Delete trainer_state.json
Browse files- trainer_state.json +0 -532
trainer_state.json
DELETED
|
@@ -1,532 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": 0.440588116645813,
|
| 3 |
-
"best_model_checkpoint": "./results/checkpoint-1500",
|
| 4 |
-
"epoch": 1.9920318725099602,
|
| 5 |
-
"global_step": 3000,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.03,
|
| 12 |
-
"eval_loss": 0.4968358278274536,
|
| 13 |
-
"eval_runtime": 22.8103,
|
| 14 |
-
"eval_samples_per_second": 29.373,
|
| 15 |
-
"eval_steps_per_second": 29.373,
|
| 16 |
-
"step": 50
|
| 17 |
-
},
|
| 18 |
-
{
|
| 19 |
-
"epoch": 0.07,
|
| 20 |
-
"eval_loss": 0.4962530732154846,
|
| 21 |
-
"eval_runtime": 22.8045,
|
| 22 |
-
"eval_samples_per_second": 29.38,
|
| 23 |
-
"eval_steps_per_second": 29.38,
|
| 24 |
-
"step": 100
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.1,
|
| 28 |
-
"eval_loss": 0.5020118951797485,
|
| 29 |
-
"eval_runtime": 22.8025,
|
| 30 |
-
"eval_samples_per_second": 29.383,
|
| 31 |
-
"eval_steps_per_second": 29.383,
|
| 32 |
-
"step": 150
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.13,
|
| 36 |
-
"eval_loss": 0.4915091395378113,
|
| 37 |
-
"eval_runtime": 22.7983,
|
| 38 |
-
"eval_samples_per_second": 29.388,
|
| 39 |
-
"eval_steps_per_second": 29.388,
|
| 40 |
-
"step": 200
|
| 41 |
-
},
|
| 42 |
-
{
|
| 43 |
-
"epoch": 0.17,
|
| 44 |
-
"eval_loss": 0.48744314908981323,
|
| 45 |
-
"eval_runtime": 22.7937,
|
| 46 |
-
"eval_samples_per_second": 29.394,
|
| 47 |
-
"eval_steps_per_second": 29.394,
|
| 48 |
-
"step": 250
|
| 49 |
-
},
|
| 50 |
-
{
|
| 51 |
-
"epoch": 0.2,
|
| 52 |
-
"eval_loss": 0.48627975583076477,
|
| 53 |
-
"eval_runtime": 22.8,
|
| 54 |
-
"eval_samples_per_second": 29.386,
|
| 55 |
-
"eval_steps_per_second": 29.386,
|
| 56 |
-
"step": 300
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.23,
|
| 60 |
-
"eval_loss": 0.47990691661834717,
|
| 61 |
-
"eval_runtime": 22.793,
|
| 62 |
-
"eval_samples_per_second": 29.395,
|
| 63 |
-
"eval_steps_per_second": 29.395,
|
| 64 |
-
"step": 350
|
| 65 |
-
},
|
| 66 |
-
{
|
| 67 |
-
"epoch": 0.27,
|
| 68 |
-
"eval_loss": 0.47654837369918823,
|
| 69 |
-
"eval_runtime": 22.799,
|
| 70 |
-
"eval_samples_per_second": 29.387,
|
| 71 |
-
"eval_steps_per_second": 29.387,
|
| 72 |
-
"step": 400
|
| 73 |
-
},
|
| 74 |
-
{
|
| 75 |
-
"epoch": 0.3,
|
| 76 |
-
"eval_loss": 0.475556343793869,
|
| 77 |
-
"eval_runtime": 22.8029,
|
| 78 |
-
"eval_samples_per_second": 29.382,
|
| 79 |
-
"eval_steps_per_second": 29.382,
|
| 80 |
-
"step": 450
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.33,
|
| 84 |
-
"learning_rate": 4.732839838492598e-05,
|
| 85 |
-
"loss": 0.5362,
|
| 86 |
-
"step": 500
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.33,
|
| 90 |
-
"eval_loss": 0.47196289896965027,
|
