Training in progress, epoch 0
Browse files- all_results.json +11 -6
- eval_results.json +6 -6
- model.safetensors +1 -1
- runs/Dec14_22-06-10_bc148a6f3d4e/events.out.tfevents.1702595991.bc148a6f3d4e.163484.2 +3 -0
- train_results.json +6 -6
- trainer_state.json +2046 -273
all_results.json
CHANGED
|
@@ -1,8 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.9405358686257562,
|
| 4 |
+
"eval_loss": 0.3180586099624634,
|
| 5 |
+
"eval_runtime": 38.4742,
|
| 6 |
+
"eval_samples_per_second": 150.361,
|
| 7 |
+
"eval_steps_per_second": 4.704,
|
| 8 |
+
"total_flos": 1.3140619208067262e+19,
|
| 9 |
+
"train_loss": 1.0548507860728673,
|
| 10 |
+
"train_runtime": 4291.8856,
|
| 11 |
+
"train_samples_per_second": 121.301,
|
| 12 |
+
"train_steps_per_second": 0.946
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.9405358686257562,
|
| 4 |
+
"eval_loss": 0.3180586099624634,
|
| 5 |
+
"eval_runtime": 38.4742,
|
| 6 |
+
"eval_samples_per_second": 150.361,
|
| 7 |
+
"eval_steps_per_second": 4.704
|
| 8 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 112253100
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:030d3ef6ccb0a545c661afbf7c1e10d6910809218395abce306b827ed531e30e
|
| 3 |
size 112253100
|
runs/Dec14_22-06-10_bc148a6f3d4e/events.out.tfevents.1702595991.bc148a6f3d4e.163484.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17a5f6fc9345e1ebb2b6e3c5f242375fb3fdfda9006738907ee41cf3613a8a4e
|
| 3 |
+
size 36401
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss":
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"total_flos": 1.3140619208067262e+19,
|
| 4 |
+
"train_loss": 1.0548507860728673,
|
| 5 |
+
"train_runtime": 4291.8856,
|
| 6 |
+
"train_samples_per_second": 121.301,
|
| 7 |
+
"train_steps_per_second": 0.946
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,782 +1,2555 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-
|
| 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": 0.02,
|
| 13 |
-
"learning_rate":
|
| 14 |
-
"loss":
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.05,
|
| 19 |
-
"learning_rate":
|
| 20 |
-
"loss":
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.07,
|
| 25 |
-
"learning_rate":
|
| 26 |
-
"loss":
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.1,
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss":
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.12,
|
| 37 |
-
"learning_rate":
|
| 38 |
-
"loss":
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.15,
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss":
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.17,
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss":
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.2,
|
| 55 |
-
"learning_rate":
|
| 56 |
-
"loss":
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.22,
|
| 61 |
-
"learning_rate":
|
| 62 |
-
"loss":
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.25,
|
| 67 |
-
"learning_rate":
|
| 68 |
-
"loss":
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.27,
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss":
|
| 75 |
"step": 110
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.3,
|
| 79 |
-
"learning_rate":
|
| 80 |
-
"loss":
|
| 81 |
"step": 120
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.32,
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss":
|
| 87 |
"step": 130
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.34,
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss":
|
| 93 |
"step": 140
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.37,
|
| 97 |
-
"learning_rate":
|
| 98 |
-
"loss":
|
| 99 |
"step": 150
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.39,
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss":
|
| 105 |
"step": 160
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.42,
|
| 109 |
-
"learning_rate":
|
| 110 |
-
"loss":
|
| 111 |
"step": 170
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.44,
|
| 115 |
-
"learning_rate":
|
| 116 |
-
"loss":
|
| 117 |
"step": 180
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.47,
|
| 121 |
-
"learning_rate":
|
| 122 |
-
"loss":
|
| 123 |
"step": 190
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 0.49,
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss":
|
| 129 |
"step": 200
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.52,
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss":
|
| 135 |
"step": 210
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.54,
|
| 139 |
-
"learning_rate":
|
| 140 |
-
"loss":
|
| 141 |
"step": 220
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.57,
|
| 145 |
-
"learning_rate":
|
| 146 |
-
"loss":
|
| 147 |
"step": 230
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.59,
|
| 151 |
-
"learning_rate":
|
| 152 |
-
"loss":
|
| 153 |
"step": 240
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.61,
|
| 157 |
-
"learning_rate":
|
| 158 |
-
"loss":
|
| 159 |
"step": 250
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.64,
|
| 163 |
-
"learning_rate":
|
| 164 |
-
"loss":
|
| 165 |
"step": 260
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 0.66,
|
| 169 |
-
"learning_rate":
|
| 170 |
-
"loss":
|
| 171 |
"step": 270
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.69,
|
| 175 |
-
"learning_rate":
|
| 176 |
-
"loss":
|
| 177 |
"step": 280
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.71,
|
| 181 |
-
"learning_rate":
|
| 182 |
-
"loss":
|
| 183 |
"step": 290
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.74,
|
| 187 |
-
"learning_rate":
|
| 188 |
-
"loss":
|
| 189 |
"step": 300
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.76,
|
| 193 |
-
"learning_rate":
|
| 194 |
-
"loss":
|
| 195 |
"step": 310
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 0.79,
|
| 199 |
-
"learning_rate":
|
| 200 |
-
"loss":
|
| 201 |
"step": 320
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 0.81,
|
| 205 |
-
"learning_rate": 4.
|
| 206 |
-
"loss":
|
| 207 |
"step": 330
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 0.84,
|
| 211 |
-
"learning_rate": 4.
|
| 212 |
-
"loss":
|
| 213 |
"step": 340
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.86,
|
| 217 |
-
"learning_rate":
|
| 218 |
-
"loss":
|
| 219 |
"step": 350
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.89,
|
| 223 |
-
"learning_rate":
|
| 224 |
-
"loss":
|
| 225 |
"step": 360
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.91,
|
| 229 |
-
"learning_rate":
|
| 230 |
-
"loss":
|
| 231 |
"step": 370
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.93,
|
| 235 |
-
"learning_rate":
|
| 236 |
-
"loss":
|
| 237 |
"step": 380
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.96,
|
| 241 |
-
"learning_rate":
|
| 242 |
-
"loss":
|
| 243 |
"step": 390
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.98,
|
| 247 |
-
"learning_rate":
|
| 248 |
-
"loss":
|
| 249 |
"step": 400
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 1.0,
|
| 253 |
-
"eval_accuracy": 0.
|
| 254 |
-
"eval_loss":
|
| 255 |
-
"eval_runtime":
|
| 256 |
-
"eval_samples_per_second":
|
| 257 |
-
"eval_steps_per_second": 6.
|
| 258 |
"step": 406
|
| 259 |
},
|
| 260 |
{
|
| 261 |
"epoch": 1.01,
|
| 262 |
-
"learning_rate":
|
| 263 |
-
"loss":
|
| 264 |
"step": 410
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 1.03,
|
| 268 |
-
"learning_rate":
|
| 269 |
-
"loss":
|
| 270 |
"step": 420
|
| 271 |
},
|
| 272 |
{
|
| 273 |
"epoch": 1.06,
|
| 274 |
-
"learning_rate":
|
| 275 |
-
"loss":
|
| 276 |
"step": 430
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 1.08,
|
| 280 |
-
"learning_rate":
|
| 281 |
-
"loss":
|
| 282 |
"step": 440
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 1.11,
|
| 286 |
-
"learning_rate":
|
| 287 |
-
"loss":
|
| 288 |
"step": 450
|
| 289 |
},
|
| 290 |
{
|
| 291 |
"epoch": 1.13,
|
| 292 |
-
"learning_rate":
|
| 293 |
-
"loss":
|
| 294 |
"step": 460
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 1.16,
|
| 298 |
-
"learning_rate":
|
| 299 |
-
"loss":
|
| 300 |
"step": 470
|
| 301 |
},
|
| 302 |
{
|
| 303 |
"epoch": 1.18,
|
| 304 |
-
"learning_rate":
|
| 305 |
-
"loss":
|
| 306 |
"step": 480
|
| 307 |
},
|
| 308 |
{
|
| 309 |
"epoch": 1.2,
|
| 310 |
-
"learning_rate":
|
| 311 |
-
"loss":
|
| 312 |
"step": 490
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 1.23,
|
| 316 |
-
"learning_rate":
|
| 317 |
-
"loss":
|
| 318 |
"step": 500
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 1.25,
|
| 322 |
-
"learning_rate":
|
| 323 |
-
"loss":
|
| 324 |
"step": 510
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 1.28,
|
| 328 |
-
"learning_rate":
|
| 329 |
-
"loss":
|
| 330 |
"step": 520
|
| 331 |
},
|
| 332 |
{
|
| 333 |
"epoch": 1.3,
|
| 334 |
-
"learning_rate":
|
| 335 |
-
"loss":
|
| 336 |
"step": 530
|
| 337 |
},
|
| 338 |
{
|
| 339 |
"epoch": 1.33,
|
| 340 |
-
"learning_rate":
|
| 341 |
-
"loss":
|
| 342 |
"step": 540
|
| 343 |
},
|
| 344 |
{
|
| 345 |
"epoch": 1.35,
|
| 346 |
-
"learning_rate":
|
| 347 |
-
"loss":
|
| 348 |
"step": 550
|
| 349 |
},
|
| 350 |
{
|
| 351 |
"epoch": 1.38,
|
| 352 |
-
"learning_rate":
|
| 353 |
-
"loss":
|
| 354 |
"step": 560
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 1.4,
|
| 358 |
-
"learning_rate":
|
| 359 |
-
"loss":
|
| 360 |
"step": 570
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 1.43,
|
| 364 |
-
"learning_rate":
|
| 365 |
-
"loss":
|
| 366 |
"step": 580
|
| 367 |
},
|
| 368 |
{
|
| 369 |
"epoch": 1.45,
|
| 370 |
-
"learning_rate":
|
| 371 |
-
"loss":
|
| 372 |
"step": 590
|
| 373 |
},
|
| 374 |
{
|
| 375 |
"epoch": 1.48,
|
| 376 |
-
"learning_rate":
|
| 377 |
-
"loss":
|
| 378 |
"step": 600
|
| 379 |
},
|
| 380 |
{
|
| 381 |
"epoch": 1.5,
|
| 382 |
-
"learning_rate":
|
| 383 |
-
"loss":
|
| 384 |
"step": 610
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 1.52,
|
| 388 |
-
"learning_rate":
|
| 389 |
-
"loss":
|
| 390 |
"step": 620
|
| 391 |
},
|
| 392 |
{
|
| 393 |
"epoch": 1.55,
|
| 394 |
-
"learning_rate":
|
| 395 |
-
"loss": 1.
|
| 396 |
"step": 630
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 1.57,
|
| 400 |
-
"learning_rate":
|
| 401 |
-
"loss":
|
| 402 |
"step": 640
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 1.6,
|
| 406 |
-
"learning_rate":
|
| 407 |
-
"loss":
|
| 408 |
"step": 650
|
| 409 |
},
|
| 410 |
{
|
| 411 |
"epoch": 1.62,
|
| 412 |
-
"learning_rate":
|
| 413 |
-
"loss":
|
| 414 |
"step": 660
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 1.65,
|
| 418 |
-
"learning_rate":
|
| 419 |
-
"loss":
|
| 420 |
"step": 670
|
| 421 |
},
|
| 422 |
{
|
| 423 |
"epoch": 1.67,
|
| 424 |
-
"learning_rate":
|
| 425 |
-
"loss":
|
| 426 |
"step": 680
|
| 427 |
},
|
| 428 |
{
|
| 429 |
"epoch": 1.7,
|
| 430 |
-
"learning_rate":
|
| 431 |
-
"loss": 1.
|
| 432 |
"step": 690
|
| 433 |
},
|
| 434 |
{
|
| 435 |
"epoch": 1.72,
|
| 436 |
-
"learning_rate":
|
| 437 |
-
"loss":
|
| 438 |
"step": 700
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 1.75,
|
| 442 |
-
"learning_rate":
|
| 443 |
-
"loss": 1.
|
| 444 |
"step": 710
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 1.77,
|
| 448 |
-
"learning_rate":
|
| 449 |
-
"loss":
|
| 450 |
"step": 720
|
| 451 |
},
|
| 452 |
{
|
| 453 |
"epoch": 1.79,
|
| 454 |
-
"learning_rate":
|
| 455 |
-
"loss": 1.
|
| 456 |
"step": 730
|
| 457 |
},
|
| 458 |
{
|
| 459 |
"epoch": 1.82,
|
| 460 |
-
"learning_rate":
|
| 461 |
-
"loss": 1.
|
| 462 |
"step": 740
|
| 463 |
},
|
| 464 |
{
|
| 465 |
"epoch": 1.84,
|
| 466 |
-
"learning_rate":
|
| 467 |
-
"loss": 1.
|
| 468 |
"step": 750
|
| 469 |
},
|
| 470 |
{
|
| 471 |
"epoch": 1.87,
|
| 472 |
-
"learning_rate":
|
| 473 |
-
"loss": 1.
|
| 474 |
"step": 760
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 1.89,
|
| 478 |
-
"learning_rate":
|
| 479 |
-
"loss":
|
| 480 |
"step": 770
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 1.92,
|
| 484 |
-
"learning_rate":
|
| 485 |
-
"loss": 1.
