Training in progress, step 100
Browse files- all_results.json +5 -5
- pytorch_model.bin +1 -1
- runs/Mar10_21-05-02_a4c3c39bee63/1678482309.2431283/events.out.tfevents.1678482309.a4c3c39bee63.128.3 +3 -0
- runs/Mar10_21-05-02_a4c3c39bee63/events.out.tfevents.1678482309.a4c3c39bee63.128.2 +3 -0
- train_results.json +5 -5
- trainer_state.json +140 -497
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 1.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
+
"total_flos": 4.665792694694584e+17,
|
| 4 |
+
"train_loss": 0.15108026370958047,
|
| 5 |
+
"train_runtime": 222.8357,
|
| 6 |
+
"train_samples_per_second": 27.02,
|
| 7 |
+
"train_steps_per_second": 1.696
|
| 8 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343268717
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:176c26fcece8fa93ca3638cc5a44ff81006609d9ef30cd57f914e8dba43a457e
|
| 3 |
size 343268717
|
runs/Mar10_21-05-02_a4c3c39bee63/1678482309.2431283/events.out.tfevents.1678482309.a4c3c39bee63.128.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:04ddefbee056805dcbfa2ceef715fdeb8f19116303e5c0aa4bbcecedeab538b0
|
| 3 |
+
size 5695
|
runs/Mar10_21-05-02_a4c3c39bee63/events.out.tfevents.1678482309.a4c3c39bee63.128.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:675c928040cca04ec1fe14efc37e49d0020bb21865ae0937c33f66f0f7ae281d
|
| 3 |
+
size 5808
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 1.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
+
"total_flos": 4.665792694694584e+17,
|
| 4 |
+
"train_loss": 0.15108026370958047,
|
| 5 |
+
"train_runtime": 222.8357,
|
| 6 |
+
"train_samples_per_second": 27.02,
|
| 7 |
+
"train_steps_per_second": 1.696
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,631 +1,274 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "./output/vit-base-riego/checkpoint-
|
| 4 |
"epoch": 3.0,
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
-
"epoch": 0.
|
| 12 |
-
"learning_rate": 0.
|
| 13 |
-
"loss": 0.
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
-
"epoch": 0.
|
| 18 |
-
"learning_rate": 0.
|
| 19 |
-
"loss": 0.
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
-
"epoch": 0.
|
| 24 |
-
"learning_rate": 0.
|
| 25 |
-
"loss": 0.
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
-
"learning_rate": 0.
|
| 31 |
-
"loss": 0.
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
-
"epoch": 0.
|
| 36 |
-
"learning_rate": 0.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 0.
|
| 42 |
-
"learning_rate": 0.
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"learning_rate": 0.
|
| 49 |
-
"loss": 0.
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"learning_rate": 0.
|
| 55 |
-
"loss": 0.
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
-
"epoch": 0.
|
| 60 |
-
"learning_rate": 0.
|
| 61 |
-
"loss": 0.
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
-
"epoch": 0.
|
| 66 |
-
"learning_rate": 0.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
-
"epoch": 0.
|
| 72 |
-
"eval_f1": 0.
|
| 73 |
-
"eval_loss": 0.
|
| 74 |
-
"eval_runtime":
|
| 75 |
-
"eval_samples_per_second":
|
| 76 |
-
"eval_steps_per_second": 16.
|
| 77 |
"step": 100
|
| 78 |
},
|
| 79 |
{
|
| 80 |
-
"epoch": 0.
|
| 81 |
-
"learning_rate": 0.
|
| 82 |
-
"loss": 0.
|
| 83 |
"step": 110
|
| 84 |
},
|
| 85 |
{
|
| 86 |
-
"epoch": 0.
|
| 87 |
-
"learning_rate": 0.
|
| 88 |
-
"loss": 0.
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
-
"epoch":
|
| 93 |
-
"learning_rate": 0.
|
| 94 |
-
"loss": 0.
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
-
"epoch":
|
| 99 |
-
"learning_rate": 0.
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
-
"epoch":
|
| 105 |
-
"learning_rate": 0.
