Training in progress, step 500, checkpoint
Browse files
last-checkpoint/adapter_config.json
CHANGED
|
@@ -12,12 +12,12 @@
|
|
| 12 |
"lora_dropout": 0.05,
|
| 13 |
"modules_to_save": null,
|
| 14 |
"peft_type": "LORA",
|
| 15 |
-
"r":
|
| 16 |
"rank_pattern": {},
|
| 17 |
"revision": null,
|
| 18 |
"target_modules": [
|
| 19 |
-
"q_proj",
|
| 20 |
"k_proj",
|
|
|
|
| 21 |
"v_proj"
|
| 22 |
],
|
| 23 |
"task_type": "CAUSAL_LM"
|
|
|
|
| 12 |
"lora_dropout": 0.05,
|
| 13 |
"modules_to_save": null,
|
| 14 |
"peft_type": "LORA",
|
| 15 |
+
"r": 16,
|
| 16 |
"rank_pattern": {},
|
| 17 |
"revision": null,
|
| 18 |
"target_modules": [
|
|
|
|
| 19 |
"k_proj",
|
| 20 |
+
"q_proj",
|
| 21 |
"v_proj"
|
| 22 |
],
|
| 23 |
"task_type": "CAUSAL_LM"
|
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a8a809056b973f3df74a8e0401d2f6b3f48855ca88f92c074a282b2fc872456
|
| 3 |
+
size 5323528
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cace089f6f62f3f0c39575883a3ef27ad996c274acf20a6fd9d2d6ee12a92918
|
| 3 |
+
size 10707706
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5e4a99a8f1604eb88b6b6efd9bc6ced6e284090d6037a758d0561d10a951f004
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6c82bedf0a611f290596df2fde142fbda2afa059d93dc846b92ee4f876380a79
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "bart_lora_outputs\\checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 100,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -11,4089 +11,349 @@
|
|
| 11 |
{
|
| 12 |
"epoch": 0.02,
|
| 13 |
"learning_rate": 2e-05,
|
| 14 |
-
"loss": 2.
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.03,
|
| 19 |
"learning_rate": 4e-05,
|
| 20 |
-
"loss": 2.
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.05,
|
| 25 |
"learning_rate": 6e-05,
|
| 26 |
-
"loss": 2.
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.07,
|
| 31 |
"learning_rate": 8e-05,
|
| 32 |
-
"loss": 2.
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.08,
|
| 37 |
"learning_rate": 0.0001,
|
| 38 |
-
"loss": 2.
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.1,
|
| 43 |
"learning_rate": 0.00012,
|
| 44 |
-
"loss": 1.
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.11,
|
| 49 |
"learning_rate": 0.00014000000000000001,
|
| 50 |
-
"loss": 1.
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.13,
|
| 55 |
"learning_rate": 0.00016,
|
| 56 |
-
"loss": 1.
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.15,
|
| 61 |
"learning_rate": 0.00017999999999999998,
|
| 62 |
-
"loss": 1.
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.16,
|
| 67 |
"learning_rate": 0.0002,
|
| 68 |
-
"loss": 1.
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.16,
|
| 73 |
-
"eval_loss": 0.
|
| 74 |
-
"eval_runtime": 5.
|
| 75 |
-
"eval_samples_per_second":
|
| 76 |
-
"eval_steps_per_second":
|
| 77 |
"step": 100
|
| 78 |
},
|
| 79 |
{
|
| 80 |
"epoch": 0.18,
|
| 81 |
"learning_rate": 0.00022,
|
| 82 |
-
"loss":
|
| 83 |
"step": 110
|
| 84 |
},
|
| 85 |
{
|
| 86 |
"epoch": 0.2,
|
| 87 |
"learning_rate": 0.00024,
|
| 88 |
-
"loss":
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 0.21,
|
| 93 |
"learning_rate": 0.00026000000000000003,
|
| 94 |
-
"loss": 0.
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 0.23,
|
| 99 |
"learning_rate": 0.00028000000000000003,
|
| 100 |
-
"loss": 0.
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.24,
|
| 105 |
"learning_rate": 0.0003,
|
| 106 |
-
"loss": 0.
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.26,
|
| 111 |
"learning_rate": 0.00032,
|
| 112 |
-
"loss": 0.
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 0.28,
|
| 117 |
"learning_rate": 0.00034,
|
| 118 |
-
"loss": 0.
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 0.29,
|
| 123 |
"learning_rate": 0.00035999999999999997,
|
| 124 |
-
"loss": 0.
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 0.31,
|
| 129 |
"learning_rate": 0.00038,
|
| 130 |
-
"loss": 0.
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 0.33,
|
| 135 |
"learning_rate": 0.0004,
|
| 136 |
-
"loss": 0.
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 0.33,
|
| 141 |
-
"eval_loss": 0.
|
| 142 |
-
"eval_runtime":
|
| 143 |
-
"eval_samples_per_second":
|
| 144 |
-
"eval_steps_per_second":
|
| 145 |
"step": 200
|
| 146 |
},
|
| 147 |
{
|
| 148 |
"epoch": 0.34,
|
| 149 |
"learning_rate": 0.00042,
|
| 150 |
-
"loss": 0.
|
| 151 |
"step": 210
|
| 152 |
},
|
| 153 |
{
|
| 154 |
"epoch": 0.36,
|
| 155 |
"learning_rate": 0.00044,
|
| 156 |
-
"loss": 0.
|
| 157 |
"step": 220
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.38,
|
| 161 |
"learning_rate": 0.00046,
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 230
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.39,
|
| 167 |
"learning_rate": 0.00048,
|
| 168 |
-
"loss": 0.
|
| 169 |
"step": 240
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 0.41,
|
| 173 |
"learning_rate": 0.0005,
|
| 174 |
-
"loss": 0.
|
| 175 |
"step": 250
|
| 176 |
},
|
| 177 |
{
|
| 178 |
"epoch": 0.42,
|
| 179 |
"learning_rate": 0.0005200000000000001,
|
| 180 |
-
"loss": 0.
|
| 181 |
"step": 260
|
| 182 |
},
|
| 183 |
{
|
| 184 |
"epoch": 0.44,
|
| 185 |
"learning_rate": 0.00054,
|
| 186 |
-
"loss": 0.
|
| 187 |
"step": 270
|
| 188 |
},
|
| 189 |
{
|
| 190 |
"epoch": 0.46,
|
| 191 |
"learning_rate": 0.0005600000000000001,
|
| 192 |
-
"loss": 0.
|
| 193 |
"step": 280
|
| 194 |
},
|
| 195 |
{
|
| 196 |
"epoch": 0.47,
|
| 197 |
"learning_rate": 0.00058,
|
| 198 |
-
"loss": 0.
|
| 199 |
"step": 290
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.49,
|
| 203 |
"learning_rate": 0.0006,
|
| 204 |
-
"loss": 0.
|
| 205 |
"step": 300
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.49,
|
| 209 |
-
"eval_loss": 0.
|
| 210 |
-
"eval_runtime":
|
| 211 |
-
"eval_samples_per_second":
|
| 212 |
-
"eval_steps_per_second":
|
| 213 |
"step": 300
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.51,
|
| 217 |
"learning_rate": 0.00062,
|
| 218 |
-
"loss": 0.
|
| 219 |
"step": 310
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.52,
|
| 223 |
"learning_rate": 0.00064,
|
| 224 |
-
"loss": 0.
|
| 225 |
"step": 320
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.54,
|
| 229 |
"learning_rate": 0.00066,
|
| 230 |
-
"loss": 0.
|
| 231 |
"step": 330
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.55,
|
| 235 |
"learning_rate": 0.00068,
|
| 236 |
-
"loss": 0.
|
| 237 |
"step": 340
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.57,
|
| 241 |
"learning_rate": 0.0007,
|
| 242 |
-
"loss": 0.
|
| 243 |
"step": 350
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.59,
|
| 247 |
"learning_rate": 0.0007199999999999999,
|
| 248 |
-
"loss": 0.
|
| 249 |
"step": 360
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.6,
|
| 253 |
"learning_rate": 0.00074,
|
| 254 |
-
"loss": 0.
|
| 255 |
"step": 370
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.62,
|
| 259 |
"learning_rate": 0.00076,
|
| 260 |
-
"loss": 0.
|
| 261 |
"step": 380
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.64,
|
| 265 |
"learning_rate": 0.0007800000000000001,
|
| 266 |
-
"loss": 0.
|
| 267 |
"step": 390
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 0.65,
|
| 271 |
"learning_rate": 0.0008,
|
| 272 |
-
"loss": 0.
|
| 273 |
"step": 400
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 0.65,
|
| 277 |
-
"eval_loss": 0.
|
| 278 |
-
"eval_runtime":
|
| 279 |
-
"eval_samples_per_second":
|
| 280 |
-
"eval_steps_per_second":
|
| 281 |
"step": 400
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 0.67,
|
| 285 |
"learning_rate": 0.00082,
|
| 286 |
-
"loss": 0.
|
| 287 |
"step": 410
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 0.69,
|
| 291 |
"learning_rate": 0.00084,
|
| 292 |
-
"loss": 0.
|
| 293 |
"step": 420
|
| 294 |
},
|
| 295 |
{
|
| 296 |
"epoch": 0.7,
|
| 297 |
"learning_rate": 0.00086,
|
| 298 |
-
"loss": 0.
|
| 299 |
"step": 430
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 0.72,
|
| 303 |
"learning_rate": 0.00088,
|
| 304 |
-
"loss": 0.
|
| 305 |
"step": 440
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.73,
|
| 309 |
"learning_rate": 0.0009000000000000001,
|
| 310 |
-
"loss": 0.
|
| 311 |
"step": 450
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.75,
|
| 315 |
"learning_rate": 0.00092,
|
| 316 |
-
"loss": 0.
|
| 317 |
"step": 460
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"epoch": 0.77,
|
| 321 |
"learning_rate": 0.00094,
|
| 322 |
-
"loss": 0.
|
| 323 |
"step": 470
|
| 324 |
},
|
| 325 |
{
|
| 326 |
"epoch": 0.78,
|
| 327 |
"learning_rate": 0.00096,
|
| 328 |
-
"loss": 0.
|
| 329 |
"step": 480
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 0.8,
|
| 333 |
"learning_rate": 0.00098,
|
| 334 |
-
"loss": 0.
|
| 335 |
"step": 490
|
| 336 |
},
|
| 337 |
{
|
| 338 |
"epoch": 0.82,
|
| 339 |
"learning_rate": 0.001,
|
| 340 |
-
"loss": 0.
|
| 341 |
"step": 500
|
| 342 |
},
|
| 343 |
{
|
| 344 |
"epoch": 0.82,
|
| 345 |
-
"eval_loss": 0.
|
| 346 |
-
"eval_runtime":
|
| 347 |
-
"eval_samples_per_second":
|
| 348 |
-
"eval_steps_per_second":
|
| 349 |
"step": 500
|
| 350 |
-
},
|
| 351 |
-
{
|
| 352 |
-
"epoch": 0.83,
|
| 353 |
-
"learning_rate": 0.0009982238010657195,
|
| 354 |
-
"loss": 0.6562,
|
| 355 |
-
"step": 510
|
| 356 |
-
},
|
| 357 |
-
{
|
| 358 |
-
"epoch": 0.85,
|
| 359 |
-
"learning_rate": 0.0009964476021314388,
|
| 360 |
-
"loss": 0.6576,
|
| 361 |
-
"step": 520
|
| 362 |
-
},
|
| 363 |
-
{
|
| 364 |
-
"epoch": 0.86,
|
| 365 |
-
"learning_rate": 0.000994671403197158,
|
| 366 |
-
"loss": 0.6182,
|
| 367 |
-
"step": 530
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.88,
|
| 371 |
-
"learning_rate": 0.0009928952042628776,
|
| 372 |
-
"loss": 0.6067,
|
| 373 |
-
"step": 540
|
| 374 |
-
},
|
| 375 |
-
{
|
| 376 |
-
"epoch": 0.9,
|
| 377 |
-
"learning_rate": 0.0009911190053285969,
|
| 378 |
-
"loss": 0.5993,
|
| 379 |
-
"step": 550
|
| 380 |
-
},
|
| 381 |
-
{
|
| 382 |
-
"epoch": 0.91,
|
| 383 |
-
"learning_rate": 0.0009893428063943162,
|
| 384 |
-
"loss": 0.6218,
|
| 385 |
-
"step": 560
|
| 386 |
-
},
|
| 387 |
-
{
|
| 388 |
-
"epoch": 0.93,
|
| 389 |
-
"learning_rate": 0.0009875666074600357,
|
| 390 |
-
"loss": 0.6154,
|
| 391 |
-
"step": 570
|
| 392 |
-
},
|
| 393 |
-
{
|
| 394 |
-
"epoch": 0.95,
|
| 395 |
-
"learning_rate": 0.000985790408525755,
|
| 396 |
-
"loss": 0.612,
|
| 397 |
-
"step": 580
|
| 398 |
-
},
|
| 399 |
-
{
|
| 400 |
-
"epoch": 0.96,
|
| 401 |
-
"learning_rate": 0.0009840142095914742,
|
| 402 |
-
"loss": 0.6217,
|
| 403 |
-
"step": 590
|
| 404 |
-
},
|
| 405 |
-
{
|
| 406 |
-
"epoch": 0.98,
|
| 407 |
-
"learning_rate": 0.0009822380106571937,
|
| 408 |
-
"loss": 0.6161,
|
| 409 |
-
"step": 600
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 0.98,
|
| 413 |
-
"eval_loss": 0.5353884100914001,
|
| 414 |
-
"eval_runtime": 5.655,
|
| 415 |
-
"eval_samples_per_second": 207.073,
|
| 416 |
-
"eval_steps_per_second": 13.086,
|
| 417 |
-
"step": 600
|
| 418 |
-
},
|
| 419 |
-
{
|
| 420 |
-
"epoch": 1.0,
|
| 421 |
-
"learning_rate": 0.000980461811722913,
|
| 422 |
-
"loss": 0.6054,
|
| 423 |
-
"step": 610
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 1.01,
|
| 427 |
-
"learning_rate": 0.0009786856127886323,
|
| 428 |
-
"loss": 0.5417,