|
| 486 |
"step": 780
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 1.94,
|
| 490 |
-
"learning_rate":
|
| 491 |
-
"loss": 1.
|
| 492 |
"step": 790
|
| 493 |
},
|
| 494 |
{
|
| 495 |
"epoch": 1.97,
|
| 496 |
-
"learning_rate":
|
| 497 |
-
"loss": 1.
|
| 498 |
"step": 800
|
| 499 |
},
|
| 500 |
{
|
| 501 |
"epoch": 1.99,
|
| 502 |
-
"learning_rate":
|
| 503 |
-
"loss": 1.
|
| 504 |
"step": 810
|
| 505 |
},
|
| 506 |
{
|
| 507 |
"epoch": 2.0,
|
| 508 |
-
"eval_accuracy": 0.
|
| 509 |
-
"eval_loss":
|
| 510 |
-
"eval_runtime": 27.
|
| 511 |
-
"eval_samples_per_second":
|
| 512 |
-
"eval_steps_per_second": 6.
|
| 513 |
"step": 813
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 2.02,
|
| 517 |
-
"learning_rate":
|
| 518 |
-
"loss": 1.
|
| 519 |
"step": 820
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 2.04,
|
| 523 |
-
"learning_rate":
|
| 524 |
-
"loss": 1.
|
| 525 |
"step": 830
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 2.07,
|
| 529 |
-
"learning_rate":
|
| 530 |
-
"loss": 1.
|
| 531 |
"step": 840
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 2.09,
|
| 535 |
-
"learning_rate":
|
| 536 |
-
"loss": 1.
|
| 537 |
"step": 850
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 2.11,
|
| 541 |
-
"learning_rate":
|
| 542 |
-
"loss": 1.
|
| 543 |
"step": 860
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 2.14,
|
| 547 |
-
"learning_rate":
|
| 548 |
-
"loss": 1.
|
| 549 |
"step": 870
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 2.16,
|
| 553 |
-
"learning_rate":
|
| 554 |
-
"loss": 1.
|
| 555 |
"step": 880
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 2.19,
|
| 559 |
-
"learning_rate":
|
| 560 |
-
"loss": 1.
|
| 561 |
"step": 890
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 2.21,
|
| 565 |
-
"learning_rate":
|
| 566 |
-
"loss": 1.
|
| 567 |
"step": 900
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 2.24,
|
| 571 |
-
"learning_rate":
|
| 572 |
-
"loss": 1.
|
| 573 |
"step": 910
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 2.26,
|
| 577 |
-
"learning_rate":
|
| 578 |
-
"loss": 1.
|
| 579 |
"step": 920
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 2.29,
|
| 583 |
-
"learning_rate":
|
| 584 |
-
"loss": 1.
|
| 585 |
"step": 930
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 2.31,
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss": 1.
|
| 591 |
"step": 940
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 2.34,
|
| 595 |
-
"learning_rate":
|
| 596 |
-
"loss": 1.
|
| 597 |
"step": 950
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 2.36,
|
| 601 |
-
"learning_rate":
|
| 602 |
-
"loss": 1.
|
| 603 |
"step": 960
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 2.38,
|
| 607 |
-
"learning_rate":
|
| 608 |
-
"loss": 1.
|
| 609 |
"step": 970
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 2.41,
|
| 613 |
-
"learning_rate":
|
| 614 |
-
"loss": 1.
|
| 615 |
"step": 980
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 2.43,
|
| 619 |
-
"learning_rate":
|
| 620 |
-
"loss": 1.
|
| 621 |
"step": 990
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 2.46,
|
| 625 |
-
"learning_rate":
|
| 626 |
-
"loss": 1.
|
| 627 |
"step": 1000
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 2.48,
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss": 1.
|
| 633 |
"step": 1010
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 2.51,
|
| 637 |
-
"learning_rate":
|
| 638 |
-
"loss": 1.
|
| 639 |
"step": 1020
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 2.53,
|
| 643 |
-
"learning_rate":
|
| 644 |
-
"loss": 1.
|
| 645 |
"step": 1030
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 2.56,
|
| 649 |
-
"learning_rate":
|
| 650 |
-
"loss": 1.
|
| 651 |
"step": 1040
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 2.58,
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 1.
|
| 657 |
"step": 1050
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 2.61,
|
| 661 |
-
"learning_rate":
|
| 662 |
-
"loss": 1.
|
| 663 |
"step": 1060
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 2.63,
|
| 667 |
-
"learning_rate":
|
| 668 |
-
"loss": 1.
|
| 669 |
"step": 1070
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 2.66,
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss": 1.
|
| 675 |
"step": 1080
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 2.68,
|
| 679 |
-
"learning_rate":
|
| 680 |
-
"loss": 1.
|
| 681 |
"step": 1090
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 2.7,
|
| 685 |
-
"learning_rate":
|
| 686 |
-
"loss": 1.
|
| 687 |
"step": 1100
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 2.73,
|
| 691 |
-
"learning_rate": 4.
|
| 692 |
-
"loss": 1.
|
| 693 |
"step": 1110
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 2.75,
|
| 697 |
-
"learning_rate": 4.
|
| 698 |
-
"loss": 1.
|
| 699 |
"step": 1120
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 2.78,
|
| 703 |
-
"learning_rate": 4.
|
| 704 |
-
"loss": 1.
|
| 705 |
"step": 1130
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 2.8,
|
| 709 |
-
"learning_rate": 3.
|
| 710 |
-
"loss": 1.
|
| 711 |
"step": 1140
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 2.83,
|
| 715 |
-
"learning_rate": 3.
|
| 716 |
-
"loss": 1.
|
| 717 |
"step": 1150
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 2.85,
|
| 721 |
-
"learning_rate":
|
| 722 |
-
"loss": 1.
|
| 723 |
"step": 1160
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 2.88,
|
| 727 |
-
"learning_rate":
|
| 728 |
-
"loss": 1.
|
| 729 |
"step": 1170
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 2.9,
|
| 733 |
-
"learning_rate":
|
| 734 |
-
"loss": 1.
|
| 735 |
"step": 1180
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 2.93,
|
| 739 |
-
"learning_rate":
|
| 740 |
-
"loss": 1.
|
| 741 |
"step": 1190
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 2.95,
|
| 745 |
-
"learning_rate":
|
| 746 |
-
"loss": 1.
|
| 747 |
"step": 1200
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 2.97,
|
| 751 |
-
"learning_rate": 3.
|
| 752 |
-
"loss": 1.
|
| 753 |
"step": 1210
|
| 754 |
},
|
| 755 |
{
|
| 756 |
-
"epoch":
|
| 757 |
-
"
|
| 758 |
-
"
|
| 759 |
-
"
|
| 760 |
-
"eval_samples_per_second": 209.909,
|
| 761 |
-
"eval_steps_per_second": 6.568,
|
| 762 |
-
"step": 1218
|
| 763 |
},
|
| 764 |
{
|
| 765 |
-
"epoch":
|
| 766 |
-
"
|
| 767 |
-
"
|
| 768 |
-
"
|
| 769 |
-
"
|
| 770 |
-
"
|
| 771 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 772 |
}
|
| 773 |
],
|
| 774 |
"logging_steps": 10,
|
| 775 |
-
"max_steps":
|
| 776 |
"num_input_tokens_seen": 0,
|
| 777 |
-
"num_train_epochs":
|
| 778 |
"save_steps": 500,
|
| 779 |
-
"total_flos":
|
| 780 |
"train_batch_size": 32,
|
| 781 |
"trial_name": null,
|
| 782 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.9405358686257562,
|
| 3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat/checkpoint-3660",
|
| 4 |
+
"epoch": 9.981561155500922,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 4060,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.02,
|
| 13 |
+
"learning_rate": 1.2315270935960593e-06,
|
| 14 |
+
"loss": 1.8272,
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.05,
|
| 19 |
+
"learning_rate": 2.4630541871921186e-06,
|
| 20 |
+
"loss": 1.7835,
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.07,
|
| 25 |
+
"learning_rate": 3.6945812807881777e-06,
|
| 26 |
+
"loss": 1.751,
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.1,
|
| 31 |
+
"learning_rate": 4.926108374384237e-06,
|
| 32 |
+
"loss": 1.7539,
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.12,
|
| 37 |
+
"learning_rate": 6.157635467980296e-06,
|
| 38 |
+
"loss": 1.7125,
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.15,
|
| 43 |
+
"learning_rate": 7.3891625615763555e-06,
|
| 44 |
+
"loss": 1.6708,
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.17,
|
| 49 |
+
"learning_rate": 8.620689655172414e-06,
|
| 50 |
+
"loss": 1.7727,
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.2,
|
| 55 |
+
"learning_rate": 9.852216748768475e-06,
|
| 56 |
+
"loss": 1.6377,
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.22,
|
| 61 |
+
"learning_rate": 1.1083743842364533e-05,
|
| 62 |
+
"loss": 1.8024,
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.25,
|
| 67 |
+
"learning_rate": 1.2315270935960592e-05,
|
| 68 |
+
"loss": 1.7476,
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.27,
|
| 73 |
+
"learning_rate": 1.354679802955665e-05,
|
| 74 |
+
"loss": 1.6509,
|
| 75 |
"step": 110
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.3,
|
| 79 |
+
"learning_rate": 1.4778325123152711e-05,
|
| 80 |
+
"loss": 1.7052,
|
| 81 |
"step": 120
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.32,
|
| 85 |
+
"learning_rate": 1.6009852216748768e-05,
|
| 86 |
+
"loss": 1.7003,
|
| 87 |
"step": 130
|
| 88 |
},
|
| 89 |
{
|
| 90 |
"epoch": 0.34,
|
| 91 |
+
"learning_rate": 1.7241379310344828e-05,
|
| 92 |
+
"loss": 1.7424,
|
| 93 |
"step": 140
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.37,
|
| 97 |
+
"learning_rate": 1.847290640394089e-05,
|
| 98 |
+
"loss": 1.7223,
|
| 99 |
"step": 150
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.39,
|
| 103 |
+
"learning_rate": 1.970443349753695e-05,
|
| 104 |
+
"loss": 1.6932,
|
| 105 |
"step": 160
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.42,
|
| 109 |
+
"learning_rate": 2.0935960591133006e-05,
|
| 110 |
+
"loss": 1.6393,
|
| 111 |
"step": 170
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.44,
|
| 115 |
+
"learning_rate": 2.2167487684729066e-05,
|
| 116 |
+
"loss": 1.7023,
|
| 117 |
"step": 180
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.47,
|
| 121 |
+
"learning_rate": 2.3399014778325123e-05,
|
| 122 |
+
"loss": 1.6219,
|
| 123 |
"step": 190
|
| 124 |
},
|
| 125 |
{
|
| 126 |
"epoch": 0.49,
|
| 127 |
+
"learning_rate": 2.4630541871921184e-05,
|
| 128 |
+
"loss": 1.7031,
|
| 129 |
"step": 200
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.52,
|
| 133 |
+
"learning_rate": 2.5862068965517244e-05,
|
| 134 |
+
"loss": 1.7309,
|
| 135 |
"step": 210
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.54,
|
| 139 |
+
"learning_rate": 2.70935960591133e-05,
|
| 140 |
+
"loss": 1.7251,
|
| 141 |
"step": 220
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.57,
|
| 145 |
+
"learning_rate": 2.8325123152709358e-05,
|
| 146 |
+
"loss": 1.7039,
|
| 147 |
"step": 230
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.59,
|
| 151 |
+
"learning_rate": 2.9556650246305422e-05,
|
| 152 |
+
"loss": 1.6753,
|
| 153 |
"step": 240
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.61,
|
| 157 |
+
"learning_rate": 3.078817733990148e-05,
|
| 158 |
+
"loss": 1.7149,
|
| 159 |
"step": 250
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.64,
|
| 163 |
+
"learning_rate": 3.2019704433497536e-05,
|
| 164 |
+
"loss": 1.7101,
|
| 165 |
"step": 260
|
| 166 |
},
|
| 167 |
{
|
| 168 |
"epoch": 0.66,
|
| 169 |
+
"learning_rate": 3.3251231527093596e-05,
|
| 170 |
+
"loss": 1.6208,
|
| 171 |
"step": 270
|
| 172 |
},