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch":
|
| 111 |
-
"learning_rate": 0.
|
| 112 |
-
"loss": 0.
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
-
"epoch":
|
| 117 |
-
"learning_rate": 0.
|
| 118 |
-
"loss": 0.
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
-
"epoch":
|
| 123 |
-
"learning_rate": 0.
|
| 124 |
-
"loss": 0.
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
-
"epoch":
|
| 129 |
-
"learning_rate": 0.
|
| 130 |
-
"loss": 0.
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
-
"epoch":
|
| 135 |
-
"learning_rate":
|
| 136 |
-
"loss": 0.
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
-
"epoch":
|
| 141 |
-
"eval_f1": 0.
|
| 142 |
-
"eval_loss": 0.
|
| 143 |
-
"eval_runtime":
|
| 144 |
-
"eval_samples_per_second":
|
| 145 |
-
"eval_steps_per_second":
|
| 146 |
"step": 200
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch":
|
| 150 |
-
"learning_rate":
|
| 151 |
-
"loss": 0.
|
| 152 |
"step": 210
|
| 153 |
},
|
| 154 |
{
|
| 155 |
-
"epoch":
|
| 156 |
-
"learning_rate":
|
| 157 |
-
"loss": 0.
|
| 158 |
"step": 220
|
| 159 |
},
|
| 160 |
{
|
| 161 |
-
"epoch":
|
| 162 |
-
"learning_rate":
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 230
|
| 165 |
},
|
| 166 |
{
|
| 167 |
-
"epoch":
|
| 168 |
-
"learning_rate":
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 240
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch":
|
| 174 |
-
"learning_rate":
|
| 175 |
-
"loss": 0.
|
| 176 |
"step": 250
|
| 177 |
},
|
| 178 |
{
|
| 179 |
-
"epoch":
|
| 180 |
-
"learning_rate":
|
| 181 |
-
"loss": 0.
|
| 182 |
"step": 260
|
| 183 |
},
|
| 184 |
{
|
| 185 |
-
"epoch":
|
| 186 |
-
"learning_rate":
|
| 187 |
-
"loss": 0.
|
| 188 |
"step": 270
|
| 189 |
},
|
| 190 |
{
|
| 191 |
-
"epoch":
|
| 192 |
-
"learning_rate":
|
| 193 |
-
"loss": 0.
|
| 194 |
"step": 280
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch":
|
| 198 |
-
"learning_rate":
|
| 199 |
-
"loss": 0.
|
| 200 |
"step": 290
|
| 201 |
},
|
| 202 |
{
|
| 203 |
-
"epoch":
|
| 204 |
-
"learning_rate":
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 300
|
| 207 |
},
|
| 208 |
{
|
| 209 |
-
"epoch":
|
| 210 |
-
"eval_f1": 0.
|
| 211 |
-
"eval_loss": 0.
|
| 212 |
-
"eval_runtime":
|
| 213 |
-
"eval_samples_per_second":
|
| 214 |
-
"eval_steps_per_second":
|
| 215 |
"step": 300
|
| 216 |
},
|
| 217 |
{
|
| 218 |
-
"epoch":
|
| 219 |
-
"learning_rate":
|
| 220 |
-
"loss": 0.
|
| 221 |
"step": 310
|
| 222 |
},
|
| 223 |
{
|
| 224 |
-
"epoch":
|
| 225 |
-
"learning_rate":
|
| 226 |
-
"loss": 0.
|
| 227 |
"step": 320
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"epoch":
|
| 231 |
-
"learning_rate":
|
| 232 |
-
"loss": 0.
|
| 233 |
"step": 330
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch":
|
| 237 |
-
"learning_rate":
|
| 238 |
-
"loss": 0.
|
| 239 |
"step": 340
|
| 240 |
},
|
| 241 |
{
|
| 242 |
-
"epoch":
|
| 243 |
-
"learning_rate":
|
| 244 |
-
"loss": 0.
|
| 245 |
"step": 350
|
| 246 |
},
|
| 247 |
{
|
| 248 |
-
"epoch":
|
| 249 |
-
"learning_rate":
|
| 250 |
-
"loss": 0.