|
| 429 |
-
"step": 620
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 1.03,
|
| 433 |
-
"learning_rate": 0.0009769094138543518,
|
| 434 |
-
"loss": 0.5902,
|
| 435 |
-
"step": 630
|
| 436 |
-
},
|
| 437 |
-
{
|
| 438 |
-
"epoch": 1.04,
|
| 439 |
-
"learning_rate": 0.0009751332149200711,
|
| 440 |
-
"loss": 0.6063,
|
| 441 |
-
"step": 640
|
| 442 |
-
},
|
| 443 |
-
{
|
| 444 |
-
"epoch": 1.06,
|
| 445 |
-
"learning_rate": 0.0009733570159857904,
|
| 446 |
-
"loss": 0.543,
|
| 447 |
-
"step": 650
|
| 448 |
-
},
|
| 449 |
-
{
|
| 450 |
-
"epoch": 1.08,
|
| 451 |
-
"learning_rate": 0.0009715808170515098,
|
| 452 |
-
"loss": 0.5668,
|
| 453 |
-
"step": 660
|
| 454 |
-
},
|
| 455 |
-
{
|
| 456 |
-
"epoch": 1.09,
|
| 457 |
-
"learning_rate": 0.0009698046181172292,
|
| 458 |
-
"loss": 0.5594,
|
| 459 |
-
"step": 670
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 1.11,
|
| 463 |
-
"learning_rate": 0.0009680284191829485,
|
| 464 |
-
"loss": 0.5758,
|
| 465 |
-
"step": 680
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"epoch": 1.13,
|
| 469 |
-
"learning_rate": 0.0009662522202486678,
|
| 470 |
-
"loss": 0.5845,
|
| 471 |
-
"step": 690
|
| 472 |
-
},
|
| 473 |
-
{
|
| 474 |
-
"epoch": 1.14,
|
| 475 |
-
"learning_rate": 0.0009644760213143872,
|
| 476 |
-
"loss": 0.574,
|
| 477 |
-
"step": 700
|
| 478 |
-
},
|
| 479 |
-
{
|
| 480 |
-
"epoch": 1.14,
|
| 481 |
-
"eval_loss": 0.5269472002983093,
|
| 482 |
-
"eval_runtime": 5.818,
|
| 483 |
-
"eval_samples_per_second": 201.271,
|
| 484 |
-
"eval_steps_per_second": 12.719,
|
| 485 |
-
"step": 700
|
| 486 |
-
},
|
| 487 |
-
{
|
| 488 |
-
"epoch": 1.16,
|
| 489 |
-
"learning_rate": 0.0009626998223801065,
|
| 490 |
-
"loss": 0.5216,
|
| 491 |
-
"step": 710
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 1.17,
|
| 495 |
-
"learning_rate": 0.0009609236234458259,
|
| 496 |
-
"loss": 0.6191,
|
| 497 |
-
"step": 720
|
| 498 |
-
},
|
| 499 |
-
{
|
| 500 |
-
"epoch": 1.19,
|
| 501 |
-
"learning_rate": 0.0009591474245115453,
|
| 502 |
-
"loss": 0.5832,
|
| 503 |
-
"step": 730
|
| 504 |
-
},
|
| 505 |
-
{
|
| 506 |
-
"epoch": 1.21,
|
| 507 |
-
"learning_rate": 0.0009573712255772646,
|
| 508 |
-
"loss": 0.5622,
|
| 509 |
-
"step": 740
|
| 510 |
-
},
|
| 511 |
-
{
|
| 512 |
-
"epoch": 1.22,
|
| 513 |
-
"learning_rate": 0.000955595026642984,
|
| 514 |
-
"loss": 0.599,
|
| 515 |
-
"step": 750
|
| 516 |
-
},
|
| 517 |
-
{
|
| 518 |
-
"epoch": 1.24,
|
| 519 |
-
"learning_rate": 0.0009538188277087034,
|
| 520 |
-
"loss": 0.564,
|
| 521 |
-
"step": 760
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 1.26,
|
| 525 |
-
"learning_rate": 0.0009520426287744227,
|
| 526 |
-
"loss": 0.5996,
|
| 527 |
-
"step": 770
|
| 528 |
-
},
|
| 529 |
-
{
|
| 530 |
-
"epoch": 1.27,
|
| 531 |
-
"learning_rate": 0.0009502664298401421,
|
| 532 |
-
"loss": 0.5337,
|
| 533 |
-
"step": 780
|
| 534 |
-
},
|
| 535 |
-
{
|
| 536 |
-
"epoch": 1.29,
|
| 537 |
-
"learning_rate": 0.0009484902309058615,
|
| 538 |
-
"loss": 0.6051,
|
| 539 |
-
"step": 790
|
| 540 |
-
},
|
| 541 |
-
{
|
| 542 |
-
"epoch": 1.31,
|
| 543 |
-
"learning_rate": 0.0009467140319715807,
|
| 544 |
-
"loss": 0.6023,
|
| 545 |
-
"step": 800
|
| 546 |
-
},
|
| 547 |
-
{
|
| 548 |
-
"epoch": 1.31,
|
| 549 |
-
"eval_loss": 0.5132582187652588,
|
| 550 |
-
"eval_runtime": 5.778,
|
| 551 |
-
"eval_samples_per_second": 202.665,
|
| 552 |
-
"eval_steps_per_second": 12.807,
|
| 553 |
-
"step": 800
|
| 554 |
-
},
|
| 555 |
-
{
|
| 556 |
-
"epoch": 1.32,
|
| 557 |
-
"learning_rate": 0.0009449378330373001,
|
| 558 |
-
"loss": 0.5531,
|
| 559 |
-
"step": 810
|
| 560 |
-
},
|
| 561 |
-
{
|
| 562 |
-
"epoch": 1.34,
|
| 563 |
-
"learning_rate": 0.0009431616341030196,
|
| 564 |
-
"loss": 0.5623,
|
| 565 |
-
"step": 820
|
| 566 |
-
},
|
| 567 |
-
{
|
| 568 |
-
"epoch": 1.35,
|
| 569 |
-
"learning_rate": 0.0009413854351687389,
|
| 570 |
-
"loss": 0.5389,
|
| 571 |
-
"step": 830
|
| 572 |
-
},
|
| 573 |
-
{
|
| 574 |
-
"epoch": 1.37,
|
| 575 |
-
"learning_rate": 0.0009396092362344583,
|
| 576 |
-
"loss": 0.5446,
|
| 577 |
-
"step": 840
|
| 578 |
-
},
|
| 579 |
-
{
|
| 580 |
-
"epoch": 1.39,
|
| 581 |
-
"learning_rate": 0.0009378330373001777,
|
| 582 |
-
"loss": 0.5512,
|
| 583 |
-
"step": 850
|
| 584 |
-
},
|
| 585 |
-
{
|
| 586 |
-
"epoch": 1.4,
|
| 587 |
-
"learning_rate": 0.000936056838365897,
|
| 588 |
-
"loss": 0.571,
|
| 589 |
-
"step": 860
|
| 590 |
-
},
|
| 591 |
-
{
|
| 592 |
-
"epoch": 1.42,
|
| 593 |
-
"learning_rate": 0.0009342806394316164,
|
| 594 |
-
"loss": 0.5415,
|
| 595 |
-
"step": 870
|
| 596 |
-
},
|
| 597 |
-
{
|
| 598 |
-
"epoch": 1.44,
|
| 599 |
-
"learning_rate": 0.0009325044404973358,
|
| 600 |
-
"loss": 0.5807,
|
| 601 |
-
"step": 880
|
| 602 |
-
},
|
| 603 |
-
{
|
| 604 |
-
"epoch": 1.45,
|
| 605 |
-
"learning_rate": 0.0009307282415630552,
|
| 606 |
-
"loss": 0.5304,
|
| 607 |
-
"step": 890
|
| 608 |
-
},
|
| 609 |
-
{
|
| 610 |
-
"epoch": 1.47,
|
| 611 |
-
"learning_rate": 0.0009289520426287745,
|
| 612 |
-
"loss": 0.5468,
|
| 613 |
-
"step": 900
|
| 614 |
-
},
|
| 615 |
-
{
|
| 616 |
-
"epoch": 1.47,
|
| 617 |
-
"eval_loss": 0.4914248585700989,
|
| 618 |
-
"eval_runtime": 5.762,
|
| 619 |
-
"eval_samples_per_second": 203.227,
|
| 620 |
-
"eval_steps_per_second": 12.843,
|
| 621 |
-
"step": 900
|
| 622 |
-
},
|
| 623 |
-
{
|
| 624 |
-
"epoch": 1.48,
|
| 625 |
-
"learning_rate": 0.0009271758436944939,
|
| 626 |
-
"loss": 0.6005,
|
| 627 |
-
"step": 910
|
| 628 |
-
},
|
| 629 |
-
{
|
| 630 |
-
"epoch": 1.5,
|
| 631 |
-
"learning_rate": 0.0009253996447602132,
|
| 632 |
-
"loss": 0.59,
|
| 633 |
-
"step": 920
|
| 634 |
-
},
|
| 635 |
-
{
|
| 636 |
-
"epoch": 1.52,
|
| 637 |
-
"learning_rate": 0.0009236234458259325,
|
| 638 |
-
"loss": 0.5487,
|
| 639 |
-
"step": 930
|
| 640 |
-
},
|
| 641 |
-
{
|
| 642 |
-
"epoch": 1.53,
|
| 643 |
-
"learning_rate": 0.0009218472468916519,
|
| 644 |
-
"loss": 0.5783,
|
| 645 |
-
"step": 940
|
| 646 |
-
},
|
| 647 |
-
{
|
| 648 |
-
"epoch": 1.55,
|
| 649 |
-
"learning_rate": 0.0009200710479573713,
|
| 650 |
-
"loss": 0.5579,
|
| 651 |
-
"step": 950
|
| 652 |
-
},
|
| 653 |
-
{
|
| 654 |
-
"epoch": 1.57,
|
| 655 |
-
"learning_rate": 0.0009182948490230906,
|
| 656 |
-
"loss": 0.5618,
|
| 657 |
-
"step": 960
|
| 658 |
-
},
|
| 659 |
-
{
|
| 660 |
-
"epoch": 1.58,
|
| 661 |
-
"learning_rate": 0.00091651865008881,
|
| 662 |
-
"loss": 0.5434,
|
| 663 |
-
"step": 970
|
| 664 |
-
},
|
| 665 |
-
{
|
| 666 |
-
"epoch": 1.6,
|
| 667 |
-
"learning_rate": 0.0009147424511545294,
|
| 668 |
-
"loss": 0.5715,
|
| 669 |
-
"step": 980
|
| 670 |
-
},
|
| 671 |
-
{
|
| 672 |
-
"epoch": 1.62,
|
| 673 |
-
"learning_rate": 0.0009129662522202487,
|
| 674 |
-
"loss": 0.5257,
|
| 675 |
-
"step": 990
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 1.63,
|
| 679 |
-
"learning_rate": 0.0009111900532859681,
|
| 680 |
-
"loss": 0.5391,
|
| 681 |
-
"step": 1000
|
| 682 |
-
},
|
| 683 |
-
{
|
| 684 |
-
"epoch": 1.63,
|
| 685 |
-
"eval_loss": 0.4895668029785156,
|
| 686 |
-
"eval_runtime": 5.802,
|
| 687 |
-
"eval_samples_per_second": 201.826,
|
| 688 |
-
"eval_steps_per_second": 12.754,
|
| 689 |
-
"step": 1000
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 1.65,
|
| 693 |
-
"learning_rate": 0.0009094138543516875,
|
| 694 |
-
"loss": 0.5178,
|
| 695 |
-
"step": 1010
|
| 696 |
-
},
|
| 697 |
-
{
|
| 698 |
-
"epoch": 1.66,
|
| 699 |
-
"learning_rate": 0.0009076376554174067,
|
| 700 |
-
"loss": 0.5655,
|
| 701 |
-
"step": 1020
|
| 702 |
-
},
|
| 703 |
-
{
|
| 704 |
-
"epoch": 1.68,
|
| 705 |
-
"learning_rate": 0.0009058614564831261,
|
| 706 |
-
"loss": 0.5352,
|
| 707 |
-
"step": 1030
|
| 708 |
-
},
|
| 709 |
-
{
|
| 710 |
-
"epoch": 1.7,
|
| 711 |
-
"learning_rate": 0.0009040852575488455,
|
| 712 |
-
"loss": 0.5355,
|
| 713 |
-
"step": 1040
|
| 714 |
-
},
|
| 715 |
-
{
|
| 716 |
-
"epoch": 1.71,
|
| 717 |
-
"learning_rate": 0.0009023090586145648,
|
| 718 |
-
"loss": 0.5322,
|
| 719 |
-
"step": 1050
|
| 720 |
-
},
|
| 721 |
-
{
|
| 722 |
-
"epoch": 1.73,
|
| 723 |
-
"learning_rate": 0.0009005328596802842,
|
| 724 |
-
"loss": 0.5427,
|
| 725 |
-
"step": 1060
|
| 726 |
-
},
|
| 727 |
-
{
|
| 728 |
-
"epoch": 1.75,
|
| 729 |
-
"learning_rate": 0.0008987566607460036,
|
| 730 |
-
"loss": 0.5531,
|
| 731 |
-
"step": 1070
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 1.76,
|
| 735 |
-
"learning_rate": 0.0008969804618117229,
|
| 736 |
-
"loss": 0.5485,
|
| 737 |
-
"step": 1080
|
| 738 |
-
},
|
| 739 |
-
{
|
| 740 |
-
"epoch": 1.78,
|
| 741 |
-
"learning_rate": 0.0008952042628774423,
|
| 742 |
-
"loss": 0.5366,
|
| 743 |
-
"step": 1090
|
| 744 |
-
},
|
| 745 |
-
{
|
| 746 |
-
"epoch": 1.79,
|
| 747 |
-
"learning_rate": 0.0008934280639431617,
|
| 748 |
-
"loss": 0.5455,
|
| 749 |
-
"step": 1100
|
| 750 |
-
},
|
| 751 |
-
{
|
| 752 |
-
"epoch": 1.79,
|
| 753 |
-
"eval_loss": 0.4829932451248169,
|
| 754 |
-
"eval_runtime": 5.871,
|
| 755 |
-
"eval_samples_per_second": 199.454,
|
| 756 |
-
"eval_steps_per_second": 12.604,
|
| 757 |
-
"step": 1100
|
| 758 |
-
},
|
| 759 |
-
{
|
| 760 |
-
"epoch": 1.81,
|
| 761 |
-
"learning_rate": 0.000891651865008881,
|
| 762 |
-
"loss": 0.5378,
|
| 763 |
-
"step": 1110
|
| 764 |
-
},
|
| 765 |
-
{
|
| 766 |
-
"epoch": 1.83,
|
| 767 |
-
"learning_rate": 0.0008898756660746004,
|
| 768 |
-
"loss": 0.5516,
|
| 769 |
-
"step": 1120
|
| 770 |
-
},
|
| 771 |
-
{
|
| 772 |
-
"epoch": 1.84,
|
| 773 |
-
"learning_rate": 0.0008880994671403197,
|
| 774 |
-
"loss": 0.541,
|
| 775 |
-
"step": 1130
|
| 776 |
-
},
|
| 777 |
-
{
|
| 778 |
-
"epoch": 1.86,
|
| 779 |
-
"learning_rate": 0.0008863232682060391,
|
| 780 |
-
"loss": 0.55,
|
| 781 |
-
"step": 1140
|
| 782 |
-
},
|
| 783 |
-
{
|
| 784 |
-
"epoch": 1.88,
|
| 785 |
-
"learning_rate": 0.0008845470692717584,
|
| 786 |
-
"loss": 0.5451,
|
| 787 |
-
"step": 1150
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"epoch": 1.89,
|
| 791 |
-
"learning_rate": 0.0008827708703374778,
|
| 792 |
-
"loss": 0.5508,
|
| 793 |
-
"step": 1160
|
| 794 |
-
},
|
| 795 |
-
{
|
| 796 |
-
"epoch": 1.91,
|
| 797 |
-
"learning_rate": 0.0008809946714031972,
|
| 798 |
-
"loss": 0.4816,
|
| 799 |
-
"step": 1170
|
| 800 |
-
},
|
| 801 |
-
{
|
| 802 |
-
"epoch": 1.92,
|
| 803 |
-
"learning_rate": 0.0008792184724689165,
|
| 804 |
-
"loss": 0.5529,
|
| 805 |
-
"step": 1180
|
| 806 |
-
},
|
| 807 |
-
{
|
| 808 |
-
"epoch": 1.94,
|
| 809 |
-
"learning_rate": 0.0008774422735346359,
|
| 810 |
-
"loss": 0.5119,
|
| 811 |
-
"step": 1190
|
| 812 |
-
},
|
| 813 |
-
{
|
| 814 |
-
"epoch": 1.96,
|
| 815 |
-
"learning_rate": 0.0008756660746003553,
|
| 816 |
-
"loss": 0.5431,
|
| 817 |
-
"step": 1200
|
| 818 |
-
},
|
| 819 |
-
{
|
| 820 |
-
"epoch": 1.96,
|
| 821 |
-
"eval_loss": 0.4757298529148102,
|
| 822 |
-
"eval_runtime": 5.601,
|
| 823 |
-
"eval_samples_per_second": 209.069,
|
| 824 |
-
"eval_steps_per_second": 13.212,
|
| 825 |
-
"step": 1200
|
| 826 |
-
},
|
| 827 |
-
{
|
| 828 |
-
"epoch": 1.97,
|
| 829 |
-
"learning_rate": 0.0008738898756660746,
|
| 830 |
-
"loss": 0.5315,
|
| 831 |
-
"step": 1210
|
| 832 |
-
},
|
| 833 |
-
{
|
| 834 |
-
"epoch": 1.99,
|
| 835 |
-
"learning_rate": 0.000872113676731794,
|
| 836 |
-
"loss": 0.5455,
|
| 837 |
-
"step": 1220
|
| 838 |
-
},
|
| 839 |
-