| 91 |
-
"eval_runtime": 22.7979,
|
| 92 |
-
"eval_samples_per_second": 29.389,
|
| 93 |
-
"eval_steps_per_second": 29.389,
|
| 94 |
-
"step": 500
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.37,
|
| 98 |
-
"eval_loss": 0.470073401927948,
|
| 99 |
-
"eval_runtime": 22.8065,
|
| 100 |
-
"eval_samples_per_second": 29.378,
|
| 101 |
-
"eval_steps_per_second": 29.378,
|
| 102 |
-
"step": 550
|
| 103 |
-
},
|
| 104 |
-
{
|
| 105 |
-
"epoch": 0.4,
|
| 106 |
-
"eval_loss": 0.4680205285549164,
|
| 107 |
-
"eval_runtime": 22.7971,
|
| 108 |
-
"eval_samples_per_second": 29.39,
|
| 109 |
-
"eval_steps_per_second": 29.39,
|
| 110 |
-
"step": 600
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.43,
|
| 114 |
-
"eval_loss": 0.46594592928886414,
|
| 115 |
-
"eval_runtime": 22.7927,
|
| 116 |
-
"eval_samples_per_second": 29.395,
|
| 117 |
-
"eval_steps_per_second": 29.395,
|
| 118 |
-
"step": 650
|
| 119 |
-
},
|
| 120 |
-
{
|
| 121 |
-
"epoch": 0.46,
|
| 122 |
-
"eval_loss": 0.463152676820755,
|
| 123 |
-
"eval_runtime": 22.7891,
|
| 124 |
-
"eval_samples_per_second": 29.4,
|
| 125 |
-
"eval_steps_per_second": 29.4,
|
| 126 |
-
"step": 700
|
| 127 |
-
},
|
| 128 |
-
{
|
| 129 |
-
"epoch": 0.5,
|
| 130 |
-
"eval_loss": 0.4629766345024109,
|
| 131 |
-
"eval_runtime": 22.7924,
|
| 132 |
-
"eval_samples_per_second": 29.396,
|
| 133 |
-
"eval_steps_per_second": 29.396,
|
| 134 |
-
"step": 750
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 0.53,
|
| 138 |
-
"eval_loss": 0.45898592472076416,
|
| 139 |
-
"eval_runtime": 22.7951,
|
| 140 |
-
"eval_samples_per_second": 29.392,
|
| 141 |
-
"eval_steps_per_second": 29.392,
|
| 142 |
-
"step": 800
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.56,
|
| 146 |
-
"eval_loss": 0.4574301540851593,
|
| 147 |
-
"eval_runtime": 22.7963,
|
| 148 |
-
"eval_samples_per_second": 29.391,
|
| 149 |
-
"eval_steps_per_second": 29.391,
|
| 150 |
-
"step": 850
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.6,
|
| 154 |
-
"eval_loss": 0.4551514983177185,
|
| 155 |
-
"eval_runtime": 22.7965,
|
| 156 |
-
"eval_samples_per_second": 29.39,
|
| 157 |
-
"eval_steps_per_second": 29.39,
|
| 158 |
-
"step": 900
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.63,
|
| 162 |
-
"eval_loss": 0.4532181918621063,
|
| 163 |
-
"eval_runtime": 22.7913,
|
| 164 |
-
"eval_samples_per_second": 29.397,
|
| 165 |
-
"eval_steps_per_second": 29.397,
|
| 166 |
-
"step": 950
|
| 167 |
-
},
|
| 168 |
-
{
|
| 169 |
-
"epoch": 0.66,
|
| 170 |
-
"learning_rate": 4.396366083445492e-05,
|
| 171 |
-
"loss": 0.4595,
|
| 172 |
-
"step": 1000
|
| 173 |
-
},
|
| 174 |
-
{
|
| 175 |
-
"epoch": 0.66,
|
| 176 |
-
"eval_loss": 0.4534400999546051,
|
| 177 |
-
"eval_runtime": 22.7971,
|
| 178 |
-
"eval_samples_per_second": 29.39,
|
| 179 |
-
"eval_steps_per_second": 29.39,