|
| 251 |
"step": 360
|
| 252 |
},
|
| 253 |
{
|
| 254 |
-
"epoch":
|
| 255 |
-
"learning_rate":
|
| 256 |
-
"loss": 0.
|
| 257 |
"step": 370
|
| 258 |
},
|
| 259 |
-
{
|
| 260 |
-
"epoch": 1.28,
|
| 261 |
-
"learning_rate": 0.00011515151515151516,
|
| 262 |
-
"loss": 0.0973,
|
| 263 |
-
"step": 380
|
| 264 |
-
},
|
| 265 |
-
{
|
| 266 |
-
"epoch": 1.31,
|
| 267 |
-
"learning_rate": 0.00011290684624017958,
|
| 268 |
-
"loss": 0.1067,
|
| 269 |
-
"step": 390
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 1.35,
|
| 273 |
-
"learning_rate": 0.00011066217732884401,
|
| 274 |
-
"loss": 0.128,
|
| 275 |
-
"step": 400
|
| 276 |
-
},
|
| 277 |
-
{
|
| 278 |
-
"epoch": 1.35,
|
| 279 |
-
"eval_f1": 0.9573863636363636,
|
| 280 |
-
"eval_loss": 0.11194377392530441,
|
| 281 |
-
"eval_runtime": 8.4731,
|
| 282 |
-
"eval_samples_per_second": 99.137,
|
| 283 |
-
"eval_steps_per_second": 12.392,
|
| 284 |
-
"step": 400
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 1.38,
|
| 288 |
-
"learning_rate": 0.00010841750841750841,
|
| 289 |
-
"loss": 0.1299,
|
| 290 |
-
"step": 410
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 1.41,
|
| 294 |
-
"learning_rate": 0.00010617283950617284,
|
| 295 |
-
"loss": 0.1451,
|
| 296 |
-
"step": 420
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 1.45,
|
| 300 |
-
"learning_rate": 0.00010392817059483726,
|
| 301 |
-
"loss": 0.0638,
|
| 302 |
-
"step": 430
|
| 303 |
-
},
|
| 304 |
-
{
|
| 305 |
-
"epoch": 1.48,
|
| 306 |
-
"learning_rate": 0.00010168350168350169,
|
| 307 |
-
"loss": 0.0818,
|
| 308 |
-
"step": 440
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 1.52,
|
| 312 |
-
"learning_rate": 9.943883277216612e-05,
|
| 313 |
-
"loss": 0.3606,
|
| 314 |
-
"step": 450
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 1.55,
|
| 318 |
-
"learning_rate": 9.719416386083054e-05,
|
| 319 |
-
"loss": 0.092,
|
| 320 |
-
"step": 460
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 1.58,
|
| 324 |
-
"learning_rate": 9.494949494949495e-05,
|
| 325 |
-
"loss": 0.1516,
|
| 326 |
-
"step": 470
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 1.62,
|
| 330 |
-
"learning_rate": 9.270482603815938e-05,
|
| 331 |
-
"loss": 0.0527,
|
| 332 |
-
"step": 480
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 1.65,
|
| 336 |
-
"learning_rate": 9.04601571268238e-05,
|
| 337 |
-
"loss": 0.1143,
|
| 338 |
-
"step": 490
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 1.68,
|
| 342 |
-
"learning_rate": 8.821548821548821e-05,
|
| 343 |
-
"loss": 0.1387,
|
| 344 |
-
"step": 500
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 1.68,
|
| 348 |
-
"eval_f1": 0.9495798319327731,
|
| 349 |
-
"eval_loss": 0.12221043556928635,
|
| 350 |
-
"eval_runtime": 5.8532,
|
| 351 |
-
"eval_samples_per_second": 143.512,
|
| 352 |
-
"eval_steps_per_second": 17.939,
|
| 353 |
-
"step": 500
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 1.72,
|
| 357 |
-
"learning_rate": 8.597081930415264e-05,
|
| 358 |