|
| 173 |
{
|
| 174 |
"epoch": 0.69,
|
| 175 |
+
"learning_rate": 3.4482758620689657e-05,
|
| 176 |
+
"loss": 1.5855,
|
| 177 |
"step": 280
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.71,
|
| 181 |
+
"learning_rate": 3.571428571428572e-05,
|
| 182 |
+
"loss": 1.741,
|
| 183 |
"step": 290
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.74,
|
| 187 |
+
"learning_rate": 3.694581280788178e-05,
|
| 188 |
+
"loss": 1.6925,
|
| 189 |
"step": 300
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.76,
|
| 193 |
+
"learning_rate": 3.817733990147783e-05,
|
| 194 |
+
"loss": 1.6607,
|
| 195 |
"step": 310
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 0.79,
|
| 199 |
+
"learning_rate": 3.94088669950739e-05,
|
| 200 |
+
"loss": 1.5157,
|
| 201 |
"step": 320
|
| 202 |
},
|
| 203 |
{
|
| 204 |
"epoch": 0.81,
|
| 205 |
+
"learning_rate": 4.064039408866995e-05,
|
| 206 |
+
"loss": 1.5652,
|
| 207 |
"step": 330
|
| 208 |
},
|
| 209 |
{
|
| 210 |
"epoch": 0.84,
|
| 211 |
+
"learning_rate": 4.187192118226601e-05,
|
| 212 |
+
"loss": 1.6547,
|
| 213 |
"step": 340
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.86,
|
| 217 |
+
"learning_rate": 4.3103448275862066e-05,
|
| 218 |
+
"loss": 1.5618,
|
| 219 |
"step": 350
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.89,
|
| 223 |
+
"learning_rate": 4.433497536945813e-05,
|
| 224 |
+
"loss": 1.6555,
|
| 225 |
"step": 360
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.91,
|
| 229 |
+
"learning_rate": 4.5566502463054186e-05,
|
| 230 |
+
"loss": 1.5538,
|
| 231 |
"step": 370
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.93,
|
| 235 |
+
"learning_rate": 4.679802955665025e-05,
|
| 236 |
+
"loss": 1.5496,
|
| 237 |
"step": 380
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.96,
|
| 241 |
+
"learning_rate": 4.802955665024631e-05,
|
| 242 |
+
"loss": 1.5248,
|
| 243 |
"step": 390
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.98,
|
| 247 |
+
"learning_rate": 4.926108374384237e-05,
|
| 248 |
+
"loss": 1.5645,
|
| 249 |
"step": 400
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 1.0,
|
| 253 |
+
"eval_accuracy": 0.7904926534140018,
|
| 254 |
+
"eval_loss": 1.0879656076431274,
|
| 255 |
+
"eval_runtime": 27.6968,
|
| 256 |
+
"eval_samples_per_second": 208.869,
|
| 257 |
+
"eval_steps_per_second": 6.535,
|
| 258 |
"step": 406
|
| 259 |
},
|
| 260 |
{
|
| 261 |
"epoch": 1.01,
|
| 262 |
+
"learning_rate": 4.9945265462506846e-05,
|
| 263 |
+
"loss": 1.5933,
|
| 264 |
"step": 410
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 1.03,
|
| 268 |
+
"learning_rate": 4.980842911877395e-05,
|
| 269 |
+
"loss": 1.5284,
|
| 270 |
"step": 420
|
| 271 |
},
|
| 272 |
{
|
| 273 |
"epoch": 1.06,
|
| 274 |
+
"learning_rate": 4.9671592775041054e-05,
|
| 275 |
+
"loss": 1.5239,
|
| 276 |
"step": 430
|
| 277 |
},
|
| 278 |
{
|
| 279 |
"epoch": 1.08,
|
| 280 |
+
"learning_rate": 4.953475643130816e-05,
|
| 281 |
+
"loss": 1.4781,
|
| 282 |
"step": 440
|
| 283 |
},
|
| 284 |
{
|
| 285 |
"epoch": 1.11,
|
| 286 |
+
"learning_rate": 4.939792008757526e-05,
|
| 287 |
+
"loss": 1.468,
|
| 288 |
"step": 450
|
| 289 |
},
|
| 290 |
{
|
| 291 |
"epoch": 1.13,
|
| 292 |
+
"learning_rate": 4.926108374384237e-05,
|
| 293 |
+
"loss": 1.5165,
|
| 294 |
"step": 460
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 1.16,
|
| 298 |
+
"learning_rate": 4.912424740010947e-05,
|
| 299 |
+
"loss": 1.5641,
|
| 300 |
"step": 470
|
| 301 |
},
|
| 302 |
{
|
| 303 |
"epoch": 1.18,
|
| 304 |
+
"learning_rate": 4.8987411056376576e-05,
|
| 305 |
+
"loss": 1.53,
|
| 306 |
"step": 480
|
| 307 |
},
|
| 308 |
{
|
| 309 |
"epoch": 1.2,
|
| 310 |
+
"learning_rate": 4.885057471264368e-05,
|
| 311 |
+
"loss": 1.4946,
|
| 312 |
"step": 490
|
| 313 |
},
|
| 314 |
{
|
| 315 |
"epoch": 1.23,
|
| 316 |
+
"learning_rate": 4.8713738368910785e-05,
|
| 317 |
+
"loss": 1.5378,
|
| 318 |
"step": 500
|
| 319 |
},
|
| 320 |
{
|
| 321 |
"epoch": 1.25,
|
| 322 |
+
"learning_rate": 4.857690202517789e-05,
|
| 323 |
+
"loss": 1.4228,
|
| 324 |
"step": 510
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 1.28,
|
| 328 |
+
"learning_rate": 4.8440065681444994e-05,
|
| 329 |
+
"loss": 1.4621,
|
| 330 |
"step": 520
|
| 331 |
},
|
| 332 |
{
|
| 333 |
"epoch": 1.3,
|
| 334 |
+
"learning_rate": 4.83032293377121e-05,
|
| 335 |
+
"loss": 1.5405,
|
| 336 |
"step": 530
|
| 337 |
},
|
| 338 |
{
|
| 339 |
"epoch": 1.33,
|
| 340 |
+
"learning_rate": 4.81663929939792e-05,
|
| 341 |
+
"loss": 1.4338,
|
| 342 |
"step": 540
|
| 343 |
},
|
| 344 |
{
|
| 345 |
"epoch": 1.35,
|
| 346 |
+
"learning_rate": 4.802955665024631e-05,
|
| 347 |
+
"loss": 1.3429,
|
| 348 |
"step": 550
|
| 349 |
},
|
| 350 |
{
|
| 351 |
"epoch": 1.38,
|
| 352 |
+
"learning_rate": 4.789272030651341e-05,
|
| 353 |
+
"loss": 1.4912,
|
| 354 |
"step": 560
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 1.4,
|
| 358 |
+
"learning_rate": 4.7755883962780516e-05,
|
| 359 |
+
"loss": 1.4859,
|
| 360 |
"step": 570
|
| 361 |
},
|
| 362 |
{
|
| 363 |
"epoch": 1.43,
|
| 364 |
+
"learning_rate": 4.761904761904762e-05,
|
| 365 |
+
"loss": 1.3696,
|
| 366 |
"step": 580
|
| 367 |
},
|
| 368 |
{
|
| 369 |
"epoch": 1.45,
|
| 370 |
+
"learning_rate": 4.7482211275314725e-05,
|
| 371 |
+
"loss": 1.3818,
|
| 372 |
"step": 590
|
| 373 |
},
|
| 374 |
{
|
| 375 |
"epoch": 1.48,
|
| 376 |
+
"learning_rate": 4.734537493158183e-05,
|
| 377 |
+
"loss": 1.395,
|
| 378 |
"step": 600
|
| 379 |
},
|
| 380 |
{
|
| 381 |
"epoch": 1.5,
|
| 382 |
+
"learning_rate": 4.7208538587848934e-05,
|
| 383 |
+
"loss": 1.5301,
|
| 384 |
"step": 610
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 1.52,
|
| 388 |
+
"learning_rate": 4.707170224411604e-05,
|
| 389 |
+
"loss": 1.3008,
|
| 390 |
"step": 620
|
| 391 |
},
|
| 392 |
{
|
| 393 |
"epoch": 1.55,
|
| 394 |
+
"learning_rate": 4.693486590038315e-05,
|
| 395 |
+
"loss": 1.3089,
|
| 396 |
"step": 630
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 1.57,
|
| 400 |
+
"learning_rate": 4.679802955665025e-05,
|
| 401 |
+
"loss": 1.4767,
|
| 402 |
"step": 640
|
| 403 |
},
|
| 404 |
{
|
| 405 |
"epoch": 1.6,
|
| 406 |
+
"learning_rate": 4.666119321291735e-05,
|
| 407 |
+
"loss": 1.3715,
|
| 408 |
"step": 650
|
| 409 |
},
|
| 410 |
{
|
| 411 |
"epoch": 1.62,
|
| 412 |
+
"learning_rate": 4.652435686918446e-05,
|
| 413 |
+
"loss": 1.3359,
|
| 414 |
"step": 660
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 1.65,
|
| 418 |
+
"learning_rate": 4.638752052545156e-05,
|
| 419 |
+
"loss": 1.38,
|
| 420 |
"step": 670
|
| 421 |
},
|
| 422 |
{
|
| 423 |
"epoch": 1.67,
|
| 424 |
+
"learning_rate": 4.6250684181718664e-05,
|
| 425 |
+
"loss": 1.4252,
|
| 426 |
"step": 680
|
| 427 |
},
|
| 428 |
{
|
| 429 |
"epoch": 1.7,
|
| 430 |
+
"learning_rate": 4.611384783798577e-05,
|
| 431 |
+
"loss": 1.2751,
|
| 432 |
"step": 690
|
| 433 |
},
|
| 434 |
{
|
| 435 |
"epoch": 1.72,
|
| 436 |
+
"learning_rate": 4.597701149425287e-05,
|
| 437 |
+
"loss": 1.4558,
|
| 438 |
"step": 700
|
| 439 |
},
|
| 440 |
{
|
| 441 |
"epoch": 1.75,
|
| 442 |
+
"learning_rate": 4.5840175150519984e-05,
|
| 443 |
+
"loss": 1.3744,
|
| 444 |
"step": 710
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 1.77,
|
| 448 |
+
"learning_rate": 4.570333880678708e-05,
|
| 449 |
+
"loss": 1.3676,
|
| 450 |
"step": 720
|
| 451 |
},
|
| 452 |
{
|
| 453 |
"epoch": 1.79,
|
| 454 |
+
"learning_rate": 4.5566502463054186e-05,
|
| 455 |
+
"loss": 1.3031,
|
| 456 |
"step": 730
|
| 457 |
},
|
| 458 |
{
|
| 459 |
"epoch": 1.82,
|
| 460 |
+
"learning_rate": 4.54296661193213e-05,
|
| 461 |
+
"loss": 1.346,
|
| 462 |
"step": 740
|
| 463 |
},
|
| 464 |
{
|
| 465 |
"epoch": 1.84,
|
| 466 |
+
"learning_rate": 4.5292829775588395e-05,
|
| 467 |
+
"loss": 1.337,
|
| 468 |
"step": 750
|
| 469 |
},
|
| 470 |
{
|
| 471 |
"epoch": 1.87,
|
| 472 |
+
"learning_rate": 4.5155993431855506e-05,
|
| 473 |
+
"loss": 1.318,
|
| 474 |
"step": 760
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 1.89,
|
| 478 |
+
"learning_rate": 4.501915708812261e-05,
|
| 479 |
+
"loss": 1.4677,
|
| 480 |
"step": 770
|
| 481 |
},
|
| 482 |
{
|
| 483 |
"epoch": 1.92,
|
| 484 |
+
"learning_rate": 4.488232074438971e-05,
|
| 485 |
+
"loss": 1.2769,
|
| 486 |
"step": 780
|
| 487 |
},
|
| 488 |
{
|
| 489 |
"epoch": 1.94,
|
| 490 |
+
"learning_rate": 4.474548440065682e-05,
|
| 491 |
+
"loss": 1.2428,
|
| 492 |
"step": 790
|
| 493 |
},
|
| 494 |
{
|
| 495 |
"epoch": 1.97,
|
| 496 |
+
"learning_rate": 4.460864805692392e-05,
|
| 497 |
+
"loss": 1.2961,
|
| 498 |
"step": 800
|
| 499 |
},
|
| 500 |
{
|
| 501 |
"epoch": 1.99,
|
| 502 |
+
"learning_rate": 4.447181171319103e-05,
|
| 503 |
+
"loss": 1.3126,
|
| 504 |
"step": 810
|
| 505 |
},
|
| 506 |
{
|
| 507 |
"epoch": 2.0,
|
| 508 |
+
"eval_accuracy": 0.8399308556611927,
|
| 509 |
+
"eval_loss": 0.7987398505210876,
|
| 510 |
+
"eval_runtime": 27.7317,
|
| 511 |
+
"eval_samples_per_second": 208.606,
|
| 512 |
+
"eval_steps_per_second": 6.527,
|
| 513 |
"step": 813
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"epoch": 2.02,
|
| 517 |
+
"learning_rate": 4.433497536945813e-05,
|
| 518 |
+
"loss": 1.271,
|
| 519 |
"step": 820
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 2.04,
|
| 523 |
+
"learning_rate": 4.419813902572523e-05,
|
| 524 |
+
"loss": 1.3539,
|
| 525 |
"step": 830
|
| 526 |
},
|
| 527 |
{
|
| 528 |
"epoch": 2.07,
|
| 529 |
+
"learning_rate": 4.406130268199234e-05,
|
| 530 |
+
"loss": 1.2938,
|
| 531 |
"step": 840
|
| 532 |
},
|
| 533 |
{
|
| 534 |
"epoch": 2.09,
|
| 535 |
+
"learning_rate": 4.3924466338259446e-05,
|
| 536 |
+
"loss": 1.2817,
|
| 537 |
"step": 850
|
| 538 |
},
|
| 539 |
{
|
| 540 |
"epoch": 2.11,
|
| 541 |
+
"learning_rate": 4.3787629994526544e-05,
|
| 542 |
+
"loss": 1.2808,
|
| 543 |
"step": 860
|
| 544 |
},
|
| 545 |
{
|
| 546 |
"epoch": 2.14,
|
| 547 |
+
"learning_rate": 4.3650793650793655e-05,
|
| 548 |
+
"loss": 1.2517,
|
| 549 |
"step": 870
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 2.16,
|
| 553 |
+
"learning_rate": 4.351395730706076e-05,
|
| 554 |
+
"loss": 1.2371,
|
| 555 |
"step": 880
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"epoch": 2.19,
|
| 559 |
+
"learning_rate": 4.3377120963327864e-05,
|
| 560 |
+
"loss": 1.277,
|
| 561 |
"step": 890
|
| 562 |
},
|
| 563 |
{
|
| 564 |
"epoch": 2.21,
|
| 565 |
+
"learning_rate": 4.324028461959497e-05,
|
| 566 |
+
"loss": 1.2152,
|
| 567 |
"step": 900
|
| 568 |
},
|
| 569 |
{
|
| 570 |
"epoch": 2.24,
|
| 571 |
+
"learning_rate": 4.3103448275862066e-05,
|
| 572 |
+
"loss": 1.1733,
|
| 573 |
"step": 910
|
| 574 |
},
|
| 575 |
{
|
| 576 |
"epoch": 2.26,
|
| 577 |
+
"learning_rate": 4.296661193212918e-05,
|
| 578 |
+
"loss": 1.3563,
|
| 579 |
"step": 920
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 2.29,
|
| 583 |
+
"learning_rate": 4.282977558839628e-05,
|
| 584 |
+
"loss": 1.4114,
|
| 585 |
"step": 930
|
| 586 |
},
|
| 587 |
{
|
| 588 |
"epoch": 2.31,
|
| 589 |
+
"learning_rate": 4.2692939244663386e-05,
|
| 590 |
+
"loss": 1.1794,
|
| 591 |
"step": 940
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 2.34,