{
|
| 840 |
-
"epoch": 2.01,
|
| 841 |
-
"learning_rate": 0.0008703374777975134,
|
| 842 |
-
"loss": 0.5274,
|
| 843 |
-
"step": 1230
|
| 844 |
-
},
|
| 845 |
-
{
|
| 846 |
-
"epoch": 2.02,
|
| 847 |
-
"learning_rate": 0.0008685612788632326,
|
| 848 |
-
"loss": 0.5383,
|
| 849 |
-
"step": 1240
|
| 850 |
-
},
|
| 851 |
-
{
|
| 852 |
-
"epoch": 2.04,
|
| 853 |
-
"learning_rate": 0.000866785079928952,
|
| 854 |
-
"loss": 0.5185,
|
| 855 |
-
"step": 1250
|
| 856 |
-
},
|
| 857 |
-
{
|
| 858 |
-
"epoch": 2.06,
|
| 859 |
-
"learning_rate": 0.0008650088809946714,
|
| 860 |
-
"loss": 0.5093,
|
| 861 |
-
"step": 1260
|
| 862 |
-
},
|
| 863 |
-
{
|
| 864 |
-
"epoch": 2.07,
|
| 865 |
-
"learning_rate": 0.0008632326820603907,
|
| 866 |
-
"loss": 0.5237,
|
| 867 |
-
"step": 1270
|
| 868 |
-
},
|
| 869 |
-
{
|
| 870 |
-
"epoch": 2.09,
|
| 871 |
-
"learning_rate": 0.0008614564831261101,
|
| 872 |
-
"loss": 0.4849,
|
| 873 |
-
"step": 1280
|
| 874 |
-
},
|
| 875 |
-
{
|
| 876 |
-
"epoch": 2.1,
|
| 877 |
-
"learning_rate": 0.0008596802841918295,
|
| 878 |
-
"loss": 0.5198,
|
| 879 |
-
"step": 1290
|
| 880 |
-
},
|
| 881 |
-
{
|
| 882 |
-
"epoch": 2.12,
|
| 883 |
-
"learning_rate": 0.0008579040852575488,
|
| 884 |
-
"loss": 0.4962,
|
| 885 |
-
"step": 1300
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"epoch": 2.12,
|
| 889 |
-
"eval_loss": 0.4822671711444855,
|
| 890 |
-
"eval_runtime": 5.607,
|
| 891 |
-
"eval_samples_per_second": 208.845,
|
| 892 |
-
"eval_steps_per_second": 13.198,
|
| 893 |
-
"step": 1300
|
| 894 |
-
},
|
| 895 |
-
{
|
| 896 |
-
"epoch": 2.14,
|
| 897 |
-
"learning_rate": 0.0008561278863232682,
|
| 898 |
-
"loss": 0.5036,
|
| 899 |
-
"step": 1310
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"epoch": 2.15,
|
| 903 |
-
"learning_rate": 0.0008543516873889876,
|
| 904 |
-
"loss": 0.498,
|
| 905 |
-
"step": 1320
|
| 906 |
-
},
|
| 907 |
-
{
|
| 908 |
-
"epoch": 2.17,
|
| 909 |
-
"learning_rate": 0.0008525754884547069,
|
| 910 |
-
"loss": 0.494,
|
| 911 |
-
"step": 1330
|
| 912 |
-
},
|
| 913 |
-
{
|
| 914 |
-
"epoch": 2.19,
|
| 915 |
-
"learning_rate": 0.0008507992895204263,
|
| 916 |
-
"loss": 0.5209,
|
| 917 |
-
"step": 1340
|
| 918 |
-
},
|
| 919 |
-
{
|
| 920 |
-
"epoch": 2.2,
|
| 921 |
-
"learning_rate": 0.0008490230905861456,
|
| 922 |
-
"loss": 0.485,
|
| 923 |
-
"step": 1350
|
| 924 |
-
},
|
| 925 |
-
{
|
| 926 |
-
"epoch": 2.22,
|
| 927 |
-
"learning_rate": 0.000847246891651865,
|
| 928 |
-
"loss": 0.5423,
|
| 929 |
-
"step": 1360
|
| 930 |
-
},
|
| 931 |
-
{
|
| 932 |
-
"epoch": 2.23,
|
| 933 |
-
"learning_rate": 0.0008454706927175843,
|
| 934 |
-
"loss": 0.4856,
|
| 935 |
-
"step": 1370
|
| 936 |
-
},
|
| 937 |
-
{
|
| 938 |
-
"epoch": 2.25,
|
| 939 |
-
"learning_rate": 0.0008436944937833037,
|
| 940 |
-
"loss": 0.4983,
|
| 941 |
-
"step": 1380
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 2.27,
|
| 945 |
-
"learning_rate": 0.0008419182948490231,
|
| 946 |
-
"loss": 0.5286,
|
| 947 |
-
"step": 1390
|
| 948 |
-
},
|
| 949 |
-
{
|
| 950 |
-
"epoch": 2.28,
|
| 951 |
-
"learning_rate": 0.0008401420959147424,
|
| 952 |
-
"loss": 0.5206,
|
| 953 |
-
"step": 1400
|
| 954 |
-
},
|
| 955 |
-
{
|
| 956 |
-
"epoch": 2.28,
|
| 957 |
-
"eval_loss": 0.47338804602622986,
|
| 958 |
-
"eval_runtime": 5.569,
|
| 959 |
-
"eval_samples_per_second": 210.27,
|
| 960 |
-
"eval_steps_per_second": 13.288,
|
| 961 |
-
"step": 1400
|
| 962 |
-
},
|
| 963 |
-
{
|
| 964 |
-
"epoch": 2.3,
|
| 965 |
-
"learning_rate": 0.0008383658969804618,
|
| 966 |
-
"loss": 0.5231,
|
| 967 |
-
"step": 1410
|
| 968 |
-
},
|
| 969 |
-
{
|
| 970 |
-
"epoch": 2.32,
|
| 971 |
-
"learning_rate": 0.0008365896980461812,
|
| 972 |
-
"loss": 0.5336,
|
| 973 |
-
"step": 1420
|
| 974 |
-
},
|
| 975 |
-
{
|
| 976 |
-
"epoch": 2.33,
|
| 977 |
-
"learning_rate": 0.0008348134991119005,
|
| 978 |
-
"loss": 0.4807,
|
| 979 |
-
"step": 1430
|
| 980 |
-
},
|
| 981 |
-
{
|
| 982 |
-
"epoch": 2.35,
|
| 983 |
-
"learning_rate": 0.00083303730017762,
|
| 984 |
-
"loss": 0.5065,
|
| 985 |
-
"step": 1440
|
| 986 |
-
},
|
| 987 |
-
{
|
| 988 |
-
"epoch": 2.37,
|
| 989 |
-
"learning_rate": 0.0008312611012433394,
|
| 990 |
-
"loss": 0.5215,
|
| 991 |
-
"step": 1450
|
| 992 |
-
},
|
| 993 |
-
{
|
| 994 |
-
"epoch": 2.38,
|
| 995 |
-
"learning_rate": 0.0008294849023090586,
|
| 996 |
-
"loss": 0.502,
|
| 997 |
-
"step": 1460
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"epoch": 2.4,
|
| 1001 |
-
"learning_rate": 0.000827708703374778,
|
| 1002 |
-
"loss": 0.51,
|
| 1003 |
-
"step": 1470
|
| 1004 |
-
},
|
| 1005 |
-
{
|
| 1006 |
-
"epoch": 2.41,
|
| 1007 |
-
"learning_rate": 0.0008259325044404974,
|
| 1008 |
-
"loss": 0.5132,
|
| 1009 |
-
"step": 1480
|
| 1010 |
-
},
|
| 1011 |
-
{
|
| 1012 |
-
"epoch": 2.43,
|
| 1013 |
-
"learning_rate": 0.0008241563055062167,
|
| 1014 |
-
"loss": 0.4988,
|
| 1015 |
-
"step": 1490
|
| 1016 |
-
},
|
| 1017 |
-
{
|
| 1018 |
-
"epoch": 2.45,
|
| 1019 |
-
"learning_rate": 0.0008223801065719361,
|
| 1020 |
-
"loss": 0.5115,
|
| 1021 |
-
"step": 1500
|
| 1022 |
-
},
|
| 1023 |
-
{
|
| 1024 |
-
"epoch": 2.45,
|
| 1025 |
-
"eval_loss": 0.4651834964752197,
|
| 1026 |
-
"eval_runtime": 5.5836,
|
| 1027 |
-
"eval_samples_per_second": 209.72,
|
| 1028 |
-
"eval_steps_per_second": 13.253,
|
| 1029 |
-
"step": 1500
|
| 1030 |
-
},
|
| 1031 |
-
{
|
| 1032 |
-
"epoch": 2.46,
|
| 1033 |
-
"learning_rate": 0.0008206039076376555,
|
| 1034 |
-
"loss": 0.4981,
|
| 1035 |
-
"step": 1510
|
| 1036 |
-
},
|
| 1037 |
-
{
|
| 1038 |
-
"epoch": 2.48,
|
| 1039 |
-
"learning_rate": 0.0008188277087033748,
|
| 1040 |
-
"loss": 0.5488,
|
| 1041 |
-
"step": 1520
|
| 1042 |
-
},
|
| 1043 |
-
{
|
| 1044 |
-
"epoch": 2.5,
|
| 1045 |
-
"learning_rate": 0.0008170515097690942,
|
| 1046 |
-
"loss": 0.4906,
|
| 1047 |
-
"step": 1530
|
| 1048 |
-
},
|
| 1049 |
-
{
|
| 1050 |
-
"epoch": 2.51,
|
| 1051 |
-
"learning_rate": 0.0008152753108348136,
|
| 1052 |
-
"loss": 0.4722,
|
| 1053 |
-
"step": 1540
|
| 1054 |
-
},
|
| 1055 |
-
{
|
| 1056 |
-
"epoch": 2.53,
|
| 1057 |
-
"learning_rate": 0.0008134991119005329,
|
| 1058 |
-
"loss": 0.448,
|
| 1059 |
-
"step": 1550
|
| 1060 |
-
},
|
| 1061 |
-
{
|
| 1062 |
-
"epoch": 2.54,
|
| 1063 |
-
"learning_rate": 0.0008117229129662523,
|
| 1064 |
-
"loss": 0.5048,
|
| 1065 |
-
"step": 1560
|
| 1066 |
-
},
|
| 1067 |
-
{
|
| 1068 |
-
"epoch": 2.56,
|
| 1069 |
-
"learning_rate": 0.0008099467140319717,
|
| 1070 |
-
"loss": 0.4938,
|
| 1071 |
-
"step": 1570
|
| 1072 |
-
},
|
| 1073 |
-
{
|
| 1074 |
-
"epoch": 2.58,
|
| 1075 |
-
"learning_rate": 0.0008081705150976909,
|
| 1076 |
-
"loss": 0.4949,
|
| 1077 |
-
"step": 1580
|
| 1078 |
-
},
|
| 1079 |
-
{
|
| 1080 |
-
"epoch": 2.59,
|
| 1081 |
-
"learning_rate": 0.0008063943161634103,
|
| 1082 |
-
"loss": 0.5185,
|
| 1083 |
-
"step": 1590
|
| 1084 |
-
},
|
| 1085 |
-
{
|
| 1086 |
-
"epoch": 2.61,
|
| 1087 |
-
"learning_rate": 0.0008046181172291297,
|
| 1088 |
-
"loss": 0.4809,
|
| 1089 |
-
"step": 1600
|
| 1090 |
-
},
|
| 1091 |
-
{
|
| 1092 |
-
"epoch": 2.61,
|
| 1093 |
-
"eval_loss": 0.4656440317630768,
|
| 1094 |
-
"eval_runtime": 5.645,
|
| 1095 |
-
"eval_samples_per_second": 207.439,
|
| 1096 |
-
"eval_steps_per_second": 13.109,
|
| 1097 |
-
"step": 1600
|
| 1098 |
-
},
|
| 1099 |
-
{
|
| 1100 |
-
"epoch": 2.63,
|
| 1101 |
-
"learning_rate": 0.0008028419182948491,
|
| 1102 |
-
"loss": 0.4819,
|
| 1103 |
-
"step": 1610
|
| 1104 |
-
},
|
| 1105 |
-
{
|
| 1106 |
-
"epoch": 2.64,
|
| 1107 |
-
"learning_rate": 0.0008010657193605684,
|
| 1108 |
-
"loss": 0.513,
|
| 1109 |
-
"step": 1620
|
| 1110 |
-
},
|
| 1111 |
-
{
|
| 1112 |
-
"epoch": 2.66,
|
| 1113 |
-
"learning_rate": 0.0007992895204262878,
|
| 1114 |
-
"loss": 0.5312,
|
| 1115 |
-
"step": 1630
|
| 1116 |
-
},
|
| 1117 |
-
{
|
| 1118 |
-
"epoch": 2.68,
|
| 1119 |
-
"learning_rate": 0.0007975133214920072,
|
| 1120 |
-
"loss": 0.5252,
|
| 1121 |
-
"step": 1640
|
| 1122 |
-
},
|
| 1123 |
-
{
|
| 1124 |
-
"epoch": 2.69,
|
| 1125 |
-
"learning_rate": 0.0007957371225577265,
|
| 1126 |
-
"loss": 0.4977,
|
| 1127 |
-
"step": 1650
|
| 1128 |
-
},
|
| 1129 |
-
{
|
| 1130 |
-
"epoch": 2.71,
|
| 1131 |
-
"learning_rate": 0.0007939609236234459,
|
| 1132 |
-
"loss": 0.4707,
|
| 1133 |
-
"step": 1660
|
| 1134 |
-
},
|
| 1135 |
-
{
|
| 1136 |
-
"epoch": 2.72,
|
| 1137 |
-
"learning_rate": 0.0007921847246891653,
|
| 1138 |
-
"loss": 0.5138,
|
| 1139 |
-
"step": 1670
|
| 1140 |
-
},
|
| 1141 |
-
{
|
| 1142 |
-
"epoch": 2.74,
|
| 1143 |
-
"learning_rate": 0.0007904085257548845,
|
| 1144 |
-
"loss": 0.5096,
|
| 1145 |
-
"step": 1680
|
| 1146 |
-
},
|
| 1147 |
-
{
|
| 1148 |
-
"epoch": 2.76,
|
| 1149 |
-
"learning_rate": 0.0007886323268206039,
|
| 1150 |
-
"loss": 0.5299,
|
| 1151 |
-
"step": 1690
|
| 1152 |
-
},
|
| 1153 |
-
{
|
| 1154 |
-
"epoch": 2.77,
|
| 1155 |
-
"learning_rate": 0.0007868561278863233,
|
| 1156 |
-
"loss": 0.5147,
|
| 1157 |
-
"step": 1700
|
| 1158 |
-
},
|
| 1159 |
-
{
|
| 1160 |
-
"epoch": 2.77,
|
| 1161 |
-
"eval_loss": 0.47206491231918335,
|
| 1162 |
-
"eval_runtime": 5.616,
|
| 1163 |
-
"eval_samples_per_second": 208.511,
|
| 1164 |
-
"eval_steps_per_second": 13.177,
|
| 1165 |
-
"step": 1700
|
| 1166 |
-
},
|
| 1167 |
-
{
|
| 1168 |
-
"epoch": 2.79,
|
| 1169 |
-
"learning_rate": 0.0007850799289520426,
|
| 1170 |
-
"loss": 0.5177,
|
| 1171 |
-
"step": 1710
|
| 1172 |
-
},
|
| 1173 |
-
{
|
| 1174 |
-
"epoch": 2.81,
|
| 1175 |
-
"learning_rate": 0.000783303730017762,
|
| 1176 |
-
"loss": 0.4924,
|
| 1177 |
-
"step": 1720
|
| 1178 |
-
},
|
| 1179 |
-
{
|
| 1180 |
-
"epoch": 2.82,
|
| 1181 |
-
"learning_rate": 0.0007815275310834814,
|
| 1182 |
-
"loss": 0.5463,
|
| 1183 |
-
"step": 1730
|
| 1184 |
-
},
|
| 1185 |
-
{
|
| 1186 |
-
"epoch": 2.84,
|
| 1187 |
-
"learning_rate": 0.0007797513321492007,
|
| 1188 |
-
"loss": 0.5046,
|
| 1189 |
-
"step": 1740
|
| 1190 |
-
},
|
| 1191 |
-
{
|
| 1192 |
-
"epoch": 2.85,
|
| 1193 |
-
"learning_rate": 0.0007779751332149201,
|
| 1194 |
-
"loss": 0.4933,
|
| 1195 |
-
"step": 1750
|
| 1196 |
-
},
|
| 1197 |
-
{
|
| 1198 |
-
"epoch": 2.87,
|
| 1199 |
-
"learning_rate": 0.0007761989342806395,
|
| 1200 |
-
"loss": 0.5742,
|
| 1201 |
-
"step": 1760
|
| 1202 |
-
},
|
| 1203 |
-
{
|
| 1204 |
-
"epoch": 2.89,
|
| 1205 |
-
"learning_rate": 0.0007744227353463588,
|
| 1206 |
-
"loss": 0.4831,
|
| 1207 |
-
"step": 1770
|
| 1208 |
-
},
|
| 1209 |
-
{
|
| 1210 |
-
"epoch": 2.9,
|
| 1211 |
-
"learning_rate": 0.0007726465364120782,
|
| 1212 |
-
"loss": 0.5179,
|
| 1213 |
-
"step": 1780
|
| 1214 |
-
},
|
| 1215 |
-
{
|
| 1216 |
-
"epoch": 2.92,
|
| 1217 |
-
"learning_rate": 0.0007708703374777975,
|
| 1218 |
-
"loss": 0.52,
|
| 1219 |
-
"step": 1790
|
| 1220 |
-
},
|
| 1221 |
-
{
|
| 1222 |
-
"epoch": 2.94,
|
| 1223 |
-
"learning_rate": 0.0007690941385435168,
|
| 1224 |
-
"loss": 0.4878,
|
| 1225 |
-
"step": 1800
|
| 1226 |
-
},
|
| 1227 |
-
{
|
| 1228 |
-
"epoch": 2.94,
|
| 1229 |
-
"eval_loss": 0.45261794328689575,
|
| 1230 |
-
"eval_runtime": 5.761,
|
| 1231 |
-
"eval_samples_per_second": 203.263,
|
| 1232 |
-
"eval_steps_per_second": 12.845,
|
| 1233 |
-
"step": 1800
|
| 1234 |
-
},
|
| 1235 |
-
{
|
| 1236 |
-
"epoch": 2.95,
|
| 1237 |
-
"learning_rate": 0.0007673179396092362,
|
| 1238 |
-
"loss": 0.465,
|