-
"loss": 0.1286,
|
| 359 |
-
"step": 510
|
| 360 |
-
},
|
| 361 |
-
{
|
| 362 |
-
"epoch": 1.75,
|
| 363 |
-
"learning_rate": 8.372615039281706e-05,
|
| 364 |
-
"loss": 0.0926,
|
| 365 |
-
"step": 520
|
| 366 |
-
},
|
| 367 |
-
{
|
| 368 |
-
"epoch": 1.78,
|
| 369 |
-
"learning_rate": 8.148148148148148e-05,
|
| 370 |
-
"loss": 0.0371,
|
| 371 |
-
"step": 530
|
| 372 |
-
},
|
| 373 |
-
{
|
| 374 |
-
"epoch": 1.82,
|
| 375 |
-
"learning_rate": 7.92368125701459e-05,
|
| 376 |
-
"loss": 0.0766,
|
| 377 |
-
"step": 540
|
| 378 |
-
},
|
| 379 |
-
{
|
| 380 |
-
"epoch": 1.85,
|
| 381 |
-
"learning_rate": 7.699214365881034e-05,
|
| 382 |
-
"loss": 0.0888,
|
| 383 |
-
"step": 550
|
| 384 |
-
},
|
| 385 |
-
{
|
| 386 |
-
"epoch": 1.89,
|
| 387 |
-
"learning_rate": 7.474747474747475e-05,
|
| 388 |
-
"loss": 0.0791,
|
| 389 |
-
"step": 560
|
| 390 |
-
},
|
| 391 |
-
{
|
| 392 |
-
"epoch": 1.92,
|
| 393 |
-
"learning_rate": 7.250280583613918e-05,
|
| 394 |
-
"loss": 0.0833,
|
| 395 |
-
"step": 570
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 1.95,
|
| 399 |
-
"learning_rate": 7.02581369248036e-05,
|
| 400 |
-
"loss": 0.0651,
|
| 401 |
-
"step": 580
|
| 402 |
-
},
|
| 403 |
-
{
|
| 404 |
-
"epoch": 1.99,
|
| 405 |
-
"learning_rate": 6.801346801346801e-05,
|
| 406 |
-
"loss": 0.0134,
|
| 407 |
-
"step": 590
|
| 408 |
-
},
|
| 409 |
-
{
|
| 410 |
-
"epoch": 2.02,
|
| 411 |
-
"learning_rate": 6.576879910213244e-05,
|
| 412 |
-
"loss": 0.0899,
|
| 413 |
-
"step": 600
|
| 414 |
-
},
|
| 415 |
-
{
|
| 416 |
-
"epoch": 2.02,
|
| 417 |
-
"eval_f1": 0.9765395894428153,
|
| 418 |
-
"eval_loss": 0.0800161063671112,
|
| 419 |
-
"eval_runtime": 6.9218,
|
| 420 |
-
"eval_samples_per_second": 121.356,
|
| 421 |
-
"eval_steps_per_second": 15.17,
|
| 422 |
-
"step": 600
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 2.05,
|
| 426 |
-
"learning_rate": 6.352413019079686e-05,
|
| 427 |
-
"loss": 0.042,
|
| 428 |
-
"step": 610
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 2.09,
|
| 432 |
-
"learning_rate": 6.150392817059484e-05,
|
| 433 |
-
"loss": 0.0251,
|
| 434 |
-
"step": 620
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 2.12,
|
| 438 |
-
"learning_rate": 5.925925925925926e-05,
|
| 439 |
-
"loss": 0.0061,
|
| 440 |
-
"step": 630
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 2.15,
|
| 444 |
-
"learning_rate": 5.701459034792368e-05,
|
| 445 |
-
"loss": 0.0495,
|
| 446 |
-
"step": 640
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 2.19,
|
| 450 |
-
"learning_rate": 5.47699214365881e-05,
|
| 451 |
-
"loss": 0.0586,
|
| 452 |
-
"step": 650
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 2.22,
|
| 456 |
-
"learning_rate": 5.2525252525252536e-05,
|
| 457 |
-
"loss": 0.0062,
|
| 458 |
-
"step": 660
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 2.26,
|
| 462 |
-
"learning_rate": 5.028058361391695e-05,
|
| 463 |
-
"loss": 0.027,
|
| 464 |
-
"step": 670
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 2.29,