|
| 595 |
+
"learning_rate": 4.255610290093049e-05,
|
| 596 |
+
"loss": 1.2454,
|
| 597 |
"step": 950
|
| 598 |
},
|
| 599 |
{
|
| 600 |
"epoch": 2.36,
|
| 601 |
+
"learning_rate": 4.2419266557197594e-05,
|
| 602 |
+
"loss": 1.2245,
|
| 603 |
"step": 960
|
| 604 |
},
|
| 605 |
{
|
| 606 |
"epoch": 2.38,
|
| 607 |
+
"learning_rate": 4.22824302134647e-05,
|
| 608 |
+
"loss": 1.2197,
|
| 609 |
"step": 970
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 2.41,
|
| 613 |
+
"learning_rate": 4.21455938697318e-05,
|
| 614 |
+
"loss": 1.3025,
|
| 615 |
"step": 980
|
| 616 |
},
|
| 617 |
{
|
| 618 |
"epoch": 2.43,
|
| 619 |
+
"learning_rate": 4.200875752599891e-05,
|
| 620 |
+
"loss": 1.1717,
|
| 621 |
"step": 990
|
| 622 |
},
|
| 623 |
{
|
| 624 |
"epoch": 2.46,
|
| 625 |
+
"learning_rate": 4.187192118226601e-05,
|
| 626 |
+
"loss": 1.2077,
|
| 627 |
"step": 1000
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"epoch": 2.48,
|
| 631 |
+
"learning_rate": 4.1735084838533116e-05,
|
| 632 |
+
"loss": 1.3137,
|
| 633 |
"step": 1010
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"epoch": 2.51,
|
| 637 |
+
"learning_rate": 4.159824849480022e-05,
|
| 638 |
+
"loss": 1.2348,
|
| 639 |
"step": 1020
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 2.53,
|
| 643 |
+
"learning_rate": 4.1461412151067325e-05,
|
| 644 |
+
"loss": 1.2611,
|
| 645 |
"step": 1030
|
| 646 |
},
|
| 647 |
{
|
| 648 |
"epoch": 2.56,
|
| 649 |
+
"learning_rate": 4.132457580733443e-05,
|
| 650 |
+
"loss": 1.0912,
|
| 651 |
"step": 1040
|
| 652 |
},
|
| 653 |
{
|
| 654 |
"epoch": 2.58,
|
| 655 |
+
"learning_rate": 4.1187739463601534e-05,
|
| 656 |
+
"loss": 1.3145,
|
| 657 |
"step": 1050
|
| 658 |
},
|
| 659 |
{
|
| 660 |
"epoch": 2.61,
|
| 661 |
+
"learning_rate": 4.105090311986864e-05,
|
| 662 |
+
"loss": 1.2088,
|
| 663 |
"step": 1060
|
| 664 |
},
|
| 665 |
{
|
| 666 |
"epoch": 2.63,
|
| 667 |
+
"learning_rate": 4.091406677613574e-05,
|
| 668 |
+
"loss": 1.1283,
|
| 669 |
"step": 1070
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 2.66,
|
| 673 |
+
"learning_rate": 4.077723043240285e-05,
|
| 674 |
+
"loss": 1.1388,
|
| 675 |
"step": 1080
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"epoch": 2.68,
|
| 679 |
+
"learning_rate": 4.064039408866995e-05,
|
| 680 |
+
"loss": 1.2748,
|
| 681 |
"step": 1090
|
| 682 |
},
|
| 683 |
{
|
| 684 |
"epoch": 2.7,
|
| 685 |
+
"learning_rate": 4.050355774493706e-05,
|
| 686 |
+
"loss": 1.0486,
|
| 687 |
"step": 1100
|
| 688 |
},
|
| 689 |
{
|
| 690 |
"epoch": 2.73,
|
| 691 |
+
"learning_rate": 4.036672140120416e-05,
|
| 692 |
+
"loss": 1.1514,
|
| 693 |
"step": 1110
|
| 694 |
},
|
| 695 |
{
|
| 696 |
"epoch": 2.75,
|
| 697 |
+
"learning_rate": 4.0229885057471265e-05,
|
| 698 |
+
"loss": 1.2419,
|
| 699 |
"step": 1120
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 2.78,
|
| 703 |
+
"learning_rate": 4.009304871373837e-05,
|
| 704 |
+
"loss": 1.3025,
|
| 705 |
"step": 1130
|
| 706 |
},
|
| 707 |
{
|
| 708 |
"epoch": 2.8,
|
| 709 |
+
"learning_rate": 3.9956212370005474e-05,
|
| 710 |
+
"loss": 1.1627,
|
| 711 |
"step": 1140
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"epoch": 2.83,
|
| 715 |
+
"learning_rate": 3.981937602627258e-05,
|
| 716 |
+
"loss": 1.2263,
|
| 717 |
"step": 1150
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"epoch": 2.85,
|
| 721 |
+
"learning_rate": 3.968253968253968e-05,
|
| 722 |
+
"loss": 1.1918,
|
| 723 |
"step": 1160
|
| 724 |
},
|
| 725 |
{
|
| 726 |
"epoch": 2.88,
|
| 727 |
+
"learning_rate": 3.954570333880679e-05,
|
| 728 |
+
"loss": 1.2337,
|
| 729 |
"step": 1170
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 2.9,
|
| 733 |
+
"learning_rate": 3.94088669950739e-05,
|
| 734 |
+
"loss": 1.325,
|
| 735 |
"step": 1180
|
| 736 |
},
|
| 737 |
{
|
| 738 |
"epoch": 2.93,
|
| 739 |
+
"learning_rate": 3.9272030651340996e-05,
|
| 740 |
+
"loss": 1.1631,
|
| 741 |
"step": 1190
|
| 742 |
},
|
| 743 |
{
|
| 744 |
"epoch": 2.95,
|
| 745 |
+
"learning_rate": 3.91351943076081e-05,
|
| 746 |
+
"loss": 1.2506,
|
| 747 |
"step": 1200
|
| 748 |
},
|
| 749 |
{
|
| 750 |
"epoch": 2.97,
|
| 751 |
+
"learning_rate": 3.899835796387521e-05,
|
| 752 |
+
"loss": 1.2831,
|
| 753 |
"step": 1210
|
| 754 |
},
|
| 755 |
{
|
| 756 |
+
"epoch": 3.0,
|
| 757 |
+
"learning_rate": 3.886152162014231e-05,
|
| 758 |
+
"loss": 1.2335,
|
| 759 |
+
"step": 1220
|
|
|
|
|
|
|
|
|
|
| 760 |
},
|
| 761 |
{
|
| 762 |
+
"epoch": 3.0,
|
| 763 |
+
"eval_accuracy": 0.8741573033707866,
|
| 764 |
+
"eval_loss": 0.6398221254348755,
|
| 765 |
+
"eval_runtime": 27.2608,
|
| 766 |
+
"eval_samples_per_second": 212.209,
|
| 767 |
+
"eval_steps_per_second": 6.64,
|
| 768 |
+
"step": 1220
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.02,
|
| 772 |
+
"learning_rate": 3.872468527640942e-05,
|
| 773 |
+
"loss": 1.1144,
|
| 774 |
+
"step": 1230
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 3.05,
|
| 778 |
+
"learning_rate": 3.858784893267652e-05,
|
| 779 |
+
"loss": 1.1746,
|
| 780 |
+
"step": 1240
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 3.07,
|
| 784 |
+
"learning_rate": 3.845101258894362e-05,
|
| 785 |
+
"loss": 1.1363,
|
| 786 |
+
"step": 1250
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 3.1,
|
| 790 |
+
"learning_rate": 3.831417624521073e-05,
|
| 791 |
+
"loss": 1.1003,
|
| 792 |
+
"step": 1260
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 3.12,
|
| 796 |
+
"learning_rate": 3.817733990147783e-05,
|
| 797 |
+
"loss": 1.1743,
|
| 798 |
+
"step": 1270
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 3.15,
|
| 802 |
+
"learning_rate": 3.8040503557744935e-05,
|
| 803 |
+
"loss": 1.1832,
|
| 804 |
+
"step": 1280
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.17,
|
| 808 |
+
"learning_rate": 3.7903667214012047e-05,
|
| 809 |
+
"loss": 1.1495,
|
| 810 |
+
"step": 1290
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 3.2,
|
| 814 |
+
"learning_rate": 3.7766830870279144e-05,
|
| 815 |
+
"loss": 1.1282,
|
| 816 |
+
"step": 1300
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 3.22,
|
| 820 |
+
"learning_rate": 3.7629994526546255e-05,
|
| 821 |
+
"loss": 1.1273,
|
| 822 |
+
"step": 1310
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 3.25,
|
| 826 |
+
"learning_rate": 3.749315818281336e-05,
|
| 827 |
+
"loss": 1.0586,
|
| 828 |
+
"step": 1320
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 3.27,
|
| 832 |
+
"learning_rate": 3.735632183908046e-05,
|
| 833 |
+
"loss": 1.1353,
|
| 834 |
+
"step": 1330
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 3.29,
|
| 838 |
+
"learning_rate": 3.721948549534757e-05,
|
| 839 |
+
"loss": 1.0796,
|
| 840 |
+
"step": 1340
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 3.32,
|
| 844 |
+
"learning_rate": 3.7082649151614666e-05,
|
| 845 |
+
"loss": 1.1132,
|
| 846 |
+
"step": 1350
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 3.34,
|
| 850 |
+
"learning_rate": 3.694581280788178e-05,
|
| 851 |
+
"loss": 1.0474,
|
| 852 |
+
"step": 1360
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 3.37,
|
| 856 |
+
"learning_rate": 3.680897646414888e-05,
|
| 857 |
+
"loss": 1.1392,
|
| 858 |
+
"step": 1370
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 3.39,
|
| 862 |
+
"learning_rate": 3.667214012041598e-05,
|
| 863 |
+
"loss": 1.1334,
|
| 864 |
+
"step": 1380
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 3.42,
|
| 868 |
+
"learning_rate": 3.653530377668309e-05,
|
| 869 |
+
"loss": 1.0506,
|
| 870 |
+
"step": 1390
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 3.44,
|
| 874 |
+
"learning_rate": 3.6398467432950195e-05,
|
| 875 |
+
"loss": 1.1107,
|
| 876 |
+
"step": 1400
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 3.47,
|
| 880 |
+
"learning_rate": 3.62616310892173e-05,
|
| 881 |
+
"loss": 1.1643,
|
| 882 |
+
"step": 1410
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 3.49,
|
| 886 |
+
"learning_rate": 3.6124794745484404e-05,
|
| 887 |
+
"loss": 1.0167,
|
| 888 |
+
"step": 1420
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 3.52,
|
| 892 |
+
"learning_rate": 3.598795840175151e-05,
|
| 893 |
+
"loss": 1.2377,
|
| 894 |
+
"step": 1430
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 3.54,
|
| 898 |
+
"learning_rate": 3.585112205801861e-05,
|
| 899 |
+
"loss": 1.0392,
|
| 900 |
+
"step": 1440
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 3.56,
|
| 904 |
+
"learning_rate": 3.571428571428572e-05,
|
| 905 |
+
"loss": 1.1304,
|
| 906 |
+
"step": 1450
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 3.59,
|
| 910 |
+
"learning_rate": 3.5577449370552815e-05,
|
| 911 |
+
"loss": 1.2185,
|
| 912 |
+
"step": 1460
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 3.61,
|
| 916 |
+
"learning_rate": 3.5440613026819926e-05,
|
| 917 |
+
"loss": 1.0848,
|
| 918 |
+
"step": 1470
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 3.64,
|
| 922 |
+
"learning_rate": 3.530377668308703e-05,
|
| 923 |
+
"loss": 1.1142,
|
| 924 |
+
"step": 1480
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 3.66,
|
| 928 |
+
"learning_rate": 3.5166940339354135e-05,
|
| 929 |
+
"loss": 1.0458,
|
| 930 |
+
"step": 1490
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 3.69,
|
| 934 |
+
"learning_rate": 3.503010399562124e-05,
|
| 935 |
+
"loss": 1.041,
|
| 936 |
+
"step": 1500
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 3.71,
|
| 940 |
+
"learning_rate": 3.489326765188834e-05,
|
| 941 |
+
"loss": 1.119,
|
| 942 |
+
"step": 1510
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 3.74,
|
| 946 |
+
"learning_rate": 3.475643130815545e-05,
|
| 947 |
+
"loss": 1.0259,
|
| 948 |
+
"step": 1520
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 3.76,
|
| 952 |
+
"learning_rate": 3.461959496442255e-05,
|
| 953 |
+
"loss": 1.14,
|
| 954 |
+
"step": 1530
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 3.79,
|
| 958 |
+
"learning_rate": 3.4482758620689657e-05,
|
| 959 |
+
"loss": 0.994,
|
| 960 |
+
"step": 1540
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 3.81,
|
| 964 |
+
"learning_rate": 3.434592227695676e-05,
|
| 965 |
+
"loss": 1.011,
|
| 966 |
+
"step": 1550
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 3.84,
|
| 970 |
+
"learning_rate": 3.4209085933223865e-05,
|
| 971 |
+
"loss": 1.0745,
|
| 972 |
+
"step": 1560
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 3.86,
|
| 976 |
+
"learning_rate": 3.407224958949097e-05,
|
| 977 |
+
"loss": 1.0777,
|
| 978 |
+
"step": 1570
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 3.88,
|
| 982 |
+
"learning_rate": 3.3935413245758074e-05,
|
| 983 |
+
"loss": 1.1397,
|
| 984 |
+
"step": 1580
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 3.91,
|
| 988 |
+
"learning_rate": 3.379857690202518e-05,
|
| 989 |
+
"loss": 1.0727,
|
| 990 |
+
"step": 1590
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 3.93,
|
| 994 |
+
"learning_rate": 3.366174055829228e-05,
|
| 995 |
+
"loss": 1.0662,
|
| 996 |
+
"step": 1600
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 3.96,