| 1239 |
-
"step": 1810
|
| 1240 |
-
},
|
| 1241 |
-
{
|
| 1242 |
-
"epoch": 2.97,
|
| 1243 |
-
"learning_rate": 0.0007655417406749556,
|
| 1244 |
-
"loss": 0.5103,
|
| 1245 |
-
"step": 1820
|
| 1246 |
-
},
|
| 1247 |
-
{
|
| 1248 |
-
"epoch": 2.99,
|
| 1249 |
-
"learning_rate": 0.0007637655417406749,
|
| 1250 |
-
"loss": 0.4947,
|
| 1251 |
-
"step": 1830
|
| 1252 |
-
},
|
| 1253 |
-
{
|
| 1254 |
-
"epoch": 3.0,
|
| 1255 |
-
"learning_rate": 0.0007619893428063943,
|
| 1256 |
-
"loss": 0.4861,
|
| 1257 |
-
"step": 1840
|
| 1258 |
-
},
|
| 1259 |
-
{
|
| 1260 |
-
"epoch": 3.02,
|
| 1261 |
-
"learning_rate": 0.0007602131438721137,
|
| 1262 |
-
"loss": 0.4779,
|
| 1263 |
-
"step": 1850
|
| 1264 |
-
},
|
| 1265 |
-
{
|
| 1266 |
-
"epoch": 3.03,
|
| 1267 |
-
"learning_rate": 0.0007584369449378331,
|
| 1268 |
-
"loss": 0.483,
|
| 1269 |
-
"step": 1860
|
| 1270 |
-
},
|
| 1271 |
-
{
|
| 1272 |
-
"epoch": 3.05,
|
| 1273 |
-
"learning_rate": 0.0007566607460035524,
|
| 1274 |
-
"loss": 0.5363,
|
| 1275 |
-
"step": 1870
|
| 1276 |
-
},
|
| 1277 |
-
{
|
| 1278 |
-
"epoch": 3.07,
|
| 1279 |
-
"learning_rate": 0.0007548845470692718,
|
| 1280 |
-
"loss": 0.5,
|
| 1281 |
-
"step": 1880
|
| 1282 |
-
},
|
| 1283 |
-
{
|
| 1284 |
-
"epoch": 3.08,
|
| 1285 |
-
"learning_rate": 0.0007531083481349912,
|
| 1286 |
-
"loss": 0.492,
|
| 1287 |
-
"step": 1890
|
| 1288 |
-
},
|
| 1289 |
-
{
|
| 1290 |
-
"epoch": 3.1,
|
| 1291 |
-
"learning_rate": 0.0007513321492007104,
|
| 1292 |
-
"loss": 0.4791,
|
| 1293 |
-
"step": 1900
|
| 1294 |
-
},
|
| 1295 |
-
{
|
| 1296 |
-
"epoch": 3.1,
|
| 1297 |
-
"eval_loss": 0.45457181334495544,
|
| 1298 |
-
"eval_runtime": 5.836,
|
| 1299 |
-
"eval_samples_per_second": 200.65,
|
| 1300 |
-
"eval_steps_per_second": 12.68,
|
| 1301 |
-
"step": 1900
|
| 1302 |
-
},
|
| 1303 |
-
{
|
| 1304 |
-
"epoch": 3.12,
|
| 1305 |
-
"learning_rate": 0.0007495559502664298,
|
| 1306 |
-
"loss": 0.5009,
|
| 1307 |
-
"step": 1910
|
| 1308 |
-
},
|
| 1309 |
-
{
|
| 1310 |
-
"epoch": 3.13,
|
| 1311 |
-
"learning_rate": 0.0007477797513321492,
|
| 1312 |
-
"loss": 0.4826,
|
| 1313 |
-
"step": 1920
|
| 1314 |
-
},
|
| 1315 |
-
{
|
| 1316 |
-
"epoch": 3.15,
|
| 1317 |
-
"learning_rate": 0.0007460035523978685,
|
| 1318 |
-
"loss": 0.4623,
|
| 1319 |
-
"step": 1930
|
| 1320 |
-
},
|
| 1321 |
-
{
|
| 1322 |
-
"epoch": 3.16,
|
| 1323 |
-
"learning_rate": 0.0007442273534635879,
|
| 1324 |
-
"loss": 0.4975,
|
| 1325 |
-
"step": 1940
|
| 1326 |
-
},
|
| 1327 |
-
{
|
| 1328 |
-
"epoch": 3.18,
|
| 1329 |
-
"learning_rate": 0.0007424511545293073,
|
| 1330 |
-
"loss": 0.4911,
|
| 1331 |
-
"step": 1950
|
| 1332 |
-
},
|
| 1333 |
-
{
|
| 1334 |
-
"epoch": 3.2,
|
| 1335 |
-
"learning_rate": 0.0007406749555950266,
|
| 1336 |
-
"loss": 0.4896,
|
| 1337 |
-
"step": 1960
|
| 1338 |
-
},
|
| 1339 |
-
{
|
| 1340 |
-
"epoch": 3.21,
|
| 1341 |
-
"learning_rate": 0.000738898756660746,
|
| 1342 |
-
"loss": 0.5118,
|
| 1343 |
-
"step": 1970
|
| 1344 |
-
},
|
| 1345 |
-
{
|
| 1346 |
-
"epoch": 3.23,
|
| 1347 |
-
"learning_rate": 0.0007371225577264654,
|
| 1348 |
-
"loss": 0.4827,
|
| 1349 |
-
"step": 1980
|
| 1350 |
-
},
|
| 1351 |
-
{
|
| 1352 |
-
"epoch": 3.25,
|
| 1353 |
-
"learning_rate": 0.0007353463587921847,
|
| 1354 |
-
"loss": 0.4929,
|
| 1355 |
-
"step": 1990
|
| 1356 |
-
},
|
| 1357 |
-
{
|
| 1358 |
-
"epoch": 3.26,
|
| 1359 |
-
"learning_rate": 0.000733570159857904,
|
| 1360 |
-
"loss": 0.462,
|
| 1361 |
-
"step": 2000
|
| 1362 |
-
},
|
| 1363 |
-
{
|
| 1364 |
-
"epoch": 3.26,
|
| 1365 |
-
"eval_loss": 0.4662368595600128,
|
| 1366 |
-
"eval_runtime": 5.657,
|
| 1367 |
-
"eval_samples_per_second": 206.999,
|
| 1368 |
-
"eval_steps_per_second": 13.081,
|
| 1369 |
-
"step": 2000
|
| 1370 |
-
},
|
| 1371 |
-
{
|
| 1372 |
-
"epoch": 3.28,
|
| 1373 |
-
"learning_rate": 0.0007317939609236234,
|
| 1374 |
-
"loss": 0.505,
|
| 1375 |
-
"step": 2010
|
| 1376 |
-
},
|
| 1377 |
-
{
|
| 1378 |
-
"epoch": 3.3,
|
| 1379 |
-
"learning_rate": 0.0007300177619893427,
|
| 1380 |
-
"loss": 0.4871,
|
| 1381 |
-
"step": 2020
|
| 1382 |
-
},
|
| 1383 |
-
{
|
| 1384 |
-
"epoch": 3.31,
|
| 1385 |
-
"learning_rate": 0.0007282415630550621,
|
| 1386 |
-
"loss": 0.4841,
|
| 1387 |
-
"step": 2030
|
| 1388 |
-
},
|
| 1389 |
-
{
|
| 1390 |
-
"epoch": 3.33,
|
| 1391 |
-
"learning_rate": 0.0007264653641207815,
|
| 1392 |
-
"loss": 0.4918,
|
| 1393 |
-
"step": 2040
|
| 1394 |
-
},
|
| 1395 |
-
{
|
| 1396 |
-
"epoch": 3.34,
|
| 1397 |
-
"learning_rate": 0.0007246891651865008,
|
| 1398 |
-
"loss": 0.4559,
|
| 1399 |
-
"step": 2050
|
| 1400 |
-
},
|
| 1401 |
-
{
|
| 1402 |
-
"epoch": 3.36,
|
| 1403 |
-
"learning_rate": 0.0007229129662522202,
|
| 1404 |
-
"loss": 0.4672,
|
| 1405 |
-
"step": 2060
|
| 1406 |
-
},
|
| 1407 |
-
{
|
| 1408 |
-
"epoch": 3.38,
|
| 1409 |
-
"learning_rate": 0.0007211367673179397,
|
| 1410 |
-
"loss": 0.4624,
|
| 1411 |
-
"step": 2070
|
| 1412 |
-
},
|
| 1413 |
-
{
|
| 1414 |
-
"epoch": 3.39,
|
| 1415 |
-
"learning_rate": 0.000719360568383659,
|
| 1416 |
-
"loss": 0.4405,
|
| 1417 |
-
"step": 2080
|
| 1418 |
-
},
|
| 1419 |
-
{
|
| 1420 |
-
"epoch": 3.41,
|
| 1421 |
-
"learning_rate": 0.0007175843694493784,
|
| 1422 |
-
"loss": 0.4842,
|
| 1423 |
-
"step": 2090
|
| 1424 |
-
},
|
| 1425 |
-
{
|
| 1426 |
-
"epoch": 3.43,
|
| 1427 |
-
"learning_rate": 0.0007158081705150978,
|
| 1428 |
-
"loss": 0.4959,
|
| 1429 |
-
"step": 2100
|
| 1430 |
-
},
|
| 1431 |
-
{
|
| 1432 |
-
"epoch": 3.43,
|
| 1433 |
-
"eval_loss": 0.4539988338947296,
|
| 1434 |
-
"eval_runtime": 5.63,
|
| 1435 |
-
"eval_samples_per_second": 207.992,
|
| 1436 |
-
"eval_steps_per_second": 13.144,
|
| 1437 |
-
"step": 2100
|
| 1438 |
-
},
|
| 1439 |
-
{
|
| 1440 |
-
"epoch": 3.44,
|
| 1441 |
-
"learning_rate": 0.0007140319715808172,
|
| 1442 |
-
"loss": 0.4884,
|
| 1443 |
-
"step": 2110
|
| 1444 |
-
},
|
| 1445 |
-
{
|
| 1446 |
-
"epoch": 3.46,
|
| 1447 |
-
"learning_rate": 0.0007122557726465364,
|
| 1448 |
-
"loss": 0.4808,
|
| 1449 |
-
"step": 2120
|
| 1450 |
-
},
|
| 1451 |
-
{
|
| 1452 |
-
"epoch": 3.47,
|
| 1453 |
-
"learning_rate": 0.0007104795737122558,
|
| 1454 |
-
"loss": 0.4494,
|
| 1455 |
-
"step": 2130
|
| 1456 |
-
},
|
| 1457 |
-
{
|
| 1458 |
-
"epoch": 3.49,
|
| 1459 |
-
"learning_rate": 0.0007087033747779752,
|
| 1460 |
-
"loss": 0.4789,
|
| 1461 |
-
"step": 2140
|
| 1462 |
-
},
|
| 1463 |
-
{
|
| 1464 |
-
"epoch": 3.51,
|
| 1465 |
-
"learning_rate": 0.0007069271758436945,
|
| 1466 |
-
"loss": 0.472,
|
| 1467 |
-
"step": 2150
|
| 1468 |
-
},
|
| 1469 |
-
{
|
| 1470 |
-
"epoch": 3.52,
|
| 1471 |
-
"learning_rate": 0.0007051509769094139,
|
| 1472 |
-
"loss": 0.4731,
|
| 1473 |
-
"step": 2160
|
| 1474 |
-
},
|
| 1475 |
-
{
|
| 1476 |
-
"epoch": 3.54,
|
| 1477 |
-
"learning_rate": 0.0007033747779751333,
|
| 1478 |
-
"loss": 0.4861,
|
| 1479 |
-
"step": 2170
|
| 1480 |
-
},
|
| 1481 |
-
{
|
| 1482 |
-
"epoch": 3.56,
|
| 1483 |
-
"learning_rate": 0.0007015985790408526,
|
| 1484 |
-
"loss": 0.4876,
|
| 1485 |
-
"step": 2180
|
| 1486 |
-
},
|
| 1487 |
-
{
|
| 1488 |
-
"epoch": 3.57,
|
| 1489 |
-
"learning_rate": 0.000699822380106572,
|
| 1490 |
-
"loss": 0.5136,
|
| 1491 |
-
"step": 2190
|
| 1492 |
-
},
|
| 1493 |
-
{
|
| 1494 |
-
"epoch": 3.59,
|
| 1495 |
-
"learning_rate": 0.0006980461811722914,
|
| 1496 |
-
"loss": 0.4671,
|
| 1497 |
-
"step": 2200
|
| 1498 |
-
},
|
| 1499 |
-
{
|
| 1500 |
-
"epoch": 3.59,
|
| 1501 |
-
"eval_loss": 0.4472385346889496,
|
| 1502 |
-
"eval_runtime": 5.617,
|
| 1503 |
-
"eval_samples_per_second": 208.473,
|
| 1504 |
-
"eval_steps_per_second": 13.174,
|
| 1505 |
-
"step": 2200
|
| 1506 |
-
},
|
| 1507 |
-
{
|
| 1508 |
-
"epoch": 3.61,
|
| 1509 |
-
"learning_rate": 0.0006962699822380107,
|
| 1510 |
-
"loss": 0.4671,
|
| 1511 |
-
"step": 2210
|
| 1512 |
-
},
|
| 1513 |
-
{
|
| 1514 |
-
"epoch": 3.62,
|
| 1515 |
-
"learning_rate": 0.0006944937833037301,
|
| 1516 |
-
"loss": 0.4896,
|
| 1517 |
-
"step": 2220
|
| 1518 |
-
},
|
| 1519 |
-
{
|
| 1520 |
-
"epoch": 3.64,
|
| 1521 |
-
"learning_rate": 0.0006927175843694495,
|
| 1522 |
-
"loss": 0.4768,
|
| 1523 |
-
"step": 2230
|
| 1524 |
-
},
|
| 1525 |
-
{
|
| 1526 |
-
"epoch": 3.65,
|
| 1527 |
-
"learning_rate": 0.0006909413854351687,
|
| 1528 |
-
"loss": 0.5263,
|
| 1529 |
-
"step": 2240
|
| 1530 |
-
},
|
| 1531 |
-
{
|
| 1532 |
-
"epoch": 3.67,
|
| 1533 |
-
"learning_rate": 0.0006891651865008881,
|
| 1534 |
-
"loss": 0.4718,
|
| 1535 |
-
"step": 2250
|
| 1536 |
-
},
|
| 1537 |
-
{
|
| 1538 |
-
"epoch": 3.69,
|
| 1539 |
-
"learning_rate": 0.0006873889875666075,
|
| 1540 |
-
"loss": 0.4854,
|
| 1541 |
-
"step": 2260
|
| 1542 |
-
},
|
| 1543 |
-
{
|
| 1544 |
-
"epoch": 3.7,
|
| 1545 |
-
"learning_rate": 0.0006856127886323268,
|
| 1546 |
-
"loss": 0.4614,
|
| 1547 |
-
"step": 2270
|
| 1548 |
-
},
|
| 1549 |
-
{
|
| 1550 |
-
"epoch": 3.72,
|
| 1551 |
-
"learning_rate": 0.0006838365896980462,
|
| 1552 |
-
"loss": 0.467,
|
| 1553 |
-
"step": 2280
|
| 1554 |
-
},
|
| 1555 |
-
{
|
| 1556 |
-
"epoch": 3.74,
|
| 1557 |
-
"learning_rate": 0.0006820603907637656,
|
| 1558 |
-
"loss": 0.4824,
|
| 1559 |
-
"step": 2290
|
| 1560 |
-
},
|
| 1561 |
-
{
|
| 1562 |
-
"epoch": 3.75,
|
| 1563 |
-
"learning_rate": 0.0006802841918294849,
|
| 1564 |
-
"loss": 0.479,
|
| 1565 |
-
"step": 2300
|
| 1566 |
-
},
|
| 1567 |
-
{
|
| 1568 |
-
"epoch": 3.75,
|
| 1569 |
-
"eval_loss": 0.45660024881362915,
|
| 1570 |
-
"eval_runtime": 5.623,
|
| 1571 |
-
"eval_samples_per_second": 208.251,
|
| 1572 |
-
"eval_steps_per_second": 13.16,
|
| 1573 |
-
"step": 2300
|
| 1574 |
-
},
|
| 1575 |
-
{
|
| 1576 |
-
"epoch": 3.77,
|
| 1577 |
-
"learning_rate": 0.0006785079928952043,
|
| 1578 |
-
"loss": 0.4761,
|
| 1579 |
-
"step": 2310
|
| 1580 |
-
},
|
| 1581 |
-
{
|
| 1582 |
-
"epoch": 3.78,
|
| 1583 |
-
"learning_rate": 0.0006767317939609237,
|
| 1584 |
-
"loss": 0.4789,
|
| 1585 |
-
"step": 2320
|
| 1586 |
-
},
|
| 1587 |
-
{
|
| 1588 |
-
"epoch": 3.8,
|
| 1589 |
-
"learning_rate": 0.000674955595026643,
|
| 1590 |
-
"loss": 0.4888,
|
| 1591 |
-
"step": 2330
|
| 1592 |
-
},
|
| 1593 |
-
{
|
| 1594 |
-
"epoch": 3.82,
|
| 1595 |
-
"learning_rate": 0.0006731793960923623,
|
| 1596 |
-
"loss": 0.4774,
|
| 1597 |
-
"step": 2340
|
| 1598 |
-
},
|
| 1599 |
-
{
|
| 1600 |
-
"epoch": 3.83,
|
| 1601 |
-
"learning_rate": 0.0006714031971580817,
|
| 1602 |
-
"loss": 0.4621,
|
| 1603 |
-
"step": 2350
|
| 1604 |
-
},
|
| 1605 |
-
{
|
| 1606 |
-
"epoch": 3.85,
|
| 1607 |
-
"learning_rate": 0.0006696269982238011,
|
| 1608 |
-
"loss": 0.4804,
|
| 1609 |
-
"step": 2360
|
| 1610 |
-
},
|
| 1611 |
-
{
|
| 1612 |
-
"epoch": 3.87,
|
| 1613 |
-
"learning_rate": 0.0006678507992895204,
|
| 1614 |
-
"loss": 0.466,
|
| 1615 |
-
"step": 2370
|
| 1616 |
-
},
|
| 1617 |
-
{
|
| 1618 |
-
"epoch": 3.88,
|
| 1619 |
-
"learning_rate": 0.0006660746003552398,
|
| 1620 |
-
"loss": 0.4846,
|
| 1621 |
-
"step": 2380
|
| 1622 |
-
},
|
| 1623 |
-
{
|
| 1624 |
-
"epoch": 3.9,
|
| 1625 |
-
"learning_rate": 0.0006642984014209592,
|
| 1626 |
-
"loss": 0.4631,
|
| 1627 |
-
"step": 2390
|
| 1628 |
-
},
|
| 1629 |
-
{
|
| 1630 |
-
"epoch": 3.92,
|
| 1631 |
-
"learning_rate": 0.0006625222024866785,
|
| 1632 |
-
"loss": 0.4568,
|
| 1633 |
-
"step": 2400
|
| 1634 |
-
},
|
| 1635 |
-
{
|
| 1636 |
-
"epoch": 3.92,
|
| 1637 |