|
| 468 |
-
"learning_rate": 4.803591470258137e-05,
|
| 469 |
-
"loss": 0.0356,
|
| 470 |
-
"step": 680
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 2.32,
|
| 474 |
-
"learning_rate": 4.57912457912458e-05,
|
| 475 |
-
"loss": 0.0372,
|
| 476 |
-
"step": 690
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 2.36,
|
| 480 |
-
"learning_rate": 4.3546576879910214e-05,
|
| 481 |
-
"loss": 0.0048,
|
| 482 |
-
"step": 700
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 2.36,
|
| 486 |
-
"eval_f1": 0.9769452449567723,
|
| 487 |
-
"eval_loss": 0.07305397093296051,
|
| 488 |
-
"eval_runtime": 5.7864,
|
| 489 |
-
"eval_samples_per_second": 145.168,
|
| 490 |
-
"eval_steps_per_second": 18.146,
|
| 491 |
-
"step": 700
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 2.39,
|
| 495 |
-
"learning_rate": 4.130190796857464e-05,
|
| 496 |
-
"loss": 0.0318,
|
| 497 |
-
"step": 710
|
| 498 |
-
},
|
| 499 |
-
{
|
| 500 |
-
"epoch": 2.42,
|
| 501 |
-
"learning_rate": 3.905723905723906e-05,
|
| 502 |
-
"loss": 0.0858,
|
| 503 |
-
"step": 720
|
| 504 |
-
},
|
| 505 |
-
{
|
| 506 |
-
"epoch": 2.46,
|
| 507 |
-
"learning_rate": 3.681257014590348e-05,
|
| 508 |
-
"loss": 0.065,
|
| 509 |
-
"step": 730
|
| 510 |
-
},
|
| 511 |
-
{
|
| 512 |
-
"epoch": 2.49,
|
| 513 |
-
"learning_rate": 3.45679012345679e-05,
|
| 514 |
-
"loss": 0.1117,
|
| 515 |
-
"step": 740
|
| 516 |
-
},
|
| 517 |
-
{
|
| 518 |
-
"epoch": 2.53,
|
| 519 |
-
"learning_rate": 3.232323232323233e-05,
|
| 520 |
-
"loss": 0.0432,
|
| 521 |
-
"step": 750
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 2.56,
|
| 525 |
-
"learning_rate": 3.007856341189675e-05,
|
| 526 |
-
"loss": 0.0209,
|
| 527 |
-
"step": 760
|
| 528 |
-
},
|
| 529 |
-
{
|
| 530 |
-
"epoch": 2.59,
|
| 531 |
-
"learning_rate": 2.783389450056117e-05,
|
| 532 |
-
"loss": 0.0591,
|
| 533 |
-
"step": 770
|
| 534 |
-
},
|
| 535 |
-
{
|
| 536 |
-
"epoch": 2.63,
|
| 537 |
-
"learning_rate": 2.5589225589225592e-05,
|
| 538 |
-
"loss": 0.0276,
|
| 539 |
-
"step": 780
|
| 540 |
-
},
|
| 541 |
-
{
|
| 542 |
-
"epoch": 2.66,
|
| 543 |
-
"learning_rate": 2.3344556677890012e-05,
|
| 544 |
-
"loss": 0.0058,
|
| 545 |
-
"step": 790
|
| 546 |
-
},
|
| 547 |
-
{
|
| 548 |
-
"epoch": 2.69,
|
| 549 |
-
"learning_rate": 2.1099887766554435e-05,
|
| 550 |
-
"loss": 0.0195,
|
| 551 |
-
"step": 800
|
| 552 |
-
},
|
| 553 |
-
{
|
| 554 |
-
"epoch": 2.69,
|
| 555 |
-
"eval_f1": 0.9885057471264367,
|
| 556 |
-
"eval_loss": 0.040985796600580215,
|
| 557 |
-
"eval_runtime": 6.6774,
|
| 558 |
-
"eval_samples_per_second": 125.798,
|
| 559 |
-
"eval_steps_per_second": 15.725,
|
| 560 |
-
"step": 800
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 2.73,
|
| 564 |
-
"learning_rate": 1.8855218855218858e-05,
|
| 565 |
-
"loss": 0.0055,
|
| 566 |
-
"step": 810
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 2.76,
|
| 570 |
-
"learning_rate": 1.6610549943883278e-05,
|
| 571 |
-
"loss": 0.0351,
|
| 572 |
-
"step": 820