|
| 180 |
-
"step": 1000
|
| 181 |
-
},
|
| 182 |
-
{
|
| 183 |
-
"epoch": 0.7,
|
| 184 |
-
"eval_loss": 0.4516228139400482,
|
| 185 |
-
"eval_runtime": 22.8066,
|
| 186 |
-
"eval_samples_per_second": 29.378,
|
| 187 |
-
"eval_steps_per_second": 29.378,
|
| 188 |
-
"step": 1050
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 0.73,
|
| 192 |
-
"eval_loss": 0.4516217112541199,
|
| 193 |
-
"eval_runtime": 22.7934,
|
| 194 |
-
"eval_samples_per_second": 29.394,
|
| 195 |
-
"eval_steps_per_second": 29.394,
|
| 196 |
-
"step": 1100
|
| 197 |
-
},
|
| 198 |
-
{
|
| 199 |
-
"epoch": 0.76,
|
| 200 |
-
"eval_loss": 0.4490440785884857,
|
| 201 |
-
"eval_runtime": 22.7982,
|
| 202 |
-
"eval_samples_per_second": 29.388,
|
| 203 |
-
"eval_steps_per_second": 29.388,
|
| 204 |
-
"step": 1150
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"epoch": 0.8,
|
| 208 |
-
"eval_loss": 0.44825297594070435,
|
| 209 |
-
"eval_runtime": 22.8047,
|
| 210 |
-
"eval_samples_per_second": 29.38,
|
| 211 |
-
"eval_steps_per_second": 29.38,
|
| 212 |
-
"step": 1200
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.83,
|
| 216 |
-
"eval_loss": 0.44741538166999817,
|
| 217 |
-
"eval_runtime": 22.8005,
|
| 218 |
-
"eval_samples_per_second": 29.385,
|
| 219 |
-
"eval_steps_per_second": 29.385,
|
| 220 |
-
"step": 1250
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.86,
|
| 224 |
-
"eval_loss": 0.44648081064224243,
|
| 225 |
-
"eval_runtime": 22.7982,
|
| 226 |
-
"eval_samples_per_second": 29.388,
|
| 227 |
-
"eval_steps_per_second": 29.388,
|
| 228 |
-
"step": 1300
|
| 229 |
-
},
|
| 230 |
-
{
|
| 231 |
-
"epoch": 0.9,
|
| 232 |
-
"eval_loss": 0.44202762842178345,
|
| 233 |
-
"eval_runtime": 22.7988,
|
| 234 |
-
"eval_samples_per_second": 29.388,
|
| 235 |
-
"eval_steps_per_second": 29.388,
|
| 236 |
-
"step": 1350
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.93,
|
| 240 |
-
"eval_loss": 0.442158043384552,
|
| 241 |
-
"eval_runtime": 22.7966,
|
| 242 |
-
"eval_samples_per_second": 29.39,
|
| 243 |
-
"eval_steps_per_second": 29.39,
|
| 244 |
-
"step": 1400
|
| 245 |
-
},
|
| 246 |
-
{
|
| 247 |
-
"epoch": 0.96,
|
| 248 |
-
"eval_loss": 0.44166651368141174,
|
| 249 |
-
"eval_runtime": 22.7983,
|
| 250 |
-
"eval_samples_per_second": 29.388,
|
| 251 |
-
"eval_steps_per_second": 29.388,
|
| 252 |
-
"step": 1450
|
| 253 |
-
},
|
| 254 |
-
{
|
| 255 |
-
"epoch": 1.0,
|
| 256 |
-
"learning_rate": 4.0598923283983853e-05,
|
| 257 |
-
"loss": 0.4383,
|
| 258 |
-
"step": 1500
|
| 259 |
-
},
|
| 260 |
-
{
|
| 261 |
-
"epoch": 1.0,
|
| 262 |
-
"eval_loss": 0.440588116645813,
|
| 263 |
-
"eval_runtime": 22.7995,
|
| 264 |
-
"eval_samples_per_second": 29.387,
|
| 265 |
-
"eval_steps_per_second": 29.387,
|
| 266 |
-
"step": 1500
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 1.03,