-
"eval_loss": 0.4491855502128601,
|
| 1638 |
-
"eval_runtime": 5.621,
|
| 1639 |
-
"eval_samples_per_second": 208.325,
|
| 1640 |
-
"eval_steps_per_second": 13.165,
|
| 1641 |
-
"step": 2400
|
| 1642 |
-
},
|
| 1643 |
-
{
|
| 1644 |
-
"epoch": 3.93,
|
| 1645 |
-
"learning_rate": 0.0006607460035523979,
|
| 1646 |
-
"loss": 0.5022,
|
| 1647 |
-
"step": 2410
|
| 1648 |
-
},
|
| 1649 |
-
{
|
| 1650 |
-
"epoch": 3.95,
|
| 1651 |
-
"learning_rate": 0.0006589698046181173,
|
| 1652 |
-
"loss": 0.4546,
|
| 1653 |
-
"step": 2420
|
| 1654 |
-
},
|
| 1655 |
-
{
|
| 1656 |
-
"epoch": 3.96,
|
| 1657 |
-
"learning_rate": 0.0006571936056838366,
|
| 1658 |
-
"loss": 0.4666,
|
| 1659 |
-
"step": 2430
|
| 1660 |
-
},
|
| 1661 |
-
{
|
| 1662 |
-
"epoch": 3.98,
|
| 1663 |
-
"learning_rate": 0.000655417406749556,
|
| 1664 |
-
"loss": 0.4915,
|
| 1665 |
-
"step": 2440
|
| 1666 |
-
},
|
| 1667 |
-
{
|
| 1668 |
-
"epoch": 4.0,
|
| 1669 |
-
"learning_rate": 0.0006536412078152753,
|
| 1670 |
-
"loss": 0.4763,
|
| 1671 |
-
"step": 2450
|
| 1672 |
-
},
|
| 1673 |
-
{
|
| 1674 |
-
"epoch": 4.01,
|
| 1675 |
-
"learning_rate": 0.0006518650088809946,
|
| 1676 |
-
"loss": 0.4625,
|
| 1677 |
-
"step": 2460
|
| 1678 |
-
},
|
| 1679 |
-
{
|
| 1680 |
-
"epoch": 4.03,
|
| 1681 |
-
"learning_rate": 0.000650088809946714,
|
| 1682 |
-
"loss": 0.4447,
|
| 1683 |
-
"step": 2470
|
| 1684 |
-
},
|
| 1685 |
-
{
|
| 1686 |
-
"epoch": 4.05,
|
| 1687 |
-
"learning_rate": 0.0006483126110124334,
|
| 1688 |
-
"loss": 0.4559,
|
| 1689 |
-
"step": 2480
|
| 1690 |
-
},
|
| 1691 |
-
{
|
| 1692 |
-
"epoch": 4.06,
|
| 1693 |
-
"learning_rate": 0.0006465364120781527,
|
| 1694 |
-
"loss": 0.4881,
|
| 1695 |
-
"step": 2490
|
| 1696 |
-
},
|
| 1697 |
-
{
|
| 1698 |
-
"epoch": 4.08,
|
| 1699 |
-
"learning_rate": 0.0006447602131438721,
|
| 1700 |
-
"loss": 0.472,
|
| 1701 |
-
"step": 2500
|
| 1702 |
-
},
|
| 1703 |
-
{
|
| 1704 |
-
"epoch": 4.08,
|
| 1705 |
-
"eval_loss": 0.45457056164741516,
|
| 1706 |
-
"eval_runtime": 5.638,
|
| 1707 |
-
"eval_samples_per_second": 207.697,
|
| 1708 |
-
"eval_steps_per_second": 13.125,
|
| 1709 |
-
"step": 2500
|
| 1710 |
-
},
|
| 1711 |
-
{
|
| 1712 |
-
"epoch": 4.09,
|
| 1713 |
-
"learning_rate": 0.0006429840142095915,
|
| 1714 |
-
"loss": 0.4606,
|
| 1715 |
-
"step": 2510
|
| 1716 |
-
},
|
| 1717 |
-
{
|
| 1718 |
-
"epoch": 4.11,
|
| 1719 |
-
"learning_rate": 0.0006412078152753108,
|
| 1720 |
-
"loss": 0.4599,
|
| 1721 |
-
"step": 2520
|
| 1722 |
-
},
|
| 1723 |
-
{
|
| 1724 |
-
"epoch": 4.13,
|
| 1725 |
-
"learning_rate": 0.0006394316163410302,
|
| 1726 |
-
"loss": 0.4846,
|
| 1727 |
-
"step": 2530
|
| 1728 |
-
},
|
| 1729 |
-
{
|
| 1730 |
-
"epoch": 4.14,
|
| 1731 |
-
"learning_rate": 0.0006376554174067496,
|
| 1732 |
-
"loss": 0.4575,
|
| 1733 |
-
"step": 2540
|
| 1734 |
-
},
|
| 1735 |
-
{
|
| 1736 |
-
"epoch": 4.16,
|
| 1737 |
-
"learning_rate": 0.0006358792184724689,
|
| 1738 |
-
"loss": 0.4758,
|
| 1739 |
-
"step": 2550
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 4.18,
|
| 1743 |
-
"learning_rate": 0.0006341030195381882,
|
| 1744 |
-
"loss": 0.4963,
|
| 1745 |
-
"step": 2560
|
| 1746 |
-
},
|
| 1747 |
-
{
|
| 1748 |
-
"epoch": 4.19,
|
| 1749 |
-
"learning_rate": 0.0006323268206039076,
|
| 1750 |
-
"loss": 0.4315,
|
| 1751 |
-
"step": 2570
|
| 1752 |
-
},
|
| 1753 |
-
{
|
| 1754 |
-
"epoch": 4.21,
|
| 1755 |
-
"learning_rate": 0.0006305506216696269,
|
| 1756 |
-
"loss": 0.4759,
|
| 1757 |
-
"step": 2580
|
| 1758 |
-
},
|
| 1759 |
-
{
|
| 1760 |
-
"epoch": 4.23,
|
| 1761 |
-
"learning_rate": 0.0006287744227353463,
|
| 1762 |
-
"loss": 0.4627,
|
| 1763 |
-
"step": 2590
|
| 1764 |
-
},
|
| 1765 |
-
{
|
| 1766 |
-
"epoch": 4.24,
|
| 1767 |
-
"learning_rate": 0.0006269982238010657,
|
| 1768 |
-
"loss": 0.4411,
|
| 1769 |
-
"step": 2600
|
| 1770 |
-
},
|
| 1771 |
-
{
|
| 1772 |
-
"epoch": 4.24,
|
| 1773 |
-
"eval_loss": 0.4529522657394409,
|
| 1774 |
-
"eval_runtime": 5.693,
|
| 1775 |
-
"eval_samples_per_second": 205.69,
|
| 1776 |
-
"eval_steps_per_second": 12.998,
|
| 1777 |
-
"step": 2600
|
| 1778 |
-
},
|
| 1779 |
-
{
|
| 1780 |
-
"epoch": 4.26,
|
| 1781 |
-
"learning_rate": 0.0006252220248667851,
|
| 1782 |
-
"loss": 0.4747,
|
| 1783 |
-
"step": 2610
|
| 1784 |
-
},
|
| 1785 |
-
{
|
| 1786 |
-
"epoch": 4.27,
|
| 1787 |
-
"learning_rate": 0.0006234458259325044,
|
| 1788 |
-
"loss": 0.4977,
|
| 1789 |
-
"step": 2620
|
| 1790 |
-
},
|
| 1791 |
-
{
|
| 1792 |
-
"epoch": 4.29,
|
| 1793 |
-
"learning_rate": 0.0006216696269982238,
|
| 1794 |
-
"loss": 0.4606,
|
| 1795 |
-
"step": 2630
|
| 1796 |
-
},
|
| 1797 |
-
{
|
| 1798 |
-
"epoch": 4.31,
|
| 1799 |
-
"learning_rate": 0.0006198934280639432,
|
| 1800 |
-
"loss": 0.4678,
|
| 1801 |
-
"step": 2640
|
| 1802 |
-
},
|
| 1803 |
-
{
|
| 1804 |
-
"epoch": 4.32,
|
| 1805 |
-
"learning_rate": 0.0006181172291296625,
|
| 1806 |
-
"loss": 0.4665,
|
| 1807 |
-
"step": 2650
|
| 1808 |
-
},
|
| 1809 |
-
{
|
| 1810 |
-
"epoch": 4.34,
|
| 1811 |
-
"learning_rate": 0.0006163410301953819,
|
| 1812 |
-
"loss": 0.4571,
|
| 1813 |
-
"step": 2660
|
| 1814 |
-
},
|
| 1815 |
-
{
|
| 1816 |
-
"epoch": 4.36,
|
| 1817 |
-
"learning_rate": 0.0006145648312611012,
|
| 1818 |
-
"loss": 0.4412,
|
| 1819 |
-
"step": 2670
|
| 1820 |
-
},
|
| 1821 |
-
{
|
| 1822 |
-
"epoch": 4.37,
|
| 1823 |
-
"learning_rate": 0.0006127886323268205,
|
| 1824 |
-
"loss": 0.4521,
|
| 1825 |
-
"step": 2680
|
| 1826 |
-
},
|
| 1827 |
-
{
|
| 1828 |
-
"epoch": 4.39,
|
| 1829 |
-
"learning_rate": 0.00061101243339254,
|
| 1830 |
-
"loss": 0.4387,
|
| 1831 |
-
"step": 2690
|
| 1832 |
-
},
|
| 1833 |
-
{
|
| 1834 |
-
"epoch": 4.4,
|
| 1835 |
-
"learning_rate": 0.0006092362344582594,
|
| 1836 |
-
"loss": 0.4636,
|
| 1837 |
-
"step": 2700
|
| 1838 |
-
},
|
| 1839 |
-
{
|
| 1840 |
-
"epoch": 4.4,
|
| 1841 |
-
"eval_loss": 0.44287770986557007,
|
| 1842 |
-
"eval_runtime": 5.817,
|
| 1843 |
-
"eval_samples_per_second": 201.306,
|
| 1844 |
-
"eval_steps_per_second": 12.721,
|
| 1845 |
-
"step": 2700
|
| 1846 |
-
},
|
| 1847 |
-
{
|
| 1848 |
-
"epoch": 4.42,
|
| 1849 |
-
"learning_rate": 0.0006074600355239787,
|
| 1850 |
-
"loss": 0.4726,
|
| 1851 |
-
"step": 2710
|
| 1852 |
-
},
|
| 1853 |
-
{
|
| 1854 |
-
"epoch": 4.44,
|
| 1855 |
-
"learning_rate": 0.0006056838365896981,
|
| 1856 |
-
"loss": 0.4297,
|
| 1857 |
-
"step": 2720
|
| 1858 |
-
},
|
| 1859 |
-
{
|
| 1860 |
-
"epoch": 4.45,
|
| 1861 |
-
"learning_rate": 0.0006039076376554175,
|
| 1862 |
-
"loss": 0.4889,
|
| 1863 |
-
"step": 2730
|
| 1864 |
-
},
|
| 1865 |
-
{
|
| 1866 |
-
"epoch": 4.47,
|
| 1867 |
-
"learning_rate": 0.0006021314387211368,
|
| 1868 |
-
"loss": 0.439,
|
| 1869 |
-
"step": 2740
|
| 1870 |
-
},
|
| 1871 |
-
{
|
| 1872 |
-
"epoch": 4.49,
|
| 1873 |
-
"learning_rate": 0.0006003552397868562,
|
| 1874 |
-
"loss": 0.4527,
|
| 1875 |
-
"step": 2750
|
| 1876 |
-
},
|
| 1877 |
-
{
|
| 1878 |
-
"epoch": 4.5,
|
| 1879 |
-
"learning_rate": 0.0005985790408525756,
|
| 1880 |
-
"loss": 0.5186,
|
| 1881 |
-
"step": 2760
|
| 1882 |
-
},
|
| 1883 |
-
{
|
| 1884 |
-
"epoch": 4.52,
|
| 1885 |
-
"learning_rate": 0.0005968028419182949,
|
| 1886 |
-
"loss": 0.4882,
|
| 1887 |
-
"step": 2770
|
| 1888 |
-
},
|
| 1889 |
-
{
|
| 1890 |
-
"epoch": 4.54,
|
| 1891 |
-
"learning_rate": 0.0005950266429840142,
|
| 1892 |
-
"loss": 0.4759,
|
| 1893 |
-
"step": 2780
|
| 1894 |
-
},
|
| 1895 |
-
{
|
| 1896 |
-
"epoch": 4.55,
|
| 1897 |
-
"learning_rate": 0.0005932504440497336,
|
| 1898 |
-
"loss": 0.4744,
|
| 1899 |
-
"step": 2790
|
| 1900 |
-
},
|
| 1901 |
-
{
|
| 1902 |
-
"epoch": 4.57,
|
| 1903 |
-
"learning_rate": 0.0005914742451154529,
|
| 1904 |
-
"loss": 0.4512,
|
| 1905 |
-
"step": 2800
|
| 1906 |
-
},
|
| 1907 |
-
{
|
| 1908 |
-
"epoch": 4.57,
|
| 1909 |
-
"eval_loss": 0.4442508816719055,
|
| 1910 |
-
"eval_runtime": 5.775,
|
| 1911 |
-
"eval_samples_per_second": 202.77,
|
| 1912 |
-
"eval_steps_per_second": 12.814,
|
| 1913 |
-
"step": 2800
|
| 1914 |
-
},
|
| 1915 |
-
{
|
| 1916 |
-
"epoch": 4.58,
|
| 1917 |
-
"learning_rate": 0.0005896980461811723,
|
| 1918 |
-
"loss": 0.4589,
|
| 1919 |
-
"step": 2810
|
| 1920 |
-
},
|
| 1921 |
-
{
|
| 1922 |
-
"epoch": 4.6,
|
| 1923 |
-
"learning_rate": 0.0005879218472468917,
|
| 1924 |
-
"loss": 0.4497,
|
| 1925 |
-
"step": 2820
|
| 1926 |
-
},
|
| 1927 |
-
{
|
| 1928 |
-
"epoch": 4.62,
|
| 1929 |
-
"learning_rate": 0.0005861456483126111,
|
| 1930 |
-
"loss": 0.4618,
|
| 1931 |
-
"step": 2830
|
| 1932 |
-
},
|
| 1933 |
-
{
|
| 1934 |
-
"epoch": 4.63,
|
| 1935 |
-
"learning_rate": 0.0005843694493783304,
|
| 1936 |
-
"loss": 0.4897,
|
| 1937 |
-
"step": 2840
|
| 1938 |
-
},
|
| 1939 |
-
{
|
| 1940 |
-
"epoch": 4.65,
|
| 1941 |
-
"learning_rate": 0.0005825932504440498,
|
| 1942 |
-
"loss": 0.4577,
|
| 1943 |
-
"step": 2850
|
| 1944 |
-
},
|
| 1945 |
-
{
|
| 1946 |
-
"epoch": 4.67,
|
| 1947 |
-
"learning_rate": 0.0005808170515097692,
|
| 1948 |
-
"loss": 0.474,
|
| 1949 |
-
"step": 2860
|
| 1950 |
-
},
|
| 1951 |
-
{
|
| 1952 |
-
"epoch": 4.68,
|
| 1953 |
-
"learning_rate": 0.0005790408525754885,
|
| 1954 |
-
"loss": 0.4409,
|
| 1955 |
-
"step": 2870
|
| 1956 |
-
},
|
| 1957 |
-
{
|
| 1958 |
-
"epoch": 4.7,
|
| 1959 |
-
"learning_rate": 0.0005772646536412079,
|
| 1960 |
-
"loss": 0.4716,
|
| 1961 |
-
"step": 2880
|
| 1962 |
-
},
|
| 1963 |
-
{
|
| 1964 |
-
"epoch": 4.71,
|
| 1965 |
-
"learning_rate": 0.0005754884547069273,
|
| 1966 |
-
"loss": 0.4833,
|
| 1967 |
-
"step": 2890
|
| 1968 |
-
},
|
| 1969 |
-
{
|
| 1970 |
-
"epoch": 4.73,
|
| 1971 |
-
"learning_rate": 0.0005737122557726465,
|
| 1972 |
-
"loss": 0.4487,
|
| 1973 |
-
"step": 2900
|
| 1974 |
-
},
|
| 1975 |
-
{
|
| 1976 |
-
"epoch": 4.73,
|
| 1977 |
-
"eval_loss": 0.439519464969635,
|
| 1978 |
-
"eval_runtime": 5.776,
|
| 1979 |
-
"eval_samples_per_second": 202.735,
|
| 1980 |
-
"eval_steps_per_second": 12.812,
|
| 1981 |
-
"step": 2900
|
| 1982 |
-
},
|
| 1983 |
-
{
|
| 1984 |
-
"epoch": 4.75,
|
| 1985 |
-
"learning_rate": 0.0005719360568383659,
|
| 1986 |
-
"loss": 0.4426,
|
| 1987 |
-
"step": 2910
|
| 1988 |
-
},
|
| 1989 |
-
{
|
| 1990 |
-
"epoch": 4.76,
|
| 1991 |
-
"learning_rate": 0.0005701598579040853,
|
| 1992 |
-
"loss": 0.4588,
|
| 1993 |
-
"step": 2920
|
| 1994 |
-
},
|
| 1995 |
-
{
|
| 1996 |
-
"epoch": 4.78,
|
| 1997 |
-
"learning_rate": 0.0005683836589698046,
|
| 1998 |
-
"loss": 0.4676,
|
| 1999 |
-
"step": 2930
|
| 2000 |
-
},
|
| 2001 |
-
{
|
| 2002 |
-
"epoch": 4.8,
|
| 2003 |
-
"learning_rate": 0.000566607460035524,
|
| 2004 |
-
"loss": 0.5016,
|
| 2005 |
-
"step": 2940
|
| 2006 |
-
},
|
| 2007 |
-
{
|
| 2008 |
-
"epoch": 4.81,
|
| 2009 |
-
"learning_rate": 0.0005648312611012434,
|
| 2010 |
-
"loss": 0.4464,
|
| 2011 |
-
"step": 2950
|
| 2012 |
-
},
|
| 2013 |
-
{
|
| 2014 |
-
"epoch": 4.83,
|
| 2015 |
-
"learning_rate": 0.0005630550621669627,
|
| 2016 |
-
"loss": 0.4589,
|
| 2017 |
-
"step": 2960
|
| 2018 |
-
},
|
| 2019 |
-
{
|
| 2020 |
-
"epoch": 4.85,
|
| 2021 |
-
"learning_rate": 0.0005612788632326821,
|
| 2022 |
-
"loss": 0.4499,
|
| 2023 |
-
"step": 2970
|
| 2024 |
-
},
|
| 2025 |
-
{
|
| 2026 |
-
"epoch": 4.86,
|
| 2027 |
-
"learning_rate": 0.0005595026642984015,
|
| 2028 |
-
"loss": 0.4361,
|
| 2029 |
-
"step": 2980
|
| 2030 |
-
},
|
| 2031 |
-
{
|
| 2032 |
-
"epoch": 4.88,