|
| 1000 |
+
"learning_rate": 3.352490421455939e-05,
|
| 1001 |
+
"loss": 1.0208,
|
| 1002 |
+
"step": 1610
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 3.98,
|
| 1006 |
+
"learning_rate": 3.338806787082649e-05,
|
| 1007 |
+
"loss": 1.1244,
|
| 1008 |
+
"step": 1620
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 4.0,
|
| 1012 |
+
"eval_accuracy": 0.9040622299049266,
|
| 1013 |
+
"eval_loss": 0.497685045003891,
|
| 1014 |
+
"eval_runtime": 27.8668,
|
| 1015 |
+
"eval_samples_per_second": 207.595,
|
| 1016 |
+
"eval_steps_per_second": 6.495,
|
| 1017 |
+
"step": 1627
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 4.01,
|
| 1021 |
+
"learning_rate": 3.3251231527093596e-05,
|
| 1022 |
+
"loss": 0.9674,
|
| 1023 |
+
"step": 1630
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 4.03,
|
| 1027 |
+
"learning_rate": 3.31143951833607e-05,
|
| 1028 |
+
"loss": 1.0643,
|
| 1029 |
+
"step": 1640
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 4.06,
|
| 1033 |
+
"learning_rate": 3.297755883962781e-05,
|
| 1034 |
+
"loss": 1.0174,
|
| 1035 |
+
"step": 1650
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 4.08,
|
| 1039 |
+
"learning_rate": 3.284072249589491e-05,
|
| 1040 |
+
"loss": 1.0732,
|
| 1041 |
+
"step": 1660
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 4.11,
|
| 1045 |
+
"learning_rate": 3.2703886152162014e-05,
|
| 1046 |
+
"loss": 1.0587,
|
| 1047 |
+
"step": 1670
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 4.13,
|
| 1051 |
+
"learning_rate": 3.256704980842912e-05,
|
| 1052 |
+
"loss": 1.0531,
|
| 1053 |
+
"step": 1680
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 4.15,
|
| 1057 |
+
"learning_rate": 3.243021346469622e-05,
|
| 1058 |
+
"loss": 1.0491,
|
| 1059 |
+
"step": 1690
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 4.18,
|
| 1063 |
+
"learning_rate": 3.2293377120963334e-05,
|
| 1064 |
+
"loss": 0.9433,
|
| 1065 |
+
"step": 1700
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 4.2,
|
| 1069 |
+
"learning_rate": 3.215654077723043e-05,
|
| 1070 |
+
"loss": 0.9851,
|
| 1071 |
+
"step": 1710
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 4.23,
|
| 1075 |
+
"learning_rate": 3.2019704433497536e-05,
|
| 1076 |
+
"loss": 1.0559,
|
| 1077 |
+
"step": 1720
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 4.25,
|
| 1081 |
+
"learning_rate": 3.188286808976465e-05,
|
| 1082 |
+
"loss": 0.9604,
|
| 1083 |
+
"step": 1730
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 4.28,
|
| 1087 |
+
"learning_rate": 3.1746031746031745e-05,
|
| 1088 |
+
"loss": 0.9938,
|
| 1089 |
+
"step": 1740
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.3,
|
| 1093 |
+
"learning_rate": 3.160919540229885e-05,
|
| 1094 |
+
"loss": 1.0587,
|
| 1095 |
+
"step": 1750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 4.33,
|
| 1099 |
+
"learning_rate": 3.147235905856596e-05,
|
| 1100 |
+
"loss": 0.9771,
|
| 1101 |
+
"step": 1760
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 4.35,
|
| 1105 |
+
"learning_rate": 3.133552271483306e-05,
|
| 1106 |
+
"loss": 1.0232,
|
| 1107 |
+
"step": 1770
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 4.38,
|
| 1111 |
+
"learning_rate": 3.119868637110017e-05,
|
| 1112 |
+
"loss": 1.014,
|
| 1113 |
+
"step": 1780
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 4.4,
|
| 1117 |
+
"learning_rate": 3.1061850027367273e-05,
|
| 1118 |
+
"loss": 0.9797,
|
| 1119 |
+
"step": 1790
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 4.43,
|
| 1123 |
+
"learning_rate": 3.092501368363437e-05,
|
| 1124 |
+
"loss": 0.9898,
|
| 1125 |
+
"step": 1800
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 4.45,
|
| 1129 |
+
"learning_rate": 3.078817733990148e-05,
|
| 1130 |
+
"loss": 1.0044,
|
| 1131 |
+
"step": 1810
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 4.47,
|
| 1135 |
+
"learning_rate": 3.065134099616858e-05,
|
| 1136 |
+
"loss": 1.0723,
|
| 1137 |
+
"step": 1820
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 4.5,
|
| 1141 |
+
"learning_rate": 3.0514504652435688e-05,
|
| 1142 |
+
"loss": 1.0545,
|
| 1143 |
+
"step": 1830
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 4.52,
|
| 1147 |
+
"learning_rate": 3.0377668308702795e-05,
|
| 1148 |
+
"loss": 0.9979,
|
| 1149 |
+
"step": 1840
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 4.55,
|
| 1153 |
+
"learning_rate": 3.0240831964969896e-05,
|
| 1154 |
+
"loss": 0.9971,
|
| 1155 |
+
"step": 1850
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 4.57,
|
| 1159 |
+
"learning_rate": 3.0103995621237e-05,
|
| 1160 |
+
"loss": 1.0341,
|
| 1161 |
+
"step": 1860
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 4.6,
|
| 1165 |
+
"learning_rate": 2.996715927750411e-05,
|
| 1166 |
+
"loss": 0.9853,
|
| 1167 |
+
"step": 1870
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 4.62,
|
| 1171 |
+
"learning_rate": 2.983032293377121e-05,
|
| 1172 |
+
"loss": 1.0434,
|
| 1173 |
+
"step": 1880
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 4.65,
|
| 1177 |
+
"learning_rate": 2.9693486590038317e-05,
|
| 1178 |
+
"loss": 0.94,
|
| 1179 |
+
"step": 1890
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 4.67,
|
| 1183 |
+
"learning_rate": 2.9556650246305422e-05,
|
| 1184 |
+
"loss": 1.0002,
|
| 1185 |
+
"step": 1900
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 4.7,
|
| 1189 |
+
"learning_rate": 2.9419813902572523e-05,
|
| 1190 |
+
"loss": 0.9252,
|
| 1191 |
+
"step": 1910
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 4.72,
|
| 1195 |
+
"learning_rate": 2.928297755883963e-05,
|
| 1196 |
+
"loss": 0.8899,
|
| 1197 |
+
"step": 1920
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 4.74,
|
| 1201 |
+
"learning_rate": 2.914614121510673e-05,
|
| 1202 |
+
"loss": 1.0344,
|
| 1203 |
+
"step": 1930
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 4.77,
|
| 1207 |
+
"learning_rate": 2.900930487137384e-05,
|
| 1208 |
+
"loss": 1.008,
|
| 1209 |
+
"step": 1940
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 4.79,
|
| 1213 |
+
"learning_rate": 2.8872468527640944e-05,
|
| 1214 |
+
"loss": 0.9935,
|
| 1215 |
+
"step": 1950
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 4.82,
|
| 1219 |
+
"learning_rate": 2.8735632183908045e-05,
|
| 1220 |
+
"loss": 1.0432,
|
| 1221 |
+
"step": 1960
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 4.84,
|
| 1225 |
+
"learning_rate": 2.8598795840175153e-05,
|
| 1226 |
+
"loss": 0.9235,
|
| 1227 |
+
"step": 1970
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 4.87,
|
| 1231 |
+
"learning_rate": 2.8461959496442257e-05,
|
| 1232 |
+
"loss": 0.9564,
|
| 1233 |
+
"step": 1980
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 4.89,
|
| 1237 |
+
"learning_rate": 2.8325123152709358e-05,
|
| 1238 |
+
"loss": 0.9873,
|
| 1239 |
+
"step": 1990
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 4.92,
|
| 1243 |
+
"learning_rate": 2.8188286808976466e-05,
|
| 1244 |
+
"loss": 1.0216,
|
| 1245 |
+
"step": 2000
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 4.94,
|
| 1249 |
+
"learning_rate": 2.8051450465243574e-05,
|
| 1250 |
+
"loss": 0.9,
|
| 1251 |
+
"step": 2010
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 4.97,
|
| 1255 |
+
"learning_rate": 2.7914614121510675e-05,
|
| 1256 |
+
"loss": 0.9814,
|
| 1257 |
+
"step": 2020
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 4.99,
|
| 1261 |
+
"learning_rate": 2.777777777777778e-05,
|
| 1262 |
+
"loss": 0.9785,
|
| 1263 |
+
"step": 2030
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 5.0,
|
| 1267 |
+
"eval_accuracy": 0.9139152981849611,
|
| 1268 |
+
"eval_loss": 0.4472927749156952,
|
| 1269 |
+
"eval_runtime": 27.773,
|
| 1270 |
+
"eval_samples_per_second": 208.296,
|
| 1271 |
+
"eval_steps_per_second": 6.517,
|
| 1272 |
+
"step": 2033
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 5.02,
|
| 1276 |
+
"learning_rate": 2.764094143404488e-05,
|
| 1277 |
+
"loss": 0.97,
|
| 1278 |
+
"step": 2040
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 5.04,
|
| 1282 |
+
"learning_rate": 2.7504105090311988e-05,
|
| 1283 |
+
"loss": 0.9339,
|
| 1284 |
+
"step": 2050
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 5.06,
|
| 1288 |
+
"learning_rate": 2.7367268746579096e-05,
|
| 1289 |
+
"loss": 0.8636,
|
| 1290 |
+
"step": 2060
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 5.09,
|
| 1294 |
+
"learning_rate": 2.7230432402846197e-05,
|
| 1295 |
+
"loss": 0.8525,
|
| 1296 |
+
"step": 2070
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 5.11,
|
| 1300 |
+
"learning_rate": 2.70935960591133e-05,
|
| 1301 |
+
"loss": 0.9281,
|
| 1302 |
+
"step": 2080
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 5.14,
|
| 1306 |
+
"learning_rate": 2.695675971538041e-05,
|
| 1307 |
+
"loss": 0.9573,
|
| 1308 |
+
"step": 2090
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 5.16,
|
| 1312 |
+
"learning_rate": 2.681992337164751e-05,
|
| 1313 |
+
"loss": 1.0129,
|
| 1314 |
+
"step": 2100
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 5.19,
|
| 1318 |
+
"learning_rate": 2.6683087027914618e-05,
|
| 1319 |
+
"loss": 0.9568,
|
| 1320 |
+
"step": 2110
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 5.21,
|
| 1324 |
+
"learning_rate": 2.6546250684181722e-05,
|
| 1325 |
+
"loss": 0.8739,
|
| 1326 |
+
"step": 2120
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 5.24,
|
| 1330 |
+
"learning_rate": 2.6409414340448823e-05,
|
| 1331 |
+
"loss": 1.0142,
|
| 1332 |
+
"step": 2130
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 5.26,
|
| 1336 |
+
"learning_rate": 2.627257799671593e-05,
|
| 1337 |
+
"loss": 0.8615,
|
| 1338 |
+
"step": 2140
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 5.29,
|
| 1342 |
+
"learning_rate": 2.6135741652983032e-05,
|
| 1343 |
+
"loss": 0.9213,
|
| 1344 |
+
"step": 2150
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.31,
|
| 1348 |
+
"learning_rate": 2.5998905309250136e-05,
|
| 1349 |
+
"loss": 0.9751,
|
| 1350 |
+
"step": 2160
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 5.33,
|
| 1354 |
+
"learning_rate": 2.5862068965517244e-05,
|
| 1355 |
+
"loss": 0.9144,
|
| 1356 |
+
"step": 2170
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 5.36,
|
| 1360 |
+
"learning_rate": 2.5725232621784345e-05,
|
| 1361 |
+
"loss": 0.8717,
|
| 1362 |
+
"step": 2180
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 5.38,
|
| 1366 |
+
"learning_rate": 2.5588396278051453e-05,
|
| 1367 |
+
"loss": 1.0529,
|
| 1368 |
+
"step": 2190
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 5.41,
|
| 1372 |
+
"learning_rate": 2.5451559934318557e-05,
|
| 1373 |
+
"loss": 0.9158,
|
| 1374 |
+
"step": 2200
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 5.43,
|
| 1378 |
+
"learning_rate": 2.531472359058566e-05,
|
| 1379 |
+
"loss": 0.9584,
|
| 1380 |
+
"step": 2210
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 5.46,
|
| 1384 |
+
"learning_rate": 2.5177887246852766e-05,
|
| 1385 |
+
"loss": 0.9806,
|
| 1386 |
+
"step": 2220
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.48,