|
| 270 |
-
"eval_loss": 0.45270833373069763,
|
| 271 |
-
"eval_runtime": 22.8208,
|
| 272 |
-
"eval_samples_per_second": 29.359,
|
| 273 |
-
"eval_steps_per_second": 29.359,
|
| 274 |
-
"step": 1550
|
| 275 |
-
},
|
| 276 |
-
{
|
| 277 |
-
"epoch": 1.06,
|
| 278 |
-
"eval_loss": 0.45460081100463867,
|
| 279 |
-
"eval_runtime": 22.7988,
|
| 280 |
-
"eval_samples_per_second": 29.388,
|
| 281 |
-
"eval_steps_per_second": 29.388,
|
| 282 |
-
"step": 1600
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 1.1,
|
| 286 |
-
"eval_loss": 0.4546465575695038,
|
| 287 |
-
"eval_runtime": 22.8024,
|
| 288 |
-
"eval_samples_per_second": 29.383,
|
| 289 |
-
"eval_steps_per_second": 29.383,
|
| 290 |
-
"step": 1650
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 1.13,
|
| 294 |
-
"eval_loss": 0.45767539739608765,
|
| 295 |
-
"eval_runtime": 22.8006,
|
| 296 |
-
"eval_samples_per_second": 29.385,
|
| 297 |
-
"eval_steps_per_second": 29.385,
|
| 298 |
-
"step": 1700
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"epoch": 1.16,
|
| 302 |
-
"eval_loss": 0.4575343728065491,
|
| 303 |
-
"eval_runtime": 22.8053,
|
| 304 |
-
"eval_samples_per_second": 29.379,
|
| 305 |
-
"eval_steps_per_second": 29.379,
|
| 306 |
-
"step": 1750
|
| 307 |
-
},
|
| 308 |
-
{
|
| 309 |
-
"epoch": 1.2,
|
| 310 |
-
"eval_loss": 0.4562443494796753,
|
| 311 |
-
"eval_runtime": 22.7983,
|
| 312 |
-
"eval_samples_per_second": 29.388,
|
| 313 |
-
"eval_steps_per_second": 29.388,
|
| 314 |
-
"step": 1800
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 1.23,
|
| 318 |
-
"eval_loss": 0.4556874632835388,
|
| 319 |
-
"eval_runtime": 22.7968,
|
| 320 |
-
"eval_samples_per_second": 29.39,
|
| 321 |
-
"eval_steps_per_second": 29.39,
|
| 322 |
-
"step": 1850
|
| 323 |
-
},
|
| 324 |
-
{
|
| 325 |
-
"epoch": 1.26,
|
| 326 |
-
"eval_loss": 0.4551508128643036,
|
| 327 |
-
"eval_runtime": 22.8001,
|
| 328 |
-
"eval_samples_per_second": 29.386,
|
| 329 |
-
"eval_steps_per_second": 29.386,
|
| 330 |
-
"step": 1900
|
| 331 |
-
},
|
| 332 |
-
{
|
| 333 |
-
"epoch": 1.29,
|
| 334 |
-
"eval_loss": 0.4562254846096039,
|
| 335 |
-
"eval_runtime": 22.7972,
|
| 336 |
-
"eval_samples_per_second": 29.39,
|
| 337 |
-
"eval_steps_per_second": 29.39,
|
| 338 |
-
"step": 1950
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 1.33,
|
| 342 |
-
"learning_rate": 3.723418573351279e-05,
|
| 343 |
-
"loss": 0.2994,
|
| 344 |
-
"step": 2000
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 1.33,
|
| 348 |
-
"eval_loss": 0.4507925510406494,
|
| 349 |
-
"eval_runtime": 22.8043,
|
| 350 |
-
"eval_samples_per_second": 29.38,
|
| 351 |
-
"eval_steps_per_second": 29.38,
|
| 352 |
-
"step": 2000
|
| 353 |
-
},
|
| 354 |
-
{
|
| 355 |
-
"epoch": 1.36,
|
| 356 |
-
"eval_loss": 0.45181065797805786,
|
