
"grad_norm": 1.7143217325210571,
|
| 148 |
+
"learning_rate": 0.0002046038402479944,
|
| 149 |
+
"loss": 0.2734,
|
| 150 |
"step": 10000
|
| 151 |
},
|
| 152 |
{
|
| 153 |
+
"epoch": 0.42,
|
| 154 |
+
"grad_norm": 0.9960472583770752,
|
| 155 |
+
"learning_rate": 0.0001954460309488451,
|
| 156 |
+
"loss": 0.2824,
|
| 157 |
"step": 10500
|
| 158 |
},
|
| 159 |
{
|
| 160 |
+
"epoch": 0.44,
|
| 161 |
+
"grad_norm": 0.9902260899543762,
|
| 162 |
+
"learning_rate": 0.00018609760520228364,
|
| 163 |
+
"loss": 0.2802,
|
| 164 |
"step": 11000
|
| 165 |
},
|
| 166 |
{
|
| 167 |
+
"epoch": 0.46,
|
| 168 |
+
"grad_norm": 1.1902357339859009,
|
| 169 |
+
"learning_rate": 0.00017659777355469886,
|
| 170 |
+
"loss": 0.2848,
|
| 171 |
"step": 11500
|
| 172 |
},
|
| 173 |
{
|
| 174 |
+
"epoch": 0.48,
|
| 175 |
+
"grad_norm": 0.7066996097564697,
|
| 176 |
+
"learning_rate": 0.0001669863816014457,
|
| 177 |
+
"loss": 0.2723,
|
| 178 |
"step": 12000
|
| 179 |
},
|
| 180 |
{
|
| 181 |
+
"epoch": 0.5,
|
| 182 |
+
"grad_norm": 0.8586142063140869,
|
| 183 |
+
"learning_rate": 0.00015730374286057098,
|
| 184 |
+
"loss": 0.2661,
|
| 185 |
"step": 12500
|
| 186 |
},
|
| 187 |
{
|
| 188 |
+
"epoch": 0.52,
|
| 189 |
+
"grad_norm": 1.3139828443527222,
|
| 190 |
+
"learning_rate": 0.00014759046968390891,
|
| 191 |
+
"loss": 0.27,
|
| 192 |
"step": 13000
|
| 193 |
},
|
| 194 |
{
|
| 195 |
+
"epoch": 0.54,
|
| 196 |
+
"grad_norm": 2.092033624649048,
|
| 197 |
+
"learning_rate": 0.00013788730291476348,
|
| 198 |
+
"loss": 0.2626,
|
| 199 |
"step": 13500
|
| 200 |
},
|
| 201 |
{
|
| 202 |
+
"epoch": 0.56,
|
| 203 |
+
"grad_norm": 0.6759353876113892,
|
| 204 |
+
"learning_rate": 0.00012823494100665345,
|
| 205 |
+
"loss": 0.2713,
|
| 206 |
"step": 14000
|
| 207 |
},
|
| 208 |
{
|
| 209 |
+
"epoch": 0.58,
|
| 210 |
+
"grad_norm": 0.7745094299316406,
|
| 211 |
+
"learning_rate": 0.0001186738693198545,
|
| 212 |
+
"loss": 0.2659,
|
| 213 |
"step": 14500
|
| 214 |
},
|
| 215 |
{
|
| 216 |
+
"epoch": 0.6,
|
| 217 |
+
"grad_norm": 1.1789864301681519,
|
| 218 |
+
"learning_rate": 0.00010924419031172836,
|
| 219 |
+
"loss": 0.2621,
|
| 220 |
"step": 15000
|
| 221 |
},
|
| 222 |
{
|
| 223 |
+
"epoch": 0.62,
|
| 224 |
+
"grad_norm": 0.7983564138412476,
|
| 225 |
+
"learning_rate": 9.998545533308028e-05,
|
| 226 |
+
"loss": 0.2485,
|
| 227 |
"step": 15500
|
| 228 |
},
|
| 229 |
{
|
| 230 |
+
"epoch": 0.64,
|
| 231 |
+
"grad_norm": 2.2566275596618652,
|
| 232 |
+
"learning_rate": 9.093649873604878e-05,
|
| 233 |
+
"loss": 0.2499,
|
| 234 |
"step": 16000
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.66,
|
| 238 |
+
"grad_norm": 2.405757188796997,
|
| 239 |
+
"learning_rate": 8.213527498933654e-05,
|
| 240 |
+
"loss": 0.2648,
|
| 241 |
"step": 16500
|
| 242 |
},
|
| 243 |
{
|
| 244 |
+
"epoch": 0.68,
|
| 245 |
+
"grad_norm": 0.706387996673584,
|
| 246 |
+
"learning_rate": 7.361869948397888e-05,
|
| 247 |
+
"loss": 0.248,
|
| 248 |
"step": 17000
|
| 249 |
},
|
| 250 |
{
|
| 251 |
+
"epoch": 0.7,
|
| 252 |
+
"grad_norm": 0.9889776110649109,
|
| 253 |
+
"learning_rate": 6.542249369736452e-05,
|
| 254 |
+
"loss": 0.2533,
|
| 255 |
"step": 17500
|
| 256 |
},
|
| 257 |
{
|
| 258 |
+
"epoch": 0.72,
|
| 259 |
+
"grad_norm": 0.3792021572589874,
|
| 260 |
+
"learning_rate": 5.758103536494548e-05,
|
| 261 |
+
"loss": 0.2464,
|
| 262 |
"step": 18000
|
| 263 |
},
|
| 264 |
{
|
| 265 |
+
"epoch": 0.74,
|
| 266 |
+
"grad_norm": 1.4570865631103516,
|
| 267 |
+
"learning_rate": 5.012721428806742e-05,