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 2.79,
|
| 576 |
-
"learning_rate": 1.43658810325477e-05,
|
| 577 |
-
"loss": 0.0045,
|
| 578 |
-
"step": 830
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 2.83,
|
| 582 |
-
"learning_rate": 1.2121212121212122e-05,
|
| 583 |
-
"loss": 0.0173,
|
| 584 |
-
"step": 840
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 2.86,
|
| 588 |
-
"learning_rate": 9.876543209876543e-06,
|
| 589 |
-
"loss": 0.0042,
|
| 590 |
-
"step": 850
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 2.9,
|
| 594 |
-
"learning_rate": 7.631874298540966e-06,
|
| 595 |
-
"loss": 0.008,
|
| 596 |
-
"step": 860
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 2.93,
|
| 600 |
-
"learning_rate": 5.387205387205387e-06,
|
| 601 |
-
"loss": 0.0648,
|
| 602 |
-
"step": 870
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 2.96,
|
| 606 |
-
"learning_rate": 3.1425364758698095e-06,
|
| 607 |
-
"loss": 0.0389,
|
| 608 |
-
"step": 880
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 3.0,
|
| 612 |
-
"learning_rate": 8.978675645342313e-07,
|
| 613 |
-
"loss": 0.0332,
|
| 614 |
-
"step": 890
|
| 615 |
-
},
|
| 616 |
{
|
| 617 |
"epoch": 3.0,
|
| 618 |
-
"step":
|
| 619 |
-
"total_flos":
|
| 620 |
-
"train_loss": 0.
|
| 621 |
-
"train_runtime":
|
| 622 |
-
"train_samples_per_second":
|
| 623 |
-
"train_steps_per_second": 1.
|
| 624 |
}
|
| 625 |
],
|
| 626 |
-
"max_steps":
|
| 627 |
"num_train_epochs": 3,
|
| 628 |
-
"total_flos":
|
| 629 |
"trial_name": null,
|
| 630 |
"trial_params": null
|
| 631 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.4047260880470276,
|
| 3 |
+
"best_model_checkpoint": "./output/vit-base-riego/checkpoint-300",
|
| 4 |
"epoch": 3.0,
|
| 5 |
+
"global_step": 378,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
+
"epoch": 0.08,
|
| 12 |
+
"learning_rate": 0.0001947089947089947,
|
| 13 |
+
"loss": 0.5579,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
+
"epoch": 0.16,
|
| 18 |
+
"learning_rate": 0.00018941798941798943,
|
| 19 |
+
"loss": 0.3376,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
+
"epoch": 0.24,
|
| 24 |
+
"learning_rate": 0.00018412698412698412,
|
| 25 |
+
"loss": 0.2934,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.32,
|
| 30 |
+
"learning_rate": 0.00017883597883597884,
|
| 31 |
+
"loss": 0.3096,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
+
"epoch": 0.4,
|
| 36 |
+
"learning_rate": 0.00017354497354497354,
|
| 37 |
+
"loss": 0.2475,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.48,
|
| 42 |
+
"learning_rate": 0.00016825396825396826,
|
| 43 |
+
"loss": 0.173,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.56,
|
| 48 |
+
"learning_rate": 0.00016296296296296295,
|
| 49 |
+
"loss": 0.3087,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.63,
|
| 54 |
+
"learning_rate": 0.00015820105820105822,
|
| 55 |
+
"loss": 0.3236,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
+
"epoch": 0.71,
|
| 60 |
+