| 357 |
-
"eval_runtime": 22.8163,
|
| 358 |
-
"eval_samples_per_second": 29.365,
|
| 359 |
-
"eval_steps_per_second": 29.365,
|
| 360 |
-
"step": 2050
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 1.39,
|
| 364 |
-
"eval_loss": 0.45148006081581116,
|
| 365 |
-
"eval_runtime": 22.8045,
|
| 366 |
-
"eval_samples_per_second": 29.38,
|
| 367 |
-
"eval_steps_per_second": 29.38,
|
| 368 |
-
"step": 2100
|
| 369 |
-
},
|
| 370 |
-
{
|
| 371 |
-
"epoch": 1.43,
|
| 372 |
-
"eval_loss": 0.45323121547698975,
|
| 373 |
-
"eval_runtime": 22.805,
|
| 374 |
-
"eval_samples_per_second": 29.38,
|
| 375 |
-
"eval_steps_per_second": 29.38,
|
| 376 |
-
"step": 2150
|
| 377 |
-
},
|
| 378 |
-
{
|
| 379 |
-
"epoch": 1.46,
|
| 380 |
-
"eval_loss": 0.45210888981819153,
|
| 381 |
-
"eval_runtime": 22.8053,
|
| 382 |
-
"eval_samples_per_second": 29.379,
|
| 383 |
-
"eval_steps_per_second": 29.379,
|
| 384 |
-
"step": 2200
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 1.49,
|
| 388 |
-
"eval_loss": 0.4499363601207733,
|
| 389 |
-
"eval_runtime": 22.8014,
|
| 390 |
-
"eval_samples_per_second": 29.384,
|
| 391 |
-
"eval_steps_per_second": 29.384,
|
| 392 |
-
"step": 2250
|
| 393 |
-
},
|
| 394 |
-
{
|
| 395 |
-
"epoch": 1.53,
|
| 396 |
-
"eval_loss": 0.4511328935623169,
|
| 397 |
-
"eval_runtime": 22.8034,
|
| 398 |
-
"eval_samples_per_second": 29.382,
|
| 399 |
-
"eval_steps_per_second": 29.382,
|
| 400 |
-
"step": 2300
|
| 401 |
-
},
|
| 402 |
-
{
|
| 403 |
-
"epoch": 1.56,
|
| 404 |
-
"eval_loss": 0.44887685775756836,
|
| 405 |
-
"eval_runtime": 22.8034,
|
| 406 |
-
"eval_samples_per_second": 29.382,
|
| 407 |
-
"eval_steps_per_second": 29.382,
|
| 408 |
-
"step": 2350
|
| 409 |
-
},
|
| 410 |
-
{
|
| 411 |
-
"epoch": 1.59,
|
| 412 |
-
"eval_loss": 0.4465619921684265,
|
| 413 |
-
"eval_runtime": 22.8019,
|
| 414 |
-
"eval_samples_per_second": 29.383,
|
| 415 |
-
"eval_steps_per_second": 29.383,
|
| 416 |
-
"step": 2400
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 1.63,
|
| 420 |
-
"eval_loss": 0.4489704966545105,
|
| 421 |
-
"eval_runtime": 22.801,
|
| 422 |
-
"eval_samples_per_second": 29.385,
|
| 423 |
-
"eval_steps_per_second": 29.385,
|
| 424 |
-
"step": 2450
|
| 425 |
-
},
|
| 426 |
-
{
|
| 427 |
-
"epoch": 1.66,
|
| 428 |
-
"learning_rate": 3.386944818304172e-05,
|
| 429 |
-
"loss": 0.2946,
|
| 430 |
-
"step": 2500
|
| 431 |
-
},
|
| 432 |
-
{
|
| 433 |
-
"epoch": 1.66,
|
| 434 |
-
"eval_loss": 0.4510194659233093,
|
| 435 |
-
"eval_runtime": 22.7961,
|
| 436 |
-
"eval_samples_per_second": 29.391,
|
| 437 |
-
"eval_steps_per_second": 29.391,
|
| 438 |
-
"step": 2500
|
| 439 |
-
},
|
| 440 |
-
{
|
| 441 |
-
"epoch": 1.69,
|
| 442 |
-
"eval_loss": 0.4499521851539612,
|
| 443 |
-
"eval_runtime": 22.8095,
|
| 444 |
-
"eval_samples_per_second": 29.374,
|