|
| 1390 |
+
"learning_rate": 2.5041050903119874e-05,
|
| 1391 |
+
"loss": 0.8941,
|
| 1392 |
+
"step": 2230
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 5.51,
|
| 1396 |
+
"learning_rate": 2.4904214559386975e-05,
|
| 1397 |
+
"loss": 0.9625,
|
| 1398 |
+
"step": 2240
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 5.53,
|
| 1402 |
+
"learning_rate": 2.476737821565408e-05,
|
| 1403 |
+
"loss": 0.9277,
|
| 1404 |
+
"step": 2250
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 5.56,
|
| 1408 |
+
"learning_rate": 2.4630541871921184e-05,
|
| 1409 |
+
"loss": 0.9645,
|
| 1410 |
+
"step": 2260
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 5.58,
|
| 1414 |
+
"learning_rate": 2.4493705528188288e-05,
|
| 1415 |
+
"loss": 0.9162,
|
| 1416 |
+
"step": 2270
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 5.61,
|
| 1420 |
+
"learning_rate": 2.4356869184455393e-05,
|
| 1421 |
+
"loss": 0.8772,
|
| 1422 |
+
"step": 2280
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 5.63,
|
| 1426 |
+
"learning_rate": 2.4220032840722497e-05,
|
| 1427 |
+
"loss": 0.9556,
|
| 1428 |
+
"step": 2290
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 5.65,
|
| 1432 |
+
"learning_rate": 2.40831964969896e-05,
|
| 1433 |
+
"loss": 0.885,
|
| 1434 |
+
"step": 2300
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 5.68,
|
| 1438 |
+
"learning_rate": 2.3946360153256706e-05,
|
| 1439 |
+
"loss": 0.7871,
|
| 1440 |
+
"step": 2310
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 5.7,
|
| 1444 |
+
"learning_rate": 2.380952380952381e-05,
|
| 1445 |
+
"loss": 0.8639,
|
| 1446 |
+
"step": 2320
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 5.73,
|
| 1450 |
+
"learning_rate": 2.3672687465790915e-05,
|
| 1451 |
+
"loss": 0.9446,
|
| 1452 |
+
"step": 2330
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 5.75,
|
| 1456 |
+
"learning_rate": 2.353585112205802e-05,
|
| 1457 |
+
"loss": 0.8905,
|
| 1458 |
+
"step": 2340
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 5.78,
|
| 1462 |
+
"learning_rate": 2.3399014778325123e-05,
|
| 1463 |
+
"loss": 0.9502,
|
| 1464 |
+
"step": 2350
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 5.8,
|
| 1468 |
+
"learning_rate": 2.326217843459223e-05,
|
| 1469 |
+
"loss": 0.9842,
|
| 1470 |
+
"step": 2360
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 5.83,
|
| 1474 |
+
"learning_rate": 2.3125342090859332e-05,
|
| 1475 |
+
"loss": 0.8827,
|
| 1476 |
+
"step": 2370
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 5.85,
|
| 1480 |
+
"learning_rate": 2.2988505747126437e-05,
|
| 1481 |
+
"loss": 0.8861,
|
| 1482 |
+
"step": 2380
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 5.88,
|
| 1486 |
+
"learning_rate": 2.285166940339354e-05,
|
| 1487 |
+
"loss": 0.9701,
|
| 1488 |
+
"step": 2390
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 5.9,
|
| 1492 |
+
"learning_rate": 2.271483305966065e-05,
|
| 1493 |
+
"loss": 0.8456,
|
| 1494 |
+
"step": 2400
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 5.93,
|
| 1498 |
+
"learning_rate": 2.2577996715927753e-05,
|
| 1499 |
+
"loss": 0.8858,
|
| 1500 |
+
"step": 2410
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 5.95,
|
| 1504 |
+
"learning_rate": 2.2441160372194854e-05,
|
| 1505 |
+
"loss": 0.8431,
|
| 1506 |
+
"step": 2420
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 5.97,
|
| 1510 |
+
"learning_rate": 2.230432402846196e-05,
|
| 1511 |
+
"loss": 0.9466,
|
| 1512 |
+
"step": 2430
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 6.0,
|
| 1516 |
+
"learning_rate": 2.2167487684729066e-05,
|
| 1517 |
+
"loss": 0.9625,
|
| 1518 |
+
"step": 2440
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 6.0,
|
| 1522 |
+
"eval_accuracy": 0.9199654278305963,
|
| 1523 |
+
"eval_loss": 0.3929709494113922,
|
| 1524 |
+
"eval_runtime": 27.3929,
|
| 1525 |
+
"eval_samples_per_second": 211.186,
|
| 1526 |
+
"eval_steps_per_second": 6.608,
|
| 1527 |
+
"step": 2440
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 6.02,
|
| 1531 |
+
"learning_rate": 2.203065134099617e-05,
|
| 1532 |
+
"loss": 0.886,
|
| 1533 |
+
"step": 2450
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 6.05,
|
| 1537 |
+
"learning_rate": 2.1893814997263272e-05,
|
| 1538 |
+
"loss": 0.8862,
|
| 1539 |
+
"step": 2460
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 6.07,
|
| 1543 |
+
"learning_rate": 2.175697865353038e-05,
|
| 1544 |
+
"loss": 0.8938,
|
| 1545 |
+
"step": 2470
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 6.1,
|
| 1549 |
+
"learning_rate": 2.1620142309797484e-05,
|
| 1550 |
+
"loss": 0.8165,
|
| 1551 |
+
"step": 2480
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 6.12,
|
| 1555 |
+
"learning_rate": 2.148330596606459e-05,
|
| 1556 |
+
"loss": 0.8117,
|
| 1557 |
+
"step": 2490
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 6.15,
|
| 1561 |
+
"learning_rate": 2.1346469622331693e-05,
|
| 1562 |
+
"loss": 0.8538,
|
| 1563 |
+
"step": 2500
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 6.17,
|
| 1567 |
+
"learning_rate": 2.1209633278598797e-05,
|
| 1568 |
+
"loss": 0.8646,
|
| 1569 |
+
"step": 2510
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 6.2,
|
| 1573 |
+
"learning_rate": 2.10727969348659e-05,
|
| 1574 |
+
"loss": 0.8506,
|
| 1575 |
+
"step": 2520
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 6.22,
|
| 1579 |
+
"learning_rate": 2.0935960591133006e-05,
|
| 1580 |
+
"loss": 0.9332,
|
| 1581 |
+
"step": 2530
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 6.24,
|
| 1585 |
+
"learning_rate": 2.079912424740011e-05,
|
| 1586 |
+
"loss": 0.7804,
|
| 1587 |
+
"step": 2540
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 6.27,
|
| 1591 |
+
"learning_rate": 2.0662287903667215e-05,
|
| 1592 |
+
"loss": 0.8503,
|
| 1593 |
+
"step": 2550
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 6.29,
|
| 1597 |
+
"learning_rate": 2.052545155993432e-05,
|
| 1598 |
+
"loss": 0.8818,
|
| 1599 |
+
"step": 2560
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 6.32,
|
| 1603 |
+
"learning_rate": 2.0388615216201424e-05,
|
| 1604 |
+
"loss": 0.7721,
|
| 1605 |
+
"step": 2570
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 6.34,
|
| 1609 |
+
"learning_rate": 2.025177887246853e-05,
|
| 1610 |
+
"loss": 0.8895,
|
| 1611 |
+
"step": 2580
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 6.37,
|
| 1615 |
+
"learning_rate": 2.0114942528735632e-05,
|
| 1616 |
+
"loss": 0.8575,
|
| 1617 |
+
"step": 2590
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 6.39,
|
| 1621 |
+
"learning_rate": 1.9978106185002737e-05,
|
| 1622 |
+
"loss": 0.8596,
|
| 1623 |
+
"step": 2600
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 6.42,
|
| 1627 |
+
"learning_rate": 1.984126984126984e-05,
|
| 1628 |
+
"loss": 0.8394,
|
| 1629 |
+
"step": 2610
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 6.44,
|
| 1633 |
+
"learning_rate": 1.970443349753695e-05,
|
| 1634 |
+
"loss": 0.8782,
|
| 1635 |
+
"step": 2620
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 6.47,
|
| 1639 |
+
"learning_rate": 1.956759715380405e-05,
|
| 1640 |
+
"loss": 0.7369,
|
| 1641 |
+
"step": 2630
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 6.49,
|
| 1645 |
+
"learning_rate": 1.9430760810071154e-05,
|
| 1646 |
+
"loss": 0.8684,
|
| 1647 |
+
"step": 2640
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 6.52,
|
| 1651 |
+
"learning_rate": 1.929392446633826e-05,
|
| 1652 |
+
"loss": 0.8494,
|
| 1653 |
+
"step": 2650
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 6.54,
|
| 1657 |
+
"learning_rate": 1.9157088122605367e-05,
|
| 1658 |
+
"loss": 0.764,
|
| 1659 |
+
"step": 2660
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 6.56,
|
| 1663 |
+
"learning_rate": 1.9020251778872468e-05,
|
| 1664 |
+
"loss": 0.855,
|
| 1665 |
+
"step": 2670
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 6.59,
|
| 1669 |
+
"learning_rate": 1.8883415435139572e-05,
|
| 1670 |
+
"loss": 0.8766,
|
| 1671 |
+
"step": 2680
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 6.61,
|
| 1675 |
+
"learning_rate": 1.874657909140668e-05,
|
| 1676 |
+
"loss": 0.8737,
|
| 1677 |
+
"step": 2690
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 6.64,
|
| 1681 |
+
"learning_rate": 1.8609742747673784e-05,
|
| 1682 |
+
"loss": 0.8441,
|
| 1683 |
+
"step": 2700
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 6.66,
|
| 1687 |
+
"learning_rate": 1.847290640394089e-05,
|
| 1688 |
+
"loss": 0.8257,
|
| 1689 |
+
"step": 2710
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 6.69,
|
| 1693 |
+
"learning_rate": 1.833607006020799e-05,
|
| 1694 |
+
"loss": 0.8367,
|
| 1695 |
+
"step": 2720
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 6.71,
|
| 1699 |
+
"learning_rate": 1.8199233716475097e-05,
|
| 1700 |
+
"loss": 0.8218,
|
| 1701 |
+
"step": 2730
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 6.74,
|
| 1705 |
+
"learning_rate": 1.8062397372742202e-05,
|
| 1706 |
+
"loss": 0.7645,
|
| 1707 |
+
"step": 2740
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 6.76,
|
| 1711 |
+
"learning_rate": 1.7925561029009306e-05,
|
| 1712 |
+
"loss": 0.8305,
|
| 1713 |
+
"step": 2750
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 6.79,
|
| 1717 |
+
"learning_rate": 1.7788724685276407e-05,
|
| 1718 |
+
"loss": 0.8306,
|
| 1719 |
+
"step": 2760
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 6.81,
|
| 1723 |
+
"learning_rate": 1.7651888341543515e-05,
|
| 1724 |
+
"loss": 0.9052,
|
| 1725 |
+
"step": 2770
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 6.83,
|
| 1729 |
+
"learning_rate": 1.751505199781062e-05,
|
| 1730 |
+
"loss": 0.7477,
|
| 1731 |
+
"step": 2780
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 6.86,
|
| 1735 |
+
"learning_rate": 1.7378215654077724e-05,
|
| 1736 |
+
"loss": 0.8458,
|
| 1737 |
+
"step": 2790
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 6.88,
|
| 1741 |
+
"learning_rate": 1.7241379310344828e-05,
|
| 1742 |
+
"loss": 0.901,
|
| 1743 |
+
"step": 2800
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 6.91,
|
| 1747 |
+
"learning_rate": 1.7104542966611933e-05,
|
| 1748 |
+
"loss": 0.8307,
|
| 1749 |
+
"step": 2810
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 6.93,
|
| 1753 |
+
"learning_rate": 1.6967706622879037e-05,
|
| 1754 |
+
"loss": 0.8201,
|
| 1755 |
+
"step": 2820
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 6.96,
|
| 1759 |
+
"learning_rate": 1.683087027914614e-05,
|
| 1760 |
+
"loss": 0.8514,
|
| 1761 |
+
"step": 2830
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 6.98,
|
| 1765 |
+
"learning_rate": 1.6694033935413246e-05,
|
| 1766 |
+
"loss": 0.8414,
|
| 1767 |
+
"step": 2840
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 7.0,
|
| 1771 |
+
"eval_accuracy": 0.9327571305099395,
|
| 1772 |
+
"eval_loss": 0.3537824749946594,
|
| 1773 |
+
"eval_runtime": 27.9648,
|
| 1774 |
+
"eval_samples_per_second": 206.867,
|
| 1775 |
+
"eval_steps_per_second": 6.472,
|
| 1776 |
+
"step": 2847