"learning_rate": 0.0001529100529100529,
|
| 61 |
+
"loss": 0.2005,
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
+
"epoch": 0.79,
|
| 66 |
+
"learning_rate": 0.00014761904761904763,
|
| 67 |
+
"loss": 0.249,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
+
"epoch": 0.79,
|
| 72 |
+
"eval_f1": 0.31901840490797545,
|
| 73 |
+
"eval_loss": 0.6620292067527771,
|
| 74 |
+
"eval_runtime": 7.2847,
|
| 75 |
+
"eval_samples_per_second": 132.882,
|
| 76 |
+
"eval_steps_per_second": 16.61,
|
| 77 |
"step": 100
|
| 78 |
},
|
| 79 |
{
|
| 80 |
+
"epoch": 0.87,
|
| 81 |
+
"learning_rate": 0.00014232804232804233,
|
| 82 |
+
"loss": 0.2803,
|
| 83 |
"step": 110
|
| 84 |
},
|
| 85 |
{
|
| 86 |
+
"epoch": 0.95,
|
| 87 |
+
"learning_rate": 0.00013703703703703705,
|
| 88 |
+
"loss": 0.163,
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
+
"epoch": 1.03,
|
| 93 |
+
"learning_rate": 0.00013174603174603174,
|
| 94 |
+
"loss": 0.2306,
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
+
"epoch": 1.11,
|
| 99 |
+
"learning_rate": 0.00012645502645502646,
|
| 100 |
+
"loss": 0.0607,
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
+
"epoch": 1.19,
|
| 105 |
+
"learning_rate": 0.00012116402116402117,
|
| 106 |
+
"loss": 0.1109,
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 1.27,
|
| 111 |
+
"learning_rate": 0.0001158730158730159,
|
| 112 |
+
"loss": 0.2241,
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
+
"epoch": 1.35,
|
| 117 |
+
"learning_rate": 0.00011058201058201059,
|
| 118 |
+
"loss": 0.1394,
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
+
"epoch": 1.43,
|
| 123 |
+
"learning_rate": 0.00010529100529100531,
|
| 124 |
+
"loss": 0.153,
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
+
"epoch": 1.51,
|
| 129 |
+
"learning_rate": 0.0001,
|
| 130 |
+
"loss": 0.1361,
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
+
"epoch": 1.59,
|
| 135 |
+
"learning_rate": 9.470899470899471e-05,
|
| 136 |
+
"loss": 0.0348,
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
+
"epoch": 1.59,
|
| 141 |
+
"eval_f1": 0.427536231884058,
|
| 142 |
+
"eval_loss": 0.514543354511261,
|
| 143 |
+
"eval_runtime": 9.2968,
|
| 144 |
+
"eval_samples_per_second": 104.122,
|
| 145 |
+
"eval_steps_per_second": 13.015,
|
| 146 |
"step": 200
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 1.67,
|
| 150 |
+
"learning_rate": 8.941798941798942e-05,
|
| 151 |
+
"loss": 0.1362,
|
| 152 |
"step": 210
|
| 153 |
},
|
| 154 |
{
|
| 155 |
+
"epoch": 1.75,
|
| 156 |
+
"learning_rate": 8.412698412698413e-05,
|
| 157 |
+
"loss": 0.1059,
|
| 158 |
"step": 220
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 1.83,
|
| 162 |
+
"learning_rate": 7.883597883597884e-05,
|
| 163 |
+
"loss": 0.1051,
|
| 164 |
"step": 230
|
| 165 |
},
|
| 166 |
{
|
| 167 |
+
"epoch": 1.9,
|
| 168 |
+
"learning_rate": 7.354497354497355e-05,
|
| 169 |
+
"loss": 0.2335,
|
| 170 |
"step": 240
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 1.98,
|
| 174 |
+