|
| 268 |
+
"loss": 0.2469,
|
| 269 |
"step": 18500
|
| 270 |
},
|
| 271 |
{
|
| 272 |
+
"epoch": 0.76,
|
| 273 |
+
"grad_norm": 0.3031163811683655,
|
| 274 |
+
"learning_rate": 4.30922943827128e-05,
|
| 275 |
+
"loss": 0.2563,
|
| 276 |
"step": 19000
|
| 277 |
},
|
| 278 |
{
|
| 279 |
+
"epoch": 0.78,
|
| 280 |
+
"grad_norm": 0.5329792499542236,
|
| 281 |
+
"learning_rate": 3.6505782547772205e-05,
|
| 282 |
+
"loss": 0.2409,
|
| 283 |
"step": 19500
|
| 284 |
},
|
| 285 |
{
|
| 286 |
+
"epoch": 0.8,
|
| 287 |
+
"grad_norm": 1.900152325630188,
|
| 288 |
+
"learning_rate": 3.0395304902856126e-05,
|
| 289 |
+
"loss": 0.2413,
|
| 290 |
"step": 20000
|
| 291 |
},
|
| 292 |
{
|
| 293 |
+
"epoch": 0.82,
|
| 294 |
+
"grad_norm": 0.79344642162323,
|
| 295 |
+
"learning_rate": 2.4786490914748508e-05,
|
| 296 |
+
"loss": 0.241,
|
| 297 |
"step": 20500
|
| 298 |
},
|
| 299 |
{
|
| 300 |
+
"epoch": 0.84,
|
| 301 |
+
"grad_norm": 2.6573524475097656,
|
| 302 |
+
"learning_rate": 1.970286589851618e-05,
|
| 303 |
+
"loss": 0.2447,
|
| 304 |
"step": 21000
|
| 305 |
},
|
| 306 |
{
|
| 307 |
+
"epoch": 0.86,
|
| 308 |
+
"grad_norm": 0.92716383934021,
|
| 309 |
+
"learning_rate": 1.516575234416183e-05,
|
| 310 |
+
"loss": 0.2513,
|
| 311 |
"step": 21500
|
| 312 |
},
|
| 313 |
{
|
| 314 |
+
"epoch": 0.88,
|
| 315 |
+
"grad_norm": 0.8475908041000366,
|
| 316 |
+
"learning_rate": 1.1194180482690935e-05,
|
| 317 |
+
"loss": 0.2425,
|
| 318 |
"step": 22000
|
| 319 |
},
|
| 320 |
{
|
| 321 |
+
"epoch": 0.9,
|
| 322 |
+
"grad_norm": 0.6743385195732117,
|
| 323 |
+
"learning_rate": 7.804808466709245e-06,
|
| 324 |
+
"loss": 0.2372,
|
| 325 |
"step": 22500
|
| 326 |
},
|
| 327 |
{
|
| 328 |
+
"epoch": 0.92,
|
| 329 |
+
"grad_norm": 0.7676027417182922,
|
| 330 |
+
"learning_rate": 5.011852500341956e-06,
|
| 331 |
+
"loss": 0.2429,
|
| 332 |
"step": 23000
|
| 333 |
},
|
| 334 |
{
|
| 335 |
+
"epoch": 0.94,
|
| 336 |
+
"grad_norm": 2.600247621536255,
|
| 337 |
+
"learning_rate": 2.827027211532812e-06,
|
| 338 |
+
"loss": 0.2351,
|
| 339 |
"step": 23500
|
| 340 |
},
|
| 341 |
{
|
| 342 |
+
"epoch": 0.96,
|
| 343 |
+
"grad_norm": 0.2377513349056244,
|
| 344 |
+
"learning_rate": 1.2594965168235604e-06,
|
| 345 |
+
"loss": 0.2464,
|
| 346 |
"step": 24000
|
| 347 |
},
|
| 348 |
{
|
| 349 |
+
"epoch": 0.98,
|
| 350 |
+
"grad_norm": 3.111140012741089,
|
| 351 |
+
"learning_rate": 3.1583518470324476e-07,
|
| 352 |
+
"loss": 0.2387,
|
| 353 |
"step": 24500
|
| 354 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 355 |
{
|
| 356 |
"epoch": 1.0,
|
| 357 |
+
"grad_norm": 1.134462833404541,
|
| 358 |
+
"learning_rate": 1.2587443087852533e-12,
|
| 359 |
+
"loss": 0.2547,
|
| 360 |
+
"step": 25000
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 1.0,
|
| 364 |
+
"step": 25000,
|
| 365 |
+
"total_flos": 2.8066019659471258e+17,
|
| 366 |
+
"train_loss": 0.2806466256713867,
|
| 367 |
+
"train_runtime": 4375.9007,
|
| 368 |
+
"train_samples_per_second": 5.713,
|
| 369 |
+
"train_steps_per_second": 5.713
|
| 370 |
}
|
| 371 |
],
|
| 372 |
"logging_steps": 500,
|
| 373 |
+
"max_steps": 25000,
|
| 374 |
"num_input_tokens_seen": 0,
|
| 375 |
"num_train_epochs": 1,
|
| 376 |
+
"save_steps": 12500,
|
| 377 |
"stateful_callbacks": {
|
| 378 |
"TrainerControl": {
|
| 379 |
"args": {
|
|
|
|
| 386 |
"attributes": {}
|
| 387 |
}
|
| 388 |
},
|
| 389 |
+
"total_flos": 2.8066019659471258e+17,
|
| 390 |
"train_batch_size": 1,
|
| 391 |
"trial_name": null,
|
| 392 |
"trial_params": null
|