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 7.01,
|
| 1780 |
+
"learning_rate": 1.655719759168035e-05,
|
| 1781 |
+
"loss": 0.9236,
|
| 1782 |
+
"step": 2850
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 7.03,
|
| 1786 |
+
"learning_rate": 1.6420361247947455e-05,
|
| 1787 |
+
"loss": 0.8118,
|
| 1788 |
+
"step": 2860
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 7.06,
|
| 1792 |
+
"learning_rate": 1.628352490421456e-05,
|
| 1793 |
+
"loss": 0.8715,
|
| 1794 |
+
"step": 2870
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 7.08,
|
| 1798 |
+
"learning_rate": 1.6146688560481667e-05,
|
| 1799 |
+
"loss": 0.8181,
|
| 1800 |
+
"step": 2880
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 7.11,
|
| 1804 |
+
"learning_rate": 1.6009852216748768e-05,
|
| 1805 |
+
"loss": 0.8577,
|
| 1806 |
+
"step": 2890
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 7.13,
|
| 1810 |
+
"learning_rate": 1.5873015873015872e-05,
|
| 1811 |
+
"loss": 0.7951,
|
| 1812 |
+
"step": 2900
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 7.15,
|
| 1816 |
+
"learning_rate": 1.573617952928298e-05,
|
| 1817 |
+
"loss": 0.8414,
|
| 1818 |
+
"step": 2910
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 7.18,
|
| 1822 |
+
"learning_rate": 1.5599343185550085e-05,
|
| 1823 |
+
"loss": 0.8405,
|
| 1824 |
+
"step": 2920
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 7.2,
|
| 1828 |
+
"learning_rate": 1.5462506841817186e-05,
|
| 1829 |
+
"loss": 0.8375,
|
| 1830 |
+
"step": 2930
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 7.23,
|
| 1834 |
+
"learning_rate": 1.532567049808429e-05,
|
| 1835 |
+
"loss": 0.8256,
|
| 1836 |
+
"step": 2940
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 7.25,
|
| 1840 |
+
"learning_rate": 1.5188834154351398e-05,
|
| 1841 |
+
"loss": 0.9106,
|
| 1842 |
+
"step": 2950
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 7.28,
|
| 1846 |
+
"learning_rate": 1.50519978106185e-05,
|
| 1847 |
+
"loss": 0.7944,
|
| 1848 |
+
"step": 2960
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 7.3,
|
| 1852 |
+
"learning_rate": 1.4915161466885605e-05,
|
| 1853 |
+
"loss": 0.8647,
|
| 1854 |
+
"step": 2970
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 7.33,
|
| 1858 |
+
"learning_rate": 1.4778325123152711e-05,
|
| 1859 |
+
"loss": 0.7921,
|
| 1860 |
+
"step": 2980
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 7.35,
|
| 1864 |
+
"learning_rate": 1.4641488779419815e-05,
|
| 1865 |
+
"loss": 0.7978,
|
| 1866 |
+
"step": 2990
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 7.38,
|
| 1870 |
+
"learning_rate": 1.450465243568692e-05,
|
| 1871 |
+
"loss": 0.8136,
|
| 1872 |
+
"step": 3000
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 7.4,
|
| 1876 |
+
"learning_rate": 1.4367816091954022e-05,
|
| 1877 |
+
"loss": 0.8278,
|
| 1878 |
+
"step": 3010
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 7.42,
|
| 1882 |
+
"learning_rate": 1.4230979748221129e-05,
|
| 1883 |
+
"loss": 0.8061,
|
| 1884 |
+
"step": 3020
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 7.45,
|
| 1888 |
+
"learning_rate": 1.4094143404488233e-05,
|
| 1889 |
+
"loss": 0.7277,
|
| 1890 |
+
"step": 3030
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 7.47,
|
| 1894 |
+
"learning_rate": 1.3957307060755337e-05,
|
| 1895 |
+
"loss": 0.809,
|
| 1896 |
+
"step": 3040
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 7.5,
|
| 1900 |
+
"learning_rate": 1.382047071702244e-05,
|
| 1901 |
+
"loss": 0.8359,
|
| 1902 |
+
"step": 3050
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 7.52,
|
| 1906 |
+
"learning_rate": 1.3683634373289548e-05,
|
| 1907 |
+
"loss": 0.8107,
|
| 1908 |
+
"step": 3060
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 7.55,
|
| 1912 |
+
"learning_rate": 1.354679802955665e-05,
|
| 1913 |
+
"loss": 0.8016,
|
| 1914 |
+
"step": 3070
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 7.57,
|
| 1918 |
+
"learning_rate": 1.3409961685823755e-05,
|
| 1919 |
+
"loss": 0.8512,
|
| 1920 |
+
"step": 3080
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 7.6,
|
| 1924 |
+
"learning_rate": 1.3273125342090861e-05,
|
| 1925 |
+
"loss": 0.7948,
|
| 1926 |
+
"step": 3090
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 7.62,
|
| 1930 |
+
"learning_rate": 1.3136288998357965e-05,
|
| 1931 |
+
"loss": 0.8081,
|
| 1932 |
+
"step": 3100
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 7.65,
|
| 1936 |
+
"learning_rate": 1.2999452654625068e-05,
|
| 1937 |
+
"loss": 0.854,
|
| 1938 |
+
"step": 3110
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 7.67,
|
| 1942 |
+
"learning_rate": 1.2862616310892173e-05,
|
| 1943 |
+
"loss": 0.8504,
|
| 1944 |
+
"step": 3120
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 7.7,
|
| 1948 |
+
"learning_rate": 1.2725779967159279e-05,
|
| 1949 |
+
"loss": 0.824,
|
| 1950 |
+
"step": 3130
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 7.72,
|
| 1954 |
+
"learning_rate": 1.2588943623426383e-05,
|
| 1955 |
+
"loss": 0.8119,
|
| 1956 |
+
"step": 3140
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 7.74,
|
| 1960 |
+
"learning_rate": 1.2452107279693487e-05,
|
| 1961 |
+
"loss": 0.8304,
|
| 1962 |
+
"step": 3150
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 7.77,
|
| 1966 |
+
"learning_rate": 1.2315270935960592e-05,
|
| 1967 |
+
"loss": 0.7737,
|
| 1968 |
+
"step": 3160
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 7.79,
|
| 1972 |
+
"learning_rate": 1.2178434592227696e-05,
|
| 1973 |
+
"loss": 0.8032,
|
| 1974 |
+
"step": 3170
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 7.82,
|
| 1978 |
+
"learning_rate": 1.20415982484948e-05,
|
| 1979 |
+
"loss": 0.8448,
|
| 1980 |
+
"step": 3180
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 7.84,
|
| 1984 |
+
"learning_rate": 1.1904761904761905e-05,
|
| 1985 |
+
"loss": 0.755,
|
| 1986 |
+
"step": 3190
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 7.87,
|
| 1990 |
+
"learning_rate": 1.176792556102901e-05,
|
| 1991 |
+
"loss": 0.812,
|
| 1992 |
+
"step": 3200
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 7.89,
|
| 1996 |
+
"learning_rate": 1.1631089217296116e-05,
|
| 1997 |
+
"loss": 0.8194,
|
| 1998 |
+
"step": 3210
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 7.92,
|
| 2002 |
+
"learning_rate": 1.1494252873563218e-05,
|
| 2003 |
+
"loss": 0.8348,
|
| 2004 |
+
"step": 3220
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 7.94,
|
| 2008 |
+
"learning_rate": 1.1357416529830324e-05,
|
| 2009 |
+
"loss": 0.7492,
|
| 2010 |
+
"step": 3230
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 7.97,
|
| 2014 |
+
"learning_rate": 1.1220580186097427e-05,
|
| 2015 |
+
"loss": 0.8227,
|
| 2016 |
+
"step": 3240
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 7.99,
|
| 2020 |
+
"learning_rate": 1.1083743842364533e-05,
|
| 2021 |
+
"loss": 0.8336,
|
| 2022 |
+
"step": 3250
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 8.0,
|
| 2026 |
+
"eval_accuracy": 0.9334485738980121,
|
| 2027 |
+
"eval_loss": 0.3332645893096924,
|
| 2028 |
+
"eval_runtime": 29.4073,
|
| 2029 |
+
"eval_samples_per_second": 196.72,
|
| 2030 |
+
"eval_steps_per_second": 6.155,
|
| 2031 |
+
"step": 3254
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 8.01,
|
| 2035 |
+
"learning_rate": 1.0946907498631636e-05,
|
| 2036 |
+
"loss": 0.7665,
|
| 2037 |
+
"step": 3260
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 8.04,
|
| 2041 |
+
"learning_rate": 1.0810071154898742e-05,
|
| 2042 |
+
"loss": 0.807,
|
| 2043 |
+
"step": 3270
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 8.06,
|
| 2047 |
+
"learning_rate": 1.0673234811165846e-05,
|
| 2048 |
+
"loss": 0.8521,
|
| 2049 |
+
"step": 3280
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 8.09,
|
| 2053 |
+
"learning_rate": 1.053639846743295e-05,
|
| 2054 |
+
"loss": 0.7569,
|
| 2055 |
+
"step": 3290
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 8.11,
|
| 2059 |
+
"learning_rate": 1.0399562123700055e-05,
|
| 2060 |
+
"loss": 0.7465,
|
| 2061 |
+
"step": 3300
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 8.14,
|
| 2065 |
+
"learning_rate": 1.026272577996716e-05,
|
| 2066 |
+
"loss": 0.8432,
|
| 2067 |
+
"step": 3310
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 8.16,
|
| 2071 |
+
"learning_rate": 1.0125889436234266e-05,
|
| 2072 |
+
"loss": 0.7659,
|
| 2073 |
+
"step": 3320
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 8.19,
|
| 2077 |
+
"learning_rate": 9.989053092501368e-06,
|
| 2078 |
+
"loss": 0.7682,
|
| 2079 |
+
"step": 3330
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 8.21,
|
| 2083 |
+
"learning_rate": 9.852216748768475e-06,
|
| 2084 |
+
"loss": 0.7285,
|
| 2085 |
+
"step": 3340
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 8.24,
|
| 2089 |
+
"learning_rate": 9.715380405035577e-06,
|
| 2090 |
+
"loss": 0.8721,
|
| 2091 |
+
"step": 3350
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 8.26,
|
| 2095 |
+
"learning_rate": 9.578544061302683e-06,
|
| 2096 |
+
"loss": 0.761,
|
| 2097 |
+
"step": 3360
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 8.29,
|
| 2101 |
+
"learning_rate": 9.441707717569786e-06,
|
| 2102 |
+
"loss": 0.7812,
|
| 2103 |
+
"step": 3370
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 8.31,
|
| 2107 |
+
"learning_rate": 9.304871373836892e-06,
|
| 2108 |
+
"loss": 0.8125,
|
| 2109 |
+
"step": 3380
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 8.33,
|
| 2113 |
+
"learning_rate": 9.168035030103995e-06,
|
| 2114 |
+
"loss": 0.8344,
|
| 2115 |
+
"step": 3390
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 8.36,
|
| 2119 |
+
"learning_rate": 9.031198686371101e-06,
|
| 2120 |
+
"loss": 0.7884,
|
| 2121 |
+
"step": 3400
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 8.38,
|
| 2125 |
+
"learning_rate": 8.894362342638204e-06,
|
| 2126 |
+
"loss": 0.7692,
|
| 2127 |
+
"step": 3410
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 8.41,
|
| 2131 |
+
"learning_rate": 8.75752599890531e-06,
|
| 2132 |
+
"loss": 0.875,
|
| 2133 |
+
"step": 3420
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 8.43,
|
| 2137 |
+
"learning_rate": 8.620689655172414e-06,
|
| 2138 |
+
"loss": 0.7938,
|
| 2139 |
+
"step": 3430
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 8.46,
|
| 2143 |
+
"learning_rate": 8.483853311439519e-06,
|
| 2144 |
+
"loss": 0.7749,
|
| 2145 |
+
"step": 3440
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 8.48,
|
| 2149 |
+
"learning_rate": 8.347016967706623e-06,
|
| 2150 |
+
"loss": 0.7628,
|
| 2151 |
+
"step": 3450
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 8.51,
|
| 2155 |
+
"learning_rate": 8.210180623973727e-06,
|
| 2156 |
+
"loss": 0.8007,
|
| 2157 |
+
"step": 3460
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 8.53,
|
| 2161 |
+
"learning_rate": 8.073344280240833e-06,
|
| 2162 |
+
"loss": 0.7281,
|
| 2163 |
+
"step": 3470
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 8.56,
|
| 2167 |
+
"learning_rate": 7.936507936507936e-06,
|
| 2168 |
+
"loss": 0.7633,
|