| 445 |
-
"eval_steps_per_second": 29.374,
|
| 446 |
-
"step": 2550
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 1.73,
|
| 450 |
-
"eval_loss": 0.44297194480895996,
|
| 451 |
-
"eval_runtime": 22.8114,
|
| 452 |
-
"eval_samples_per_second": 29.371,
|
| 453 |
-
"eval_steps_per_second": 29.371,
|
| 454 |
-
"step": 2600
|
| 455 |
-
},
|
| 456 |
-
{
|
| 457 |
-
"epoch": 1.76,
|
| 458 |
-
"eval_loss": 0.4441739618778229,
|
| 459 |
-
"eval_runtime": 22.8127,
|
| 460 |
-
"eval_samples_per_second": 29.37,
|
| 461 |
-
"eval_steps_per_second": 29.37,
|
| 462 |
-
"step": 2650
|
| 463 |
-
},
|
| 464 |
-
{
|
| 465 |
-
"epoch": 1.79,
|
| 466 |
-
"eval_loss": 0.44600966572761536,
|
| 467 |
-
"eval_runtime": 22.8114,
|
| 468 |
-
"eval_samples_per_second": 29.371,
|
| 469 |
-
"eval_steps_per_second": 29.371,
|
| 470 |
-
"step": 2700
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 1.83,
|
| 474 |
-
"eval_loss": 0.4461658000946045,
|
| 475 |
-
"eval_runtime": 22.8112,
|
| 476 |
-
"eval_samples_per_second": 29.372,
|
| 477 |
-
"eval_steps_per_second": 29.372,
|
| 478 |
-
"step": 2750
|
| 479 |
-
},
|
| 480 |
-
{
|
| 481 |
-
"epoch": 1.86,
|
| 482 |
-
"eval_loss": 0.4427674114704132,
|
| 483 |
-
"eval_runtime": 22.8085,
|
| 484 |
-
"eval_samples_per_second": 29.375,
|
| 485 |
-
"eval_steps_per_second": 29.375,
|
| 486 |
-
"step": 2800
|
| 487 |
-
},
|
| 488 |
-
{
|
| 489 |
-
"epoch": 1.89,
|
| 490 |
-
"eval_loss": 0.44410833716392517,
|
| 491 |
-
"eval_runtime": 22.8091,
|
| 492 |
-
"eval_samples_per_second": 29.374,
|
| 493 |
-
"eval_steps_per_second": 29.374,
|
| 494 |
-
"step": 2850
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 1.93,
|
| 498 |
-
"eval_loss": 0.4429542124271393,
|
| 499 |
-
"eval_runtime": 22.8068,
|
| 500 |
-
"eval_samples_per_second": 29.377,
|
| 501 |
-
"eval_steps_per_second": 29.377,
|
| 502 |
-
"step": 2900
|
| 503 |
-
},
|
| 504 |
-
{
|
| 505 |
-
"epoch": 1.96,
|
| 506 |
-
"eval_loss": 0.44093453884124756,
|
| 507 |
-
"eval_runtime": 22.8017,
|
| 508 |
-
"eval_samples_per_second": 29.384,
|
| 509 |
-
"eval_steps_per_second": 29.384,
|
| 510 |
-
"step": 2950
|
| 511 |
-
},
|
| 512 |
-
{
|
| 513 |
-
"epoch": 1.99,
|
| 514 |
-
"learning_rate": 3.050471063257066e-05,
|
| 515 |
-
"loss": 0.2961,
|
| 516 |
-
"step": 3000
|
| 517 |
-
},
|
| 518 |
-
{
|
| 519 |
-
"epoch": 1.99,
|
| 520 |
-
"eval_loss": 0.44067710638046265,
|
| 521 |
-
"eval_runtime": 22.8039,
|
| 522 |
-
"eval_samples_per_second": 29.381,
|
| 523 |
-
"eval_steps_per_second": 29.381,
|
| 524 |
-
"step": 3000
|
| 525 |
-
}
|
| 526 |
-
],
|
| 527 |
-
"max_steps": 7530,
|
| 528 |
-
"num_train_epochs": 5,
|
| 529 |
-
"total_flos": 5.2553298345984e+16,
|
| 530 |
-
"trial_name": null,
|
| 531 |
-
"trial_params": null
|
| 532 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|