"learning_rate": 6.825396825396825e-05,
|
| 175 |
+
"loss": 0.0877,
|
| 176 |
"step": 250
|
| 177 |
},
|
| 178 |
{
|
| 179 |
+
"epoch": 2.06,
|
| 180 |
+
"learning_rate": 6.296296296296296e-05,
|
| 181 |
+
"loss": 0.0765,
|
| 182 |
"step": 260
|
| 183 |
},
|
| 184 |
{
|
| 185 |
+
"epoch": 2.14,
|
| 186 |
+
"learning_rate": 5.7671957671957676e-05,
|
| 187 |
+
"loss": 0.0409,
|
| 188 |
"step": 270
|
| 189 |
},
|
| 190 |
{
|
| 191 |
+
"epoch": 2.22,
|
| 192 |
+
"learning_rate": 5.2380952380952384e-05,
|
| 193 |
+
"loss": 0.0978,
|
| 194 |
"step": 280
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 2.3,
|
| 198 |
+
"learning_rate": 4.708994708994709e-05,
|
| 199 |
+
"loss": 0.048,
|
| 200 |
"step": 290
|
| 201 |
},
|
| 202 |
{
|
| 203 |
+
"epoch": 2.38,
|
| 204 |
+
"learning_rate": 4.17989417989418e-05,
|
| 205 |
+
"loss": 0.0135,
|
| 206 |
"step": 300
|
| 207 |
},
|
| 208 |
{
|
| 209 |
+
"epoch": 2.38,
|
| 210 |
+
"eval_f1": 0.580441640378549,
|
| 211 |
+
"eval_loss": 0.4047260880470276,
|
| 212 |
+
"eval_runtime": 9.0047,
|
| 213 |
+
"eval_samples_per_second": 107.499,
|
| 214 |
+
"eval_steps_per_second": 13.437,
|
| 215 |
"step": 300
|
| 216 |
},
|
| 217 |
{
|
| 218 |
+
"epoch": 2.46,
|
| 219 |
+
"learning_rate": 3.650793650793651e-05,
|
| 220 |
+
"loss": 0.0093,
|
| 221 |
"step": 310
|
| 222 |
},
|
| 223 |
{
|
| 224 |
+
"epoch": 2.54,
|
| 225 |
+
"learning_rate": 3.121693121693122e-05,
|
| 226 |
+
"loss": 0.0163,
|
| 227 |
"step": 320
|
| 228 |
},
|
| 229 |
{
|
| 230 |
+
"epoch": 2.62,
|
| 231 |
+
"learning_rate": 2.5925925925925925e-05,
|
| 232 |
+
"loss": 0.0783,
|
| 233 |
"step": 330
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 2.7,
|
| 237 |
+
"learning_rate": 2.0634920634920636e-05,
|
| 238 |
+
"loss": 0.0288,
|
| 239 |
"step": 340
|
| 240 |
},
|
| 241 |
{
|
| 242 |
+
"epoch": 2.78,
|
| 243 |
+
"learning_rate": 1.5343915343915344e-05,
|
| 244 |
+
"loss": 0.033,
|
| 245 |
"step": 350
|
| 246 |
},
|
| 247 |
{
|
| 248 |
+
"epoch": 2.86,
|
| 249 |
+
"learning_rate": 1.0052910052910053e-05,
|
| 250 |
+
"loss": 0.0308,
|
| 251 |
"step": 360
|
| 252 |
},
|
| 253 |
{
|
| 254 |
+
"epoch": 2.94,
|
| 255 |
+
"learning_rate": 4.7619047619047615e-06,
|
| 256 |
+
"loss": 0.0293,
|
| 257 |
"step": 370
|
| 258 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 259 |
{
|
| 260 |
"epoch": 3.0,
|
| 261 |
+
"step": 378,
|
| 262 |
+
"total_flos": 4.665792694694584e+17,
|
| 263 |
+
"train_loss": 0.15108026370958047,
|
| 264 |
+
"train_runtime": 222.8357,
|
| 265 |
+
"train_samples_per_second": 27.02,
|
| 266 |
+
"train_steps_per_second": 1.696
|
| 267 |
}
|
| 268 |
],
|
| 269 |
+
"max_steps": 378,
|
| 270 |
"num_train_epochs": 3,
|
| 271 |
+
"total_flos": 4.665792694694584e+17,
|
| 272 |
"trial_name": null,
|
| 273 |
"trial_params": null
|
| 274 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3515
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:34043900b560431430b7737d36e5976c630010556c272b1438037be0984bfc67
|
| 3 |
size 3515
|