|
| 2033 |
-
"learning_rate": 0.0005577264653641208,
|
| 2034 |
-
"loss": 0.4437,
|
| 2035 |
-
"step": 2990
|
| 2036 |
-
},
|
| 2037 |
-
{
|
| 2038 |
-
"epoch": 4.89,
|
| 2039 |
-
"learning_rate": 0.0005559502664298401,
|
| 2040 |
-
"loss": 0.4724,
|
| 2041 |
-
"step": 3000
|
| 2042 |
-
},
|
| 2043 |
-
{
|
| 2044 |
-
"epoch": 4.89,
|
| 2045 |
-
"eval_loss": 0.43853452801704407,
|
| 2046 |
-
"eval_runtime": 5.773,
|
| 2047 |
-
"eval_samples_per_second": 202.84,
|
| 2048 |
-
"eval_steps_per_second": 12.818,
|
| 2049 |
-
"step": 3000
|
| 2050 |
-
},
|
| 2051 |
-
{
|
| 2052 |
-
"epoch": 4.91,
|
| 2053 |
-
"learning_rate": 0.0005541740674955595,
|
| 2054 |
-
"loss": 0.4927,
|
| 2055 |
-
"step": 3010
|
| 2056 |
-
},
|
| 2057 |
-
{
|
| 2058 |
-
"epoch": 4.93,
|
| 2059 |
-
"learning_rate": 0.0005523978685612788,
|
| 2060 |
-
"loss": 0.4705,
|
| 2061 |
-
"step": 3020
|
| 2062 |
-
},
|
| 2063 |
-
{
|
| 2064 |
-
"epoch": 4.94,
|
| 2065 |
-
"learning_rate": 0.0005506216696269982,
|
| 2066 |
-
"loss": 0.4356,
|
| 2067 |
-
"step": 3030
|
| 2068 |
-
},
|
| 2069 |
-
{
|
| 2070 |
-
"epoch": 4.96,
|
| 2071 |
-
"learning_rate": 0.0005488454706927176,
|
| 2072 |
-
"loss": 0.5064,
|
| 2073 |
-
"step": 3040
|
| 2074 |
-
},
|
| 2075 |
-
{
|
| 2076 |
-
"epoch": 4.98,
|
| 2077 |
-
"learning_rate": 0.0005470692717584369,
|
| 2078 |
-
"loss": 0.4611,
|
| 2079 |
-
"step": 3050
|
| 2080 |
-
},
|
| 2081 |
-
{
|
| 2082 |
-
"epoch": 4.99,
|
| 2083 |
-
"learning_rate": 0.0005452930728241563,
|
| 2084 |
-
"loss": 0.4797,
|
| 2085 |
-
"step": 3060
|
| 2086 |
-
},
|
| 2087 |
-
{
|
| 2088 |
-
"epoch": 5.01,
|
| 2089 |
-
"learning_rate": 0.0005435168738898757,
|
| 2090 |
-
"loss": 0.432,
|
| 2091 |
-
"step": 3070
|
| 2092 |
-
},
|
| 2093 |
-
{
|
| 2094 |
-
"epoch": 5.02,
|
| 2095 |
-
"learning_rate": 0.0005417406749555951,
|
| 2096 |
-
"loss": 0.4598,
|
| 2097 |
-
"step": 3080
|
| 2098 |
-
},
|
| 2099 |
-
{
|
| 2100 |
-
"epoch": 5.04,
|
| 2101 |
-
"learning_rate": 0.0005399644760213144,
|
| 2102 |
-
"loss": 0.4524,
|
| 2103 |
-
"step": 3090
|
| 2104 |
-
},
|
| 2105 |
-
{
|
| 2106 |
-
"epoch": 5.06,
|
| 2107 |
-
"learning_rate": 0.0005381882770870338,
|
| 2108 |
-
"loss": 0.4553,
|
| 2109 |
-
"step": 3100
|
| 2110 |
-
},
|
| 2111 |
-
{
|
| 2112 |
-
"epoch": 5.06,
|
| 2113 |
-
"eval_loss": 0.43742743134498596,
|
| 2114 |
-
"eval_runtime": 5.776,
|
| 2115 |
-
"eval_samples_per_second": 202.735,
|
| 2116 |
-
"eval_steps_per_second": 12.812,
|
| 2117 |
-
"step": 3100
|
| 2118 |
-
},
|
| 2119 |
-
{
|
| 2120 |
-
"epoch": 5.07,
|
| 2121 |
-
"learning_rate": 0.0005364120781527531,
|
| 2122 |
-
"loss": 0.4454,
|
| 2123 |
-
"step": 3110
|
| 2124 |
-
},
|
| 2125 |
-
{
|
| 2126 |
-
"epoch": 5.09,
|
| 2127 |
-
"learning_rate": 0.0005346358792184724,
|
| 2128 |
-
"loss": 0.4436,
|
| 2129 |
-
"step": 3120
|
| 2130 |
-
},
|
| 2131 |
-
{
|
| 2132 |
-
"epoch": 5.11,
|
| 2133 |
-
"learning_rate": 0.0005328596802841918,
|
| 2134 |
-
"loss": 0.4414,
|
| 2135 |
-
"step": 3130
|
| 2136 |
-
},
|
| 2137 |
-
{
|
| 2138 |
-
"epoch": 5.12,
|
| 2139 |
-
"learning_rate": 0.0005310834813499112,
|
| 2140 |
-
"loss": 0.4395,
|
| 2141 |
-
"step": 3140
|
| 2142 |
-
},
|
| 2143 |
-
{
|
| 2144 |
-
"epoch": 5.14,
|
| 2145 |
-
"learning_rate": 0.0005293072824156305,
|
| 2146 |
-
"loss": 0.4259,
|
| 2147 |
-
"step": 3150
|
| 2148 |
-
},
|
| 2149 |
-
{
|
| 2150 |
-
"epoch": 5.15,
|
| 2151 |
-
"learning_rate": 0.0005275310834813499,
|
| 2152 |
-
"loss": 0.4486,
|
| 2153 |
-
"step": 3160
|
| 2154 |
-
},
|
| 2155 |
-
{
|
| 2156 |
-
"epoch": 5.17,
|
| 2157 |
-
"learning_rate": 0.0005257548845470693,
|
| 2158 |
-
"loss": 0.4612,
|
| 2159 |
-
"step": 3170
|
| 2160 |
-
},
|
| 2161 |
-
{
|
| 2162 |
-
"epoch": 5.19,
|
| 2163 |
-
"learning_rate": 0.0005239786856127886,
|
| 2164 |
-
"loss": 0.4726,
|
| 2165 |
-
"step": 3180
|
| 2166 |
-
},
|
| 2167 |
-
{
|
| 2168 |
-
"epoch": 5.2,
|
| 2169 |
-
"learning_rate": 0.000522202486678508,
|
| 2170 |
-
"loss": 0.4725,
|
| 2171 |
-
"step": 3190
|
| 2172 |
-
},
|
| 2173 |
-
{
|
| 2174 |
-
"epoch": 5.22,
|
| 2175 |
-
"learning_rate": 0.0005204262877442274,
|
| 2176 |
-
"loss": 0.4798,
|
| 2177 |
-
"step": 3200
|
| 2178 |
-
},
|
| 2179 |
-
{
|
| 2180 |
-
"epoch": 5.22,
|
| 2181 |
-
"eval_loss": 0.4378213882446289,
|
| 2182 |
-
"eval_runtime": 5.571,
|
| 2183 |
-
"eval_samples_per_second": 210.195,
|
| 2184 |
-
"eval_steps_per_second": 13.283,
|
| 2185 |
-
"step": 3200
|
| 2186 |
-
},
|
| 2187 |
-
{
|
| 2188 |
-
"epoch": 5.24,
|
| 2189 |
-
"learning_rate": 0.0005186500888099467,
|
| 2190 |
-
"loss": 0.446,
|
| 2191 |
-
"step": 3210
|
| 2192 |
-
},
|
| 2193 |
-
{
|
| 2194 |
-
"epoch": 5.25,
|
| 2195 |
-
"learning_rate": 0.000516873889875666,
|
| 2196 |
-
"loss": 0.4919,
|
| 2197 |
-
"step": 3220
|
| 2198 |
-
},
|
| 2199 |
-
{
|
| 2200 |
-
"epoch": 5.27,
|
| 2201 |
-
"learning_rate": 0.0005150976909413854,
|
| 2202 |
-
"loss": 0.4294,
|
| 2203 |
-
"step": 3230
|
| 2204 |
-
},
|
| 2205 |
-
{
|
| 2206 |
-
"epoch": 5.29,
|
| 2207 |
-
"learning_rate": 0.0005133214920071047,
|
| 2208 |
-
"loss": 0.474,
|
| 2209 |
-
"step": 3240
|
| 2210 |
-
},
|
| 2211 |
-
{
|
| 2212 |
-
"epoch": 5.3,
|
| 2213 |
-
"learning_rate": 0.0005115452930728241,
|
| 2214 |
-
"loss": 0.4951,
|
| 2215 |
-
"step": 3250
|
| 2216 |
-
},
|
| 2217 |
-
{
|
| 2218 |
-
"epoch": 5.32,
|
| 2219 |
-
"learning_rate": 0.0005097690941385435,
|
| 2220 |
-
"loss": 0.4588,
|
| 2221 |
-
"step": 3260
|
| 2222 |
-
},
|
| 2223 |
-
{
|
| 2224 |
-
"epoch": 5.33,
|
| 2225 |
-
"learning_rate": 0.0005079928952042628,
|
| 2226 |
-
"loss": 0.4543,
|
| 2227 |
-
"step": 3270
|
| 2228 |
-
},
|
| 2229 |
-
{
|
| 2230 |
-
"epoch": 5.35,
|
| 2231 |
-
"learning_rate": 0.0005062166962699822,
|
| 2232 |
-
"loss": 0.4488,
|
| 2233 |
-
"step": 3280
|
| 2234 |
-
},
|
| 2235 |
-
{
|
| 2236 |
-
"epoch": 5.37,
|
| 2237 |
-
"learning_rate": 0.0005044404973357016,
|
| 2238 |
-
"loss": 0.4653,
|
| 2239 |
-
"step": 3290
|
| 2240 |
-
},
|
| 2241 |
-
{
|
| 2242 |
-
"epoch": 5.38,
|
| 2243 |
-
"learning_rate": 0.0005026642984014209,
|
| 2244 |
-
"loss": 0.4735,
|
| 2245 |
-
"step": 3300
|
| 2246 |
-
},
|
| 2247 |
-
{
|
| 2248 |
-
"epoch": 5.38,
|
| 2249 |
-
"eval_loss": 0.43299299478530884,
|
| 2250 |
-
"eval_runtime": 5.567,
|
| 2251 |
-
"eval_samples_per_second": 210.346,
|
| 2252 |
-
"eval_steps_per_second": 13.293,
|
| 2253 |
-
"step": 3300
|
| 2254 |
-
},
|
| 2255 |
-
{
|
| 2256 |
-
"epoch": 5.4,
|
| 2257 |
-
"learning_rate": 0.0005008880994671403,
|
| 2258 |
-
"loss": 0.4339,
|
| 2259 |
-
"step": 3310
|
| 2260 |
-
},
|
| 2261 |
-
{
|
| 2262 |
-
"epoch": 5.42,
|
| 2263 |
-
"learning_rate": 0.0004991119005328598,
|
| 2264 |
-
"loss": 0.4268,
|
| 2265 |
-
"step": 3320
|
| 2266 |
-
},
|
| 2267 |
-
{
|
| 2268 |
-
"epoch": 5.43,
|
| 2269 |
-
"learning_rate": 0.000497335701598579,
|
| 2270 |
-
"loss": 0.4174,
|
| 2271 |
-
"step": 3330
|
| 2272 |
-
},
|
| 2273 |
-
{
|
| 2274 |
-
"epoch": 5.45,
|
| 2275 |
-
"learning_rate": 0.0004955595026642984,
|
| 2276 |
-
"loss": 0.4805,
|
| 2277 |
-
"step": 3340
|
| 2278 |
-
},
|
| 2279 |
-
{
|
| 2280 |
-
"epoch": 5.46,
|
| 2281 |
-
"learning_rate": 0.0004937833037300178,
|
| 2282 |
-
"loss": 0.4537,
|
| 2283 |
-
"step": 3350
|
| 2284 |
-
},
|
| 2285 |
-
{
|
| 2286 |
-
"epoch": 5.48,
|
| 2287 |
-
"learning_rate": 0.0004920071047957371,
|
| 2288 |
-
"loss": 0.4559,
|
| 2289 |
-
"step": 3360
|
| 2290 |
-
},
|
| 2291 |
-
{
|
| 2292 |
-
"epoch": 5.5,
|
| 2293 |
-
"learning_rate": 0.0004902309058614565,
|
| 2294 |
-
"loss": 0.4401,
|
| 2295 |
-
"step": 3370
|
| 2296 |
-
},
|
| 2297 |
-
{
|
| 2298 |
-
"epoch": 5.51,
|
| 2299 |
-
"learning_rate": 0.0004884547069271759,
|
| 2300 |
-
"loss": 0.4571,
|
| 2301 |
-
"step": 3380
|
| 2302 |
-
},
|
| 2303 |
-
{
|
| 2304 |
-
"epoch": 5.53,
|
| 2305 |
-
"learning_rate": 0.0004866785079928952,
|
| 2306 |
-
"loss": 0.4396,
|
| 2307 |
-
"step": 3390
|
| 2308 |
-
},
|
| 2309 |
-
{
|
| 2310 |
-
"epoch": 5.55,
|
| 2311 |
-
"learning_rate": 0.0004849023090586146,
|
| 2312 |
-
"loss": 0.4203,
|
| 2313 |
-
"step": 3400
|
| 2314 |
-
},
|
| 2315 |
-
{
|
| 2316 |
-
"epoch": 5.55,
|
| 2317 |
-
"eval_loss": 0.4426453411579132,
|
| 2318 |
-
"eval_runtime": 5.572,
|
| 2319 |
-
"eval_samples_per_second": 210.157,
|
| 2320 |
-
"eval_steps_per_second": 13.281,
|
| 2321 |
-
"step": 3400
|
| 2322 |
-
},
|
| 2323 |
-
{
|
| 2324 |
-
"epoch": 5.56,
|
| 2325 |
-
"learning_rate": 0.0004831261101243339,
|
| 2326 |
-
"loss": 0.4444,
|
| 2327 |
-
"step": 3410
|
| 2328 |
-
},
|
| 2329 |
-
{
|
| 2330 |
-
"epoch": 5.58,
|
| 2331 |
-
"learning_rate": 0.00048134991119005326,
|
| 2332 |
-
"loss": 0.4433,
|
| 2333 |
-
"step": 3420
|
| 2334 |
-
},
|
| 2335 |
-
{
|
| 2336 |
-
"epoch": 5.6,
|
| 2337 |
-
"learning_rate": 0.00047957371225577266,
|
| 2338 |
-
"loss": 0.4328,
|
| 2339 |
-
"step": 3430
|
| 2340 |
-
},
|
| 2341 |
-
{
|
| 2342 |
-
"epoch": 5.61,
|
| 2343 |
-
"learning_rate": 0.000477797513321492,
|
| 2344 |
-
"loss": 0.4257,
|
| 2345 |
-
"step": 3440
|
| 2346 |
-
},
|
| 2347 |
-
{
|
| 2348 |
-
"epoch": 5.63,
|
| 2349 |
-
"learning_rate": 0.00047602131438721133,
|
| 2350 |
-
"loss": 0.4804,
|
| 2351 |
-
"step": 3450
|
| 2352 |
-
},
|
| 2353 |
-
{
|
| 2354 |
-
"epoch": 5.64,
|
| 2355 |
-
"learning_rate": 0.00047424511545293073,
|
| 2356 |
-
"loss": 0.4396,
|
| 2357 |
-
"step": 3460
|
| 2358 |
-
},
|
| 2359 |
-
{
|
| 2360 |
-
"epoch": 5.66,
|
| 2361 |
-
"learning_rate": 0.00047246891651865007,
|
| 2362 |
-
"loss": 0.4498,
|
| 2363 |
-
"step": 3470
|
| 2364 |
-
},
|
| 2365 |
-
{
|
| 2366 |
-
"epoch": 5.68,
|
| 2367 |
-
"learning_rate": 0.00047069271758436946,
|
| 2368 |
-
"loss": 0.4799,
|
| 2369 |
-
"step": 3480
|
| 2370 |
-
},
|
| 2371 |
-
{
|
| 2372 |
-
"epoch": 5.69,
|
| 2373 |
-
"learning_rate": 0.00046891651865008885,
|
| 2374 |
-
"loss": 0.4736,
|
| 2375 |
-
"step": 3490
|
| 2376 |
-
},
|
| 2377 |
-
{
|
| 2378 |
-
"epoch": 5.71,
|
| 2379 |
-
"learning_rate": 0.0004671403197158082,
|
| 2380 |
-
"loss": 0.4252,
|
| 2381 |
-
"step": 3500
|
| 2382 |
-
},
|
| 2383 |
-
{
|
| 2384 |
-
"epoch": 5.71,
|
| 2385 |
-
"eval_loss": 0.441283255815506,
|
| 2386 |
-
"eval_runtime": 5.574,
|
| 2387 |
-
"eval_samples_per_second": 210.082,
|
| 2388 |
-
"eval_steps_per_second": 13.276,
|
| 2389 |
-
"step": 3500
|
| 2390 |
-
},
|
| 2391 |
-
{
|
| 2392 |
-
"epoch": 5.73,
|
| 2393 |
-
"learning_rate": 0.0004653641207815276,
|
| 2394 |
-
"loss": 0.4661,
|
| 2395 |
-
"step": 3510
|
| 2396 |
-
},
|
| 2397 |
-
{
|
| 2398 |
-
"epoch": 5.74,
|
| 2399 |
-
"learning_rate": 0.0004635879218472469,
|
| 2400 |
-
"loss": 0.4545,
|
| 2401 |
-
"step": 3520
|
| 2402 |
-
},
|
| 2403 |
-
{
|
| 2404 |
-
"epoch": 5.76,
|
| 2405 |
-
"learning_rate": 0.00046181172291296627,
|
| 2406 |
-
"loss": 0.4573,
|
| 2407 |
-
"step": 3530
|
| 2408 |
-
},
|
| 2409 |
-
{
|
| 2410 |
-
"epoch": 5.77,
|
| 2411 |
-
"learning_rate": 0.00046003552397868566,
|
| 2412 |
-
"loss": 0.4464,
|
| 2413 |
-
"step": 3540
|
| 2414 |
-
},
|
| 2415 |
-
{
|
| 2416 |
-
"epoch": 5.79,
|
| 2417 |
-
"learning_rate": 0.000458259325044405,
|
| 2418 |
-
"loss": 0.4089,
|
| 2419 |
-
"step": 3550
|
| 2420 |
-
},
|
| 2421 |
-
{
|
| 2422 |
-
"epoch": 5.81,
|
| 2423 |
-
"learning_rate": 0.00045648312611012434,
|
| 2424 |
-
"loss": 0.4622,
|
| 2425 |
-
"step": 3560
|
| 2426 |
-
},
|
| 2427 |
-