| 2169 |
+
"step": 3480
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 8.58,
|
| 2173 |
+
"learning_rate": 7.799671592775042e-06,
|
| 2174 |
+
"loss": 0.8202,
|
| 2175 |
+
"step": 3490
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 8.6,
|
| 2179 |
+
"learning_rate": 7.662835249042145e-06,
|
| 2180 |
+
"loss": 0.8123,
|
| 2181 |
+
"step": 3500
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 8.63,
|
| 2185 |
+
"learning_rate": 7.52599890530925e-06,
|
| 2186 |
+
"loss": 0.8223,
|
| 2187 |
+
"step": 3510
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 8.65,
|
| 2191 |
+
"learning_rate": 7.3891625615763555e-06,
|
| 2192 |
+
"loss": 0.8456,
|
| 2193 |
+
"step": 3520
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 8.68,
|
| 2197 |
+
"learning_rate": 7.25232621784346e-06,
|
| 2198 |
+
"loss": 0.752,
|
| 2199 |
+
"step": 3530
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 8.7,
|
| 2203 |
+
"learning_rate": 7.115489874110564e-06,
|
| 2204 |
+
"loss": 0.7942,
|
| 2205 |
+
"step": 3540
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 8.73,
|
| 2209 |
+
"learning_rate": 6.978653530377669e-06,
|
| 2210 |
+
"loss": 0.7635,
|
| 2211 |
+
"step": 3550
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 8.75,
|
| 2215 |
+
"learning_rate": 6.841817186644774e-06,
|
| 2216 |
+
"loss": 0.7014,
|
| 2217 |
+
"step": 3560
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 8.78,
|
| 2221 |
+
"learning_rate": 6.7049808429118775e-06,
|
| 2222 |
+
"loss": 0.7572,
|
| 2223 |
+
"step": 3570
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 8.8,
|
| 2227 |
+
"learning_rate": 6.568144499178983e-06,
|
| 2228 |
+
"loss": 0.7816,
|
| 2229 |
+
"step": 3580
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 8.83,
|
| 2233 |
+
"learning_rate": 6.431308155446086e-06,
|
| 2234 |
+
"loss": 0.7336,
|
| 2235 |
+
"step": 3590
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 8.85,
|
| 2239 |
+
"learning_rate": 6.2944718117131915e-06,
|
| 2240 |
+
"loss": 0.7665,
|
| 2241 |
+
"step": 3600
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 8.88,
|
| 2245 |
+
"learning_rate": 6.157635467980296e-06,
|
| 2246 |
+
"loss": 0.7669,
|
| 2247 |
+
"step": 3610
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 8.9,
|
| 2251 |
+
"learning_rate": 6.0207991242474e-06,
|
| 2252 |
+
"loss": 0.7587,
|
| 2253 |
+
"step": 3620
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 8.92,
|
| 2257 |
+
"learning_rate": 5.883962780514505e-06,
|
| 2258 |
+
"loss": 0.8309,
|
| 2259 |
+
"step": 3630
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 8.95,
|
| 2263 |
+
"learning_rate": 5.747126436781609e-06,
|
| 2264 |
+
"loss": 0.7201,
|
| 2265 |
+
"step": 3640
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 8.97,
|
| 2269 |
+
"learning_rate": 5.6102900930487136e-06,
|
| 2270 |
+
"loss": 0.736,
|
| 2271 |
+
"step": 3650
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 9.0,
|
| 2275 |
+
"learning_rate": 5.473453749315818e-06,
|
| 2276 |
+
"loss": 0.758,
|
| 2277 |
+
"step": 3660
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 9.0,
|
| 2281 |
+
"eval_accuracy": 0.9405358686257562,
|
| 2282 |
+
"eval_loss": 0.3180586099624634,
|
| 2283 |
+
"eval_runtime": 28.3456,
|
| 2284 |
+
"eval_samples_per_second": 204.088,
|
| 2285 |
+
"eval_steps_per_second": 6.385,
|
| 2286 |
+
"step": 3660
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 9.02,
|
| 2290 |
+
"learning_rate": 5.336617405582923e-06,
|
| 2291 |
+
"loss": 0.6856,
|
| 2292 |
+
"step": 3670
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 9.05,
|
| 2296 |
+
"learning_rate": 5.199781061850028e-06,
|
| 2297 |
+
"loss": 0.8134,
|
| 2298 |
+
"step": 3680
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 9.07,
|
| 2302 |
+
"learning_rate": 5.062944718117133e-06,
|
| 2303 |
+
"loss": 0.8841,
|
| 2304 |
+
"step": 3690
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 9.1,
|
| 2308 |
+
"learning_rate": 4.926108374384237e-06,
|
| 2309 |
+
"loss": 0.7623,
|
| 2310 |
+
"step": 3700
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 9.12,
|
| 2314 |
+
"learning_rate": 4.789272030651342e-06,
|
| 2315 |
+
"loss": 0.7488,
|
| 2316 |
+
"step": 3710
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 9.15,
|
| 2320 |
+
"learning_rate": 4.652435686918446e-06,
|
| 2321 |
+
"loss": 0.7433,
|
| 2322 |
+
"step": 3720
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 9.17,
|
| 2326 |
+
"learning_rate": 4.5155993431855505e-06,
|
| 2327 |
+
"loss": 0.7512,
|
| 2328 |
+
"step": 3730
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 9.19,
|
| 2332 |
+
"learning_rate": 4.378762999452655e-06,
|
| 2333 |
+
"loss": 0.7725,
|
| 2334 |
+
"step": 3740
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 9.22,
|
| 2338 |
+
"learning_rate": 4.241926655719759e-06,
|
| 2339 |
+
"loss": 0.7567,
|
| 2340 |
+
"step": 3750
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 9.24,
|
| 2344 |
+
"learning_rate": 4.105090311986864e-06,
|
| 2345 |
+
"loss": 0.7213,
|
| 2346 |
+
"step": 3760
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 9.27,
|
| 2350 |
+
"learning_rate": 3.968253968253968e-06,
|
| 2351 |
+
"loss": 0.75,
|
| 2352 |
+
"step": 3770
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 9.29,
|
| 2356 |
+
"learning_rate": 3.8314176245210725e-06,
|
| 2357 |
+
"loss": 0.7999,
|
| 2358 |
+
"step": 3780
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 9.32,
|
| 2362 |
+
"learning_rate": 3.6945812807881777e-06,
|
| 2363 |
+
"loss": 0.7489,
|
| 2364 |
+
"step": 3790
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 9.34,
|
| 2368 |
+
"learning_rate": 3.557744937055282e-06,
|
| 2369 |
+
"loss": 0.7591,
|
| 2370 |
+
"step": 3800
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 9.37,
|
| 2374 |
+
"learning_rate": 3.420908593322387e-06,
|
| 2375 |
+
"loss": 0.769,
|
| 2376 |
+
"step": 3810
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 9.39,
|
| 2380 |
+
"learning_rate": 3.2840722495894914e-06,
|
| 2381 |
+
"loss": 0.7171,
|
| 2382 |
+
"step": 3820
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 9.42,
|
| 2386 |
+
"learning_rate": 3.1472359058565958e-06,
|
| 2387 |
+
"loss": 0.7158,
|
| 2388 |
+
"step": 3830
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 9.44,
|
| 2392 |
+
"learning_rate": 3.0103995621237e-06,
|
| 2393 |
+
"loss": 0.7132,
|
| 2394 |
+
"step": 3840
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 9.47,
|
| 2398 |
+
"learning_rate": 2.8735632183908046e-06,
|
| 2399 |
+
"loss": 0.7236,
|
| 2400 |
+
"step": 3850
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 9.49,
|
| 2404 |
+
"learning_rate": 2.736726874657909e-06,
|
| 2405 |
+
"loss": 0.7361,
|
| 2406 |
+
"step": 3860
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 9.51,
|
| 2410 |
+
"learning_rate": 2.599890530925014e-06,
|
| 2411 |
+
"loss": 0.6774,
|
| 2412 |
+
"step": 3870
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 9.54,
|
| 2416 |
+
"learning_rate": 2.4630541871921186e-06,
|
| 2417 |
+
"loss": 0.8082,
|
| 2418 |
+
"step": 3880
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 9.56,
|
| 2422 |
+
"learning_rate": 2.326217843459223e-06,
|
| 2423 |
+
"loss": 0.776,
|
| 2424 |
+
"step": 3890
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 9.59,
|
| 2428 |
+
"learning_rate": 2.1893814997263274e-06,
|
| 2429 |
+
"loss": 0.7655,
|
| 2430 |
+
"step": 3900
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 9.61,
|
| 2434 |
+
"learning_rate": 2.052545155993432e-06,
|
| 2435 |
+
"loss": 0.7834,
|
| 2436 |
+
"step": 3910
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 9.64,
|
| 2440 |
+
"learning_rate": 1.9157088122605362e-06,
|
| 2441 |
+
"loss": 0.7236,
|
| 2442 |
+
"step": 3920
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 9.66,
|
| 2446 |
+
"learning_rate": 1.778872468527641e-06,
|
| 2447 |
+
"loss": 0.6974,
|
| 2448 |
+
"step": 3930
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 9.69,
|
| 2452 |
+
"learning_rate": 1.6420361247947457e-06,
|
| 2453 |
+
"loss": 0.73,
|
| 2454 |
+
"step": 3940
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 9.71,
|
| 2458 |
+
"learning_rate": 1.50519978106185e-06,
|
| 2459 |
+
"loss": 0.6768,
|
| 2460 |
+
"step": 3950
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 9.74,
|
| 2464 |
+
"learning_rate": 1.3683634373289545e-06,
|
| 2465 |
+
"loss": 0.7841,
|
| 2466 |
+
"step": 3960
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 9.76,
|
| 2470 |
+
"learning_rate": 1.2315270935960593e-06,
|
| 2471 |
+
"loss": 0.759,
|
| 2472 |
+
"step": 3970
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 9.78,
|
| 2476 |
+
"learning_rate": 1.0946907498631637e-06,
|
| 2477 |
+
"loss": 0.7244,
|
| 2478 |
+
"step": 3980
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 9.81,
|
| 2482 |
+
"learning_rate": 9.578544061302681e-07,
|
| 2483 |
+
"loss": 0.7326,
|
| 2484 |
+
"step": 3990
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 9.83,
|
| 2488 |
+
"learning_rate": 8.210180623973728e-07,
|
| 2489 |
+
"loss": 0.7484,
|
| 2490 |
+
"step": 4000
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 9.86,
|
| 2494 |
+
"learning_rate": 6.841817186644772e-07,
|
| 2495 |
+
"loss": 0.7589,
|
| 2496 |
+
"step": 4010
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 9.88,
|
| 2500 |
+
"learning_rate": 5.473453749315819e-07,
|
| 2501 |
+
"loss": 0.7169,
|
| 2502 |
+
"step": 4020
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 9.91,
|
| 2506 |
+
"learning_rate": 4.105090311986864e-07,
|
| 2507 |
+
"loss": 0.6762,
|
| 2508 |
+
"step": 4030
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 9.93,
|
| 2512 |
+
"learning_rate": 2.7367268746579093e-07,
|
| 2513 |
+
"loss": 0.7896,
|
| 2514 |
+
"step": 4040
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 9.96,
|
| 2518 |
+
"learning_rate": 1.3683634373289546e-07,
|
| 2519 |
+
"loss": 0.8115,
|
| 2520 |
+
"step": 4050
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 9.98,
|
| 2524 |
+
"learning_rate": 0.0,
|
| 2525 |
+
"loss": 0.7343,
|
| 2526 |
+
"step": 4060
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 9.98,
|
| 2530 |
+
"eval_accuracy": 0.9403630077787382,
|
| 2531 |
+
"eval_loss": 0.31005963683128357,
|
| 2532 |
+
"eval_runtime": 28.1514,
|
| 2533 |
+
"eval_samples_per_second": 205.496,
|
| 2534 |
+
"eval_steps_per_second": 6.43,
|
| 2535 |
+
"step": 4060
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 9.98,
|
| 2539 |
+
"step": 4060,
|
| 2540 |
+
"total_flos": 1.3140619208067262e+19,
|
| 2541 |
+
"train_loss": 1.0548507860728673,
|
| 2542 |
+
"train_runtime": 4291.8856,
|
| 2543 |
+
"train_samples_per_second": 121.301,
|
| 2544 |
+
"train_steps_per_second": 0.946
|
| 2545 |
}
|
| 2546 |
],
|
| 2547 |
"logging_steps": 10,
|
| 2548 |
+
"max_steps": 4060,
|
| 2549 |
"num_input_tokens_seen": 0,
|
| 2550 |
+
"num_train_epochs": 10,
|
| 2551 |
"save_steps": 500,
|
| 2552 |
+
"total_flos": 1.3140619208067262e+19,
|
| 2553 |
"train_batch_size": 32,
|
| 2554 |
"trial_name": null,
|
| 2555 |
"trial_params": null
|