{
|
| 2428 |
-
"epoch": 5.82,
|
| 2429 |
-
"learning_rate": 0.00045470692717584373,
|
| 2430 |
-
"loss": 0.4611,
|
| 2431 |
-
"step": 3570
|
| 2432 |
-
},
|
| 2433 |
-
{
|
| 2434 |
-
"epoch": 5.84,
|
| 2435 |
-
"learning_rate": 0.00045293072824156307,
|
| 2436 |
-
"loss": 0.4524,
|
| 2437 |
-
"step": 3580
|
| 2438 |
-
},
|
| 2439 |
-
{
|
| 2440 |
-
"epoch": 5.86,
|
| 2441 |
-
"learning_rate": 0.0004511545293072824,
|
| 2442 |
-
"loss": 0.439,
|
| 2443 |
-
"step": 3590
|
| 2444 |
-
},
|
| 2445 |
-
{
|
| 2446 |
-
"epoch": 5.87,
|
| 2447 |
-
"learning_rate": 0.0004493783303730018,
|
| 2448 |
-
"loss": 0.4246,
|
| 2449 |
-
"step": 3600
|
| 2450 |
-
},
|
| 2451 |
-
{
|
| 2452 |
-
"epoch": 5.87,
|
| 2453 |
-
"eval_loss": 0.4371810257434845,
|
| 2454 |
-
"eval_runtime": 5.588,
|
| 2455 |
-
"eval_samples_per_second": 209.555,
|
| 2456 |
-
"eval_steps_per_second": 13.243,
|
| 2457 |
-
"step": 3600
|
| 2458 |
-
},
|
| 2459 |
-
{
|
| 2460 |
-
"epoch": 5.89,
|
| 2461 |
-
"learning_rate": 0.00044760213143872114,
|
| 2462 |
-
"loss": 0.4645,
|
| 2463 |
-
"step": 3610
|
| 2464 |
-
},
|
| 2465 |
-
{
|
| 2466 |
-
"epoch": 5.91,
|
| 2467 |
-
"learning_rate": 0.0004458259325044405,
|
| 2468 |
-
"loss": 0.454,
|
| 2469 |
-
"step": 3620
|
| 2470 |
-
},
|
| 2471 |
-
{
|
| 2472 |
-
"epoch": 5.92,
|
| 2473 |
-
"learning_rate": 0.0004440497335701599,
|
| 2474 |
-
"loss": 0.45,
|
| 2475 |
-
"step": 3630
|
| 2476 |
-
},
|
| 2477 |
-
{
|
| 2478 |
-
"epoch": 5.94,
|
| 2479 |
-
"learning_rate": 0.0004422735346358792,
|
| 2480 |
-
"loss": 0.439,
|
| 2481 |
-
"step": 3640
|
| 2482 |
-
},
|
| 2483 |
-
{
|
| 2484 |
-
"epoch": 5.95,
|
| 2485 |
-
"learning_rate": 0.0004404973357015986,
|
| 2486 |
-
"loss": 0.4276,
|
| 2487 |
-
"step": 3650
|
| 2488 |
-
},
|
| 2489 |
-
{
|
| 2490 |
-
"epoch": 5.97,
|
| 2491 |
-
"learning_rate": 0.00043872113676731795,
|
| 2492 |
-
"loss": 0.4537,
|
| 2493 |
-
"step": 3660
|
| 2494 |
-
},
|
| 2495 |
-
{
|
| 2496 |
-
"epoch": 5.99,
|
| 2497 |
-
"learning_rate": 0.0004369449378330373,
|
| 2498 |
-
"loss": 0.4647,
|
| 2499 |
-
"step": 3670
|
| 2500 |
-
},
|
| 2501 |
-
{
|
| 2502 |
-
"epoch": 6.0,
|
| 2503 |
-
"learning_rate": 0.0004351687388987567,
|
| 2504 |
-
"loss": 0.4192,
|
| 2505 |
-
"step": 3680
|
| 2506 |
-
},
|
| 2507 |
-
{
|
| 2508 |
-
"epoch": 6.02,
|
| 2509 |
-
"learning_rate": 0.000433392539964476,
|
| 2510 |
-
"loss": 0.4178,
|
| 2511 |
-
"step": 3690
|
| 2512 |
-
},
|
| 2513 |
-
{
|
| 2514 |
-
"epoch": 6.04,
|
| 2515 |
-
"learning_rate": 0.00043161634103019536,
|
| 2516 |
-
"loss": 0.4158,
|
| 2517 |
-
"step": 3700
|
| 2518 |
-
},
|
| 2519 |
-
{
|
| 2520 |
-
"epoch": 6.04,
|
| 2521 |
-
"eval_loss": 0.43516233563423157,
|
| 2522 |
-
"eval_runtime": 5.574,
|
| 2523 |
-
"eval_samples_per_second": 210.082,
|
| 2524 |
-
"eval_steps_per_second": 13.276,
|
| 2525 |
-
"step": 3700
|
| 2526 |
-
},
|
| 2527 |
-
{
|
| 2528 |
-
"epoch": 6.05,
|
| 2529 |
-
"learning_rate": 0.00042984014209591475,
|
| 2530 |
-
"loss": 0.4581,
|
| 2531 |
-
"step": 3710
|
| 2532 |
-
},
|
| 2533 |
-
{
|
| 2534 |
-
"epoch": 6.07,
|
| 2535 |
-
"learning_rate": 0.0004280639431616341,
|
| 2536 |
-
"loss": 0.4422,
|
| 2537 |
-
"step": 3720
|
| 2538 |
-
},
|
| 2539 |
-
{
|
| 2540 |
-
"epoch": 6.08,
|
| 2541 |
-
"learning_rate": 0.00042628774422735343,
|
| 2542 |
-
"loss": 0.4191,
|
| 2543 |
-
"step": 3730
|
| 2544 |
-
},
|
| 2545 |
-
{
|
| 2546 |
-
"epoch": 6.1,
|
| 2547 |
-
"learning_rate": 0.0004245115452930728,
|
| 2548 |
-
"loss": 0.434,
|
| 2549 |
-
"step": 3740
|
| 2550 |
-
},
|
| 2551 |
-
{
|
| 2552 |
-
"epoch": 6.12,
|
| 2553 |
-
"learning_rate": 0.00042273534635879216,
|
| 2554 |
-
"loss": 0.416,
|
| 2555 |
-
"step": 3750
|
| 2556 |
-
},
|
| 2557 |
-
{
|
| 2558 |
-
"epoch": 6.13,
|
| 2559 |
-
"learning_rate": 0.00042095914742451156,
|
| 2560 |
-
"loss": 0.4116,
|
| 2561 |
-
"step": 3760
|
| 2562 |
-
},
|
| 2563 |
-
{
|
| 2564 |
-
"epoch": 6.15,
|
| 2565 |
-
"learning_rate": 0.0004191829484902309,
|
| 2566 |
-
"loss": 0.4545,
|
| 2567 |
-
"step": 3770
|
| 2568 |
-
},
|
| 2569 |
-
{
|
| 2570 |
-
"epoch": 6.17,
|
| 2571 |
-
"learning_rate": 0.00041740674955595023,
|
| 2572 |
-
"loss": 0.4507,
|
| 2573 |
-
"step": 3780
|
| 2574 |
-
},
|
| 2575 |
-
{
|
| 2576 |
-
"epoch": 6.18,
|
| 2577 |
-
"learning_rate": 0.0004156305506216697,
|
| 2578 |
-
"loss": 0.4345,
|
| 2579 |
-
"step": 3790
|
| 2580 |
-
},
|
| 2581 |
-
{
|
| 2582 |
-
"epoch": 6.2,
|
| 2583 |
-
"learning_rate": 0.000413854351687389,
|
| 2584 |
-
"loss": 0.4603,
|
| 2585 |
-
"step": 3800
|
| 2586 |
-
},
|
| 2587 |
-
{
|
| 2588 |
-
"epoch": 6.2,
|
| 2589 |
-
"eval_loss": 0.43294015526771545,
|
| 2590 |
-
"eval_runtime": 5.595,
|
| 2591 |
-
"eval_samples_per_second": 209.293,
|
| 2592 |
-
"eval_steps_per_second": 13.226,
|
| 2593 |
-
"step": 3800
|
| 2594 |
-
},
|
| 2595 |
-
{
|
| 2596 |
-
"epoch": 6.22,
|
| 2597 |
-
"learning_rate": 0.00041207815275310836,
|
| 2598 |
-
"loss": 0.4586,
|
| 2599 |
-
"step": 3810
|
| 2600 |
-
},
|
| 2601 |
-
{
|
| 2602 |
-
"epoch": 6.23,
|
| 2603 |
-
"learning_rate": 0.00041030195381882775,
|
| 2604 |
-
"loss": 0.4205,
|
| 2605 |
-
"step": 3820
|
| 2606 |
-
},
|
| 2607 |
-
{
|
| 2608 |
-
"epoch": 6.25,
|
| 2609 |
-
"learning_rate": 0.0004085257548845471,
|
| 2610 |
-
"loss": 0.4412,
|
| 2611 |
-
"step": 3830
|
| 2612 |
-
},
|
| 2613 |
-
{
|
| 2614 |
-
"epoch": 6.26,
|
| 2615 |
-
"learning_rate": 0.00040674955595026643,
|
| 2616 |
-
"loss": 0.4632,
|
| 2617 |
-
"step": 3840
|
| 2618 |
-
},
|
| 2619 |
-
{
|
| 2620 |
-
"epoch": 6.28,
|
| 2621 |
-
"learning_rate": 0.0004049733570159858,
|
| 2622 |
-
"loss": 0.4498,
|
| 2623 |
-
"step": 3850
|
| 2624 |
-
},
|
| 2625 |
-
{
|
| 2626 |
-
"epoch": 6.3,
|
| 2627 |
-
"learning_rate": 0.00040319715808170517,
|
| 2628 |
-
"loss": 0.4457,
|
| 2629 |
-
"step": 3860
|
| 2630 |
-
},
|
| 2631 |
-
{
|
| 2632 |
-
"epoch": 6.31,
|
| 2633 |
-
"learning_rate": 0.00040142095914742456,
|
| 2634 |
-
"loss": 0.4461,
|
| 2635 |
-
"step": 3870
|
| 2636 |
-
},
|
| 2637 |
-
{
|
| 2638 |
-
"epoch": 6.33,
|
| 2639 |
-
"learning_rate": 0.0003996447602131439,
|
| 2640 |
-
"loss": 0.4239,
|
| 2641 |
-
"step": 3880
|
| 2642 |
-
},
|
| 2643 |
-
{
|
| 2644 |
-
"epoch": 6.35,
|
| 2645 |
-
"learning_rate": 0.00039786856127886324,
|
| 2646 |
-
"loss": 0.4439,
|
| 2647 |
-
"step": 3890
|
| 2648 |
-
},
|
| 2649 |
-
{
|
| 2650 |
-
"epoch": 6.36,
|
| 2651 |
-
"learning_rate": 0.00039609236234458263,
|
| 2652 |
-
"loss": 0.4095,
|
| 2653 |
-
"step": 3900
|
| 2654 |
-
},
|
| 2655 |
-
{
|
| 2656 |
-
"epoch": 6.36,
|
| 2657 |
-
"eval_loss": 0.4296843111515045,
|
| 2658 |
-
"eval_runtime": 5.575,
|
| 2659 |
-
"eval_samples_per_second": 210.044,
|
| 2660 |
-
"eval_steps_per_second": 13.273,
|
| 2661 |
-
"step": 3900
|
| 2662 |
-
},
|
| 2663 |
-
{
|
| 2664 |
-
"epoch": 6.38,
|
| 2665 |
-
"learning_rate": 0.00039431616341030197,
|
| 2666 |
-
"loss": 0.4338,
|
| 2667 |
-
"step": 3910
|
| 2668 |
-
},
|
| 2669 |
-
{
|
| 2670 |
-
"epoch": 6.39,
|
| 2671 |
-
"learning_rate": 0.0003925399644760213,
|
| 2672 |
-
"loss": 0.4279,
|
| 2673 |
-
"step": 3920
|
| 2674 |
-
},
|
| 2675 |
-
{
|
| 2676 |
-
"epoch": 6.41,
|
| 2677 |
-
"learning_rate": 0.0003907637655417407,
|
| 2678 |
-
"loss": 0.4214,
|
| 2679 |
-
"step": 3930
|
| 2680 |
-
},
|
| 2681 |
-
{
|
| 2682 |
-
"epoch": 6.43,
|
| 2683 |
-
"learning_rate": 0.00038898756660746004,
|
| 2684 |
-
"loss": 0.4151,
|
| 2685 |
-
"step": 3940
|
| 2686 |
-
},
|
| 2687 |
-
{
|
| 2688 |
-
"epoch": 6.44,
|
| 2689 |
-
"learning_rate": 0.0003872113676731794,
|
| 2690 |
-
"loss": 0.4393,
|
| 2691 |
-
"step": 3950
|
| 2692 |
-
},
|
| 2693 |
-
{
|
| 2694 |
-
"epoch": 6.46,
|
| 2695 |
-
"learning_rate": 0.0003854351687388988,
|
| 2696 |
-
"loss": 0.4259,
|
| 2697 |
-
"step": 3960
|
| 2698 |
-
},
|
| 2699 |
-
{
|
| 2700 |
-
"epoch": 6.48,
|
| 2701 |
-
"learning_rate": 0.0003836589698046181,
|
| 2702 |
-
"loss": 0.4354,
|
| 2703 |
-
"step": 3970
|
| 2704 |
-
},
|
| 2705 |
-
{
|
| 2706 |
-
"epoch": 6.49,
|
| 2707 |
-
"learning_rate": 0.00038188277087033745,
|
| 2708 |
-
"loss": 0.4135,
|
| 2709 |
-
"step": 3980
|
| 2710 |
-
},
|
| 2711 |
-
{
|
| 2712 |
-
"epoch": 6.51,
|
| 2713 |
-
"learning_rate": 0.00038010657193605685,
|
| 2714 |
-
"loss": 0.4317,
|
| 2715 |
-
"step": 3990
|
| 2716 |
-
},
|
| 2717 |
-
{
|
| 2718 |
-
"epoch": 6.53,
|
| 2719 |
-
"learning_rate": 0.0003783303730017762,
|
| 2720 |
-
"loss": 0.453,
|
| 2721 |
-
"step": 4000
|
| 2722 |
-
},
|
| 2723 |
-
{
|
| 2724 |
-
"epoch": 6.53,
|
| 2725 |
-
"eval_loss": 0.43606194853782654,
|
| 2726 |
-
"eval_runtime": 5.568,
|
| 2727 |
-
"eval_samples_per_second": 210.308,
|
| 2728 |
-
"eval_steps_per_second": 13.29,
|
| 2729 |
-
"step": 4000
|
| 2730 |
-
},
|
| 2731 |
-
{
|
| 2732 |
-
"epoch": 6.54,
|
| 2733 |
-
"learning_rate": 0.0003765541740674956,
|
| 2734 |
-
"loss": 0.4468,
|
| 2735 |
-
"step": 4010
|
| 2736 |
-
},
|
| 2737 |
-
{
|
| 2738 |
-
"epoch": 6.56,
|
| 2739 |
-
"learning_rate": 0.0003747779751332149,
|
| 2740 |
-
"loss": 0.4104,
|
| 2741 |
-
"step": 4020
|
| 2742 |
-
},
|
| 2743 |
-
{
|
| 2744 |
-
"epoch": 6.57,
|
| 2745 |
-
"learning_rate": 0.00037300177619893426,
|
| 2746 |
-
"loss": 0.4593,
|
| 2747 |
-
"step": 4030
|
| 2748 |
-
},
|
| 2749 |
-
{
|
| 2750 |
-
"epoch": 6.59,
|
| 2751 |
-
"learning_rate": 0.00037122557726465365,
|
| 2752 |
-
"loss": 0.4376,
|
| 2753 |
-
"step": 4040
|
| 2754 |
-
},
|
| 2755 |
-
{
|
| 2756 |
-
"epoch": 6.61,
|
| 2757 |
-
"learning_rate": 0.000369449378330373,
|
| 2758 |
-
"loss": 0.4585,
|
| 2759 |
-
"step": 4050
|
| 2760 |
-
},
|
| 2761 |
-
{
|
| 2762 |
-
"epoch": 6.62,
|
| 2763 |
-
"learning_rate": 0.00036767317939609233,
|
| 2764 |
-
"loss": 0.4177,
|
| 2765 |
-
"step": 4060
|
| 2766 |
-
},
|
| 2767 |
-
{
|
| 2768 |
-
"epoch": 6.64,
|
| 2769 |
-
"learning_rate": 0.0003658969804618117,
|
| 2770 |
-
"loss": 0.4852,
|
| 2771 |
-
"step": 4070
|
| 2772 |
-
},
|
| 2773 |
-
{
|
| 2774 |
-
"epoch": 6.66,
|
| 2775 |
-
"learning_rate": 0.00036412078152753106,
|
| 2776 |
-
"loss": 0.4167,
|
| 2777 |
-
"step": 4080
|
| 2778 |
-
},
|
| 2779 |
-
{
|
| 2780 |
-
"epoch": 6.67,
|
| 2781 |
-
"learning_rate": 0.0003623445825932504,
|
| 2782 |
-
"loss": 0.4477,
|
| 2783 |
-
"step": 4090
|
| 2784 |
-
},
|
| 2785 |
-
{
|
| 2786 |
-
"epoch": 6.69,
|
| 2787 |
-
"learning_rate": 0.00036056838365896985,
|
| 2788 |
-
"loss": 0.4219,
|
| 2789 |
-
"step": 4100
|
| 2790 |
-
},
|
| 2791 |
-
{
|
| 2792 |
-
"epoch": 6.69,
|
| 2793 |
-
"eval_loss": 0.4315338730812073,
|
| 2794 |
-
"eval_runtime": 5.58,
|
| 2795 |
-
"eval_samples_per_second": 209.856,
|
| 2796 |
-
"eval_steps_per_second": 13.262,
|
| 2797 |
-
"step": 4100
|
| 2798 |
-
},
|
| 2799 |
-
{
|
| 2800 |
-
"epoch": 6.7,
|
| 2801 |
-
"learning_rate": 0.0003587921847246892,
|
| 2802 |
-
"loss": 0.4345,
|
| 2803 |
-
"step": 4110
|
| 2804 |
-
},
|
| 2805 |
-
{
|
| 2806 |
-
"epoch": 6.72,
|
| 2807 |
-
"learning_rate": 0.0003570159857904086,
|
| 2808 |
-
"loss": 0.4706,
|
| 2809 |
-
"step": 4120
|
| 2810 |
-
},
|
| 2811 |
-
{
|
| 2812 |
-
"epoch": 6.74,
|
| 2813 |
-
"learning_rate": 0.0003552397868561279,
|
| 2814 |
-
"loss": 0.4254,
|
| 2815 |
-
"step": 4130
|
| 2816 |
-
},
|
| 2817 |
-
{
|
| 2818 |
-
"epoch": 6.75,
|
| 2819 |
-
"learning_rate": 0.00035346358792184726,
|
| 2820 |
-
"loss": 0.4316,
|
| 2821 |
-
"step": 4140
|
| 2822 |
-
},
|
| 2823 |
-
{
|
| 2824 |
-
"epoch": 6.77,
|
| 2825 |
-
"learning_rate": 0.00035168738898756665,
|
| 2826 |
-
"loss": 0.4257,
|
| 2827 |
-
"step": 4150
|
| 2828 |
-
},
|
| 2829 |
-
{
|
| 2830 |
-
"epoch": 6.79,
|
| 2831 |
-
"learning_rate": 0.000349911190053286,
|
| 2832 |
-
"loss": 0.4332,
|
| 2833 |
-
"step": 4160
|
| 2834 |
-
},
|
| 2835 |
-
{
|
| 2836 |
-
"epoch": 6.8,
|
| 2837 |
-
"learning_rate": 0.00034813499111900533,
|
| 2838 |
-
"loss": 0.4652,
|
| 2839 |
-
"step": 4170
|
| 2840 |
-
},
|
| 2841 |
-
{
|
| 2842 |
-
"epoch": 6.82,
|
| 2843 |
-
"learning_rate": 0.0003463587921847247,
|
| 2844 |
-
"loss": 0.4515,
|
| 2845 |
-
"step": 4180
|
| 2846 |
-
},
|
| 2847 |
-
{
|
| 2848 |
-
"epoch": 6.84,
|
| 2849 |
-
"learning_rate": 0.00034458259325044407,
|
| 2850 |
-
"loss": 0.4579,
|
| 2851 |
-
"step": 4190
|
| 2852 |
-
},
|
| 2853 |
-
{
|
| 2854 |
-
"epoch": 6.85,
|
| 2855 |
-
"learning_rate": 0.0003428063943161634,
|
| 2856 |
-
"loss": 0.4269,
|
| 2857 |
-
"step": 4200
|
| 2858 |
-
},
|
| 2859 |
-
{
|
| 2860 |
-
"epoch": 6.85,
|
| 2861 |
-
"eval_loss": 0.4239509403705597,
|
| 2862 |
-
"eval_runtime": 5.576,
|
| 2863 |
-
"eval_samples_per_second": 210.006,
|
| 2864 |
-
"eval_steps_per_second": 13.271,
|
| 2865 |
-
"step": 4200
|
| 2866 |
-
},
|
| 2867 |
-
{
|
| 2868 |
-
"epoch": 6.87,
|
| 2869 |
-
"learning_rate": 0.0003410301953818828,
|
| 2870 |
-
"loss": 0.448,
|
| 2871 |
-
"step": 4210
|
| 2872 |
-
},
|
| 2873 |
-
{
|
| 2874 |
-
"epoch": 6.88,
|
| 2875 |
-
"learning_rate": 0.00033925399644760214,
|
| 2876 |
-
"loss": 0.4353,
|
| 2877 |
-
"step": 4220
|
| 2878 |
-
},
|
| 2879 |
-
{
|
| 2880 |
-
"epoch": 6.9,
|
| 2881 |
-
"learning_rate": 0.0003374777975133215,
|
| 2882 |
-
"loss": 0.439,
|
| 2883 |
-
"step": 4230
|
| 2884 |
-
},
|
| 2885 |
-
{
|
| 2886 |
-
"epoch": 6.92,
|
| 2887 |
-
"learning_rate": 0.00033570159857904087,
|
| 2888 |
-
"loss": 0.4548,
|
| 2889 |
-
"step": 4240
|
| 2890 |
-
},
|
| 2891 |
-
{
|
| 2892 |
-
"epoch": 6.93,
|
| 2893 |
-
"learning_rate": 0.0003339253996447602,
|
| 2894 |
-
"loss": 0.4305,
|
| 2895 |
-
"step": 4250
|
| 2896 |
-
},
|
| 2897 |
-
{
|
| 2898 |
-
"epoch": 6.95,
|
| 2899 |
-
"learning_rate": 0.0003321492007104796,
|
| 2900 |
-
"loss": 0.4385,
|
| 2901 |
-
"step": 4260
|
| 2902 |
-
},
|
| 2903 |
-
{
|
| 2904 |
-
"epoch": 6.97,
|
| 2905 |
-
"learning_rate": 0.00033037300177619894,
|
| 2906 |
-
"loss": 0.4737,
|
| 2907 |
-
"step": 4270
|
| 2908 |
-
},
|
| 2909 |
-
{
|
| 2910 |
-
"epoch": 6.98,
|
| 2911 |
-
"learning_rate": 0.0003285968028419183,
|
| 2912 |
-
"loss": 0.4615,
|
| 2913 |
-
"step": 4280
|
| 2914 |
-
},
|
| 2915 |
-
{
|
| 2916 |
-
"epoch": 7.0,
|
| 2917 |
-
"learning_rate": 0.0003268206039076377,
|
| 2918 |
-
"loss": 0.4474,
|
| 2919 |
-
"step": 4290
|
| 2920 |
-
},
|
| 2921 |
-
{
|
| 2922 |
-
"epoch": 7.01,
|
| 2923 |
-
"learning_rate": 0.000325044404973357,
|
| 2924 |
-
"loss": 0.4098,
|
| 2925 |
-
"step": 4300
|
| 2926 |
-
},
|
| 2927 |
-
{
|
| 2928 |
-
"epoch": 7.01,
|
| 2929 |
-
"eval_loss": 0.4368573725223541,
|
| 2930 |
-
"eval_runtime": 5.581,
|
| 2931 |
-
"eval_samples_per_second": 209.818,
|
| 2932 |
-
"eval_steps_per_second": 13.259,
|
| 2933 |
-
"step": 4300
|
| 2934 |
-
},
|
| 2935 |
-
{
|
| 2936 |
-
"epoch": 7.03,
|
| 2937 |
-
"learning_rate": 0.00032326820603907635,
|
| 2938 |
-
"loss": 0.4292,
|
| 2939 |
-
"step": 4310
|
| 2940 |
-
},
|
| 2941 |
-
{
|
| 2942 |
-
"epoch": 7.05,
|
| 2943 |
-
"learning_rate": 0.00032149200710479575,
|
| 2944 |
-
"loss": 0.4143,
|
| 2945 |
-
"step": 4320
|
| 2946 |
-
},
|
| 2947 |
-
{
|
| 2948 |
-
"epoch": 7.06,
|
| 2949 |
-
"learning_rate": 0.0003197158081705151,
|
| 2950 |
-
"loss": 0.439,
|
| 2951 |
-
"step": 4330
|
| 2952 |
-
},
|
| 2953 |
-
{
|
| 2954 |
-
"epoch": 7.08,
|
| 2955 |
-
"learning_rate": 0.0003179396092362344,
|
| 2956 |
-
"loss": 0.4534,
|
| 2957 |
-
"step": 4340
|
| 2958 |
-
},
|
| 2959 |
-
{
|
| 2960 |
-
"epoch": 7.1,
|
| 2961 |
-
"learning_rate": 0.0003161634103019538,
|
| 2962 |
-
"loss": 0.4361,
|
| 2963 |
-
"step": 4350
|
| 2964 |
-
},
|
| 2965 |
-
{
|
| 2966 |
-
"epoch": 7.11,
|
| 2967 |
-
"learning_rate": 0.00031438721136767316,
|
| 2968 |
-
"loss": 0.4249,
|
| 2969 |
-
"step": 4360
|
| 2970 |
-
},
|
| 2971 |
-
{
|
| 2972 |
-
"epoch": 7.13,
|
| 2973 |
-
"learning_rate": 0.00031261101243339255,
|
| 2974 |
-
"loss": 0.455,
|
| 2975 |
-
"step": 4370
|
| 2976 |
-
},
|
| 2977 |
-
{
|
| 2978 |
-
"epoch": 7.15,
|
| 2979 |
-
"learning_rate": 0.0003108348134991119,
|
| 2980 |
-
"loss": 0.4154,
|
| 2981 |
-
"step": 4380
|
| 2982 |
-
},
|
| 2983 |
-
{
|
| 2984 |
-
"epoch": 7.16,
|
| 2985 |
-
"learning_rate": 0.00030905861456483123,
|
| 2986 |
-
"loss": 0.4328,
|
| 2987 |
-
"step": 4390
|
| 2988 |
-
},
|
| 2989 |
-
{
|
| 2990 |
-
"epoch": 7.18,
|
| 2991 |
-
"learning_rate": 0.0003072824156305506,
|
| 2992 |
-
"loss": 0.4153,
|
| 2993 |
-
"step": 4400
|
| 2994 |
-
},
|
| 2995 |
-
{
|
| 2996 |
-
"epoch": 7.18,
|
| 2997 |
-
"eval_loss": 0.43138188123703003,
|
| 2998 |
-
"eval_runtime": 5.569,
|
| 2999 |
-
"eval_samples_per_second": 210.27,
|
| 3000 |
-
"eval_steps_per_second": 13.288,
|
| 3001 |
-
"step": 4400
|
| 3002 |
-
},
|
| 3003 |
-
{
|
| 3004 |
-
"epoch": 7.19,
|
| 3005 |
-
"learning_rate": 0.00030550621669627,
|
| 3006 |
-
"loss": 0.4308,
|
| 3007 |
-
"step": 4410
|
| 3008 |
-
},
|
| 3009 |
-
{
|
| 3010 |
-
"epoch": 7.21,
|
| 3011 |
-
"learning_rate": 0.00030373001776198936,
|
| 3012 |
-
"loss": 0.4271,
|
| 3013 |
-
"step": 4420
|
| 3014 |
-
},
|
| 3015 |
-
{
|
| 3016 |
-
"epoch": 7.23,
|
| 3017 |
-
"learning_rate": 0.00030195381882770875,
|
| 3018 |
-
"loss": 0.4237,
|
| 3019 |
-
"step": 4430
|
| 3020 |
-
},
|
| 3021 |
-
{
|
| 3022 |
-
"epoch": 7.24,
|
| 3023 |
-
"learning_rate": 0.0003001776198934281,
|
| 3024 |
-
"loss": 0.4563,
|
| 3025 |
-
"step": 4440
|
| 3026 |
-
},
|
| 3027 |
-
{
|
| 3028 |
-
"epoch": 7.26,
|
| 3029 |
-
"learning_rate": 0.00029840142095914743,
|
| 3030 |
-
"loss": 0.4374,
|
| 3031 |
-
"step": 4450
|
| 3032 |
-
},
|
| 3033 |
-
{
|
| 3034 |
-
"epoch": 7.28,
|
| 3035 |
-
"learning_rate": 0.0002966252220248668,
|
| 3036 |
-
"loss": 0.4167,
|
| 3037 |
-
"step": 4460
|
| 3038 |
-
},
|
| 3039 |
-
{
|
| 3040 |
-
"epoch": 7.29,
|
| 3041 |
-
"learning_rate": 0.00029484902309058616,
|
| 3042 |
-
"loss": 0.4293,
|
| 3043 |
-
"step": 4470
|
| 3044 |
-
},
|
| 3045 |
-
{
|
| 3046 |
-
"epoch": 7.31,
|
| 3047 |
-
"learning_rate": 0.00029307282415630555,
|
| 3048 |
-
"loss": 0.4065,
|
| 3049 |
-
"step": 4480
|
| 3050 |
-
},
|
| 3051 |
-
{
|
| 3052 |
-
"epoch": 7.32,
|
| 3053 |
-
"learning_rate": 0.0002912966252220249,
|
| 3054 |
-
"loss": 0.434,
|
| 3055 |
-
"step": 4490
|
| 3056 |
-
},
|
| 3057 |
-
{
|
| 3058 |
-
"epoch": 7.34,
|
| 3059 |
-
"learning_rate": 0.00028952042628774423,
|
| 3060 |
-
"loss": 0.4177,
|
| 3061 |
-
"step": 4500
|
| 3062 |
-
},
|
| 3063 |
-
{
|
| 3064 |
-
"epoch": 7.34,
|
| 3065 |
-
"eval_loss": 0.43316009640693665,
|
| 3066 |
-
"eval_runtime": 5.57,
|
| 3067 |
-
"eval_samples_per_second": 210.233,
|
| 3068 |
-
"eval_steps_per_second": 13.285,
|
| 3069 |
-
"step": 4500
|
| 3070 |
-
},
|
| 3071 |
-
{
|
| 3072 |
-
"epoch": 7.36,
|
| 3073 |
-
"learning_rate": 0.0002877442273534636,
|
| 3074 |
-
"loss": 0.4295,
|
| 3075 |
-
"step": 4510
|
| 3076 |
-
},
|
| 3077 |
-
{
|
| 3078 |
-
"epoch": 7.37,
|
| 3079 |
-
"learning_rate": 0.00028596802841918297,
|
| 3080 |
-
"loss": 0.4235,
|
| 3081 |
-
"step": 4520
|
| 3082 |
-
},
|
| 3083 |
-
{
|
| 3084 |
-
"epoch": 7.39,
|
| 3085 |
-
"learning_rate": 0.0002841918294849023,
|
| 3086 |
-
"loss": 0.4207,
|
| 3087 |
-
"step": 4530
|
| 3088 |
-
},
|
| 3089 |
-
{
|
| 3090 |
-
"epoch": 7.41,
|
| 3091 |
-
"learning_rate": 0.0002824156305506217,
|
| 3092 |
-
"loss": 0.4188,
|
| 3093 |
-
"step": 4540
|
| 3094 |
-
},
|
| 3095 |
-
{
|
| 3096 |
-
"epoch": 7.42,
|
| 3097 |
-
"learning_rate": 0.00028063943161634104,
|
| 3098 |
-
"loss": 0.4304,
|
| 3099 |
-
"step": 4550
|
| 3100 |
-
},
|
| 3101 |
-
{
|
| 3102 |
-
"epoch": 7.44,
|
| 3103 |
-
"learning_rate": 0.0002788632326820604,
|
| 3104 |
-
"loss": 0.4522,
|
| 3105 |
-
"step": 4560
|
| 3106 |
-
},
|
| 3107 |
-
{
|
| 3108 |
-
"epoch": 7.46,
|
| 3109 |
-
"learning_rate": 0.00027708703374777977,
|
| 3110 |
-
"loss": 0.4179,
|
| 3111 |
-
"step": 4570
|
| 3112 |
-
},
|
| 3113 |
-
{
|
| 3114 |
-
"epoch": 7.47,
|
| 3115 |
-
"learning_rate": 0.0002753108348134991,
|
| 3116 |
-
"loss": 0.4339,
|
| 3117 |
-
"step": 4580
|
| 3118 |
-
},
|
| 3119 |
-
{
|
| 3120 |
-
"epoch": 7.49,
|
| 3121 |
-
"learning_rate": 0.00027353463587921845,
|
| 3122 |
-
"loss": 0.4207,
|
| 3123 |
-
"step": 4590
|
| 3124 |
-
},
|
| 3125 |
-
{
|
| 3126 |
-
"epoch": 7.5,
|
| 3127 |
-
"learning_rate": 0.00027175843694493784,
|
| 3128 |
-
"loss": 0.4131,
|
| 3129 |
-
"step": 4600
|
| 3130 |
-
},
|
| 3131 |
-
{
|
| 3132 |
-
"epoch": 7.5,
|
| 3133 |
-
"eval_loss": 0.4320705831050873,
|
| 3134 |
-
"eval_runtime": 5.5625,
|
| 3135 |
-
"eval_samples_per_second": 210.516,
|
| 3136 |
-
"eval_steps_per_second": 13.303,
|
| 3137 |
-
"step": 4600
|
| 3138 |
-
},
|
| 3139 |
-
{
|
| 3140 |
-
"epoch": 7.52,
|
| 3141 |
-
"learning_rate": 0.0002699822380106572,
|
| 3142 |
-
"loss": 0.42,
|
| 3143 |
-
"step": 4610
|
| 3144 |
-
},
|
| 3145 |
-
{
|
| 3146 |
-
"epoch": 7.54,
|
| 3147 |
-
"learning_rate": 0.0002682060390763766,
|
| 3148 |
-
"loss": 0.4311,
|
| 3149 |
-
"step": 4620
|
| 3150 |
-
},
|
| 3151 |
-
{
|
| 3152 |
-
"epoch": 7.55,
|
| 3153 |
-
"learning_rate": 0.0002664298401420959,
|
| 3154 |
-
"loss": 0.413,
|
| 3155 |
-
"step": 4630
|
| 3156 |
-
},
|
| 3157 |
-
{
|
| 3158 |
-
"epoch": 7.57,
|
| 3159 |
-
"learning_rate": 0.00026465364120781525,
|
| 3160 |
-
"loss": 0.4403,
|
| 3161 |
-
"step": 4640
|
| 3162 |
-
},
|
| 3163 |
-
{
|
| 3164 |
-
"epoch": 7.59,
|
| 3165 |
-
"learning_rate": 0.00026287744227353465,
|
| 3166 |
-
"loss": 0.4359,
|
| 3167 |
-
"step": 4650
|
| 3168 |
-
},
|
| 3169 |
-
{
|
| 3170 |
-
"epoch": 7.6,
|
| 3171 |
-
"learning_rate": 0.000261101243339254,
|
| 3172 |
-
"loss": 0.4614,
|
| 3173 |
-
"step": 4660
|
| 3174 |
-
},
|
| 3175 |
-
{
|
| 3176 |
-
"epoch": 7.62,
|
| 3177 |
-
"learning_rate": 0.0002593250444049733,
|
| 3178 |
-
"loss": 0.4462,
|
| 3179 |
-
"step": 4670
|
| 3180 |
-
},
|
| 3181 |
-
{
|
| 3182 |
-
"epoch": 7.63,
|
| 3183 |
-
"learning_rate": 0.0002575488454706927,
|
| 3184 |
-
"loss": 0.4333,
|
| 3185 |
-
"step": 4680
|
| 3186 |
-
},
|
| 3187 |
-
{
|
| 3188 |
-
"epoch": 7.65,
|
| 3189 |
-
"learning_rate": 0.00025577264653641206,
|
| 3190 |
-
"loss": 0.4529,
|
| 3191 |
-
"step": 4690
|
| 3192 |
-
},
|
| 3193 |
-
{
|
| 3194 |
-
"epoch": 7.67,
|
| 3195 |
-
"learning_rate": 0.0002539964476021314,
|
| 3196 |
-
"loss": 0.4107,
|
| 3197 |
-
"step": 4700
|
| 3198 |
-
},
|
| 3199 |
-
{
|
| 3200 |
-
"epoch": 7.67,
|
| 3201 |
-
"eval_loss": 0.43221670389175415,
|
| 3202 |
-
"eval_runtime": 5.5625,
|
| 3203 |
-
"eval_samples_per_second": 210.516,
|
| 3204 |
-
"eval_steps_per_second": 13.303,
|
| 3205 |
-
"step": 4700
|
| 3206 |
-
},
|
| 3207 |
-
{
|
| 3208 |
-
"epoch": 7.68,
|
| 3209 |
-
"learning_rate": 0.0002522202486678508,
|
| 3210 |
-
"loss": 0.4248,
|
| 3211 |
-
"step": 4710
|
| 3212 |
-
},
|
| 3213 |
-
{
|
| 3214 |
-
"epoch": 7.7,
|
| 3215 |
-
"learning_rate": 0.00025044404973357013,
|
| 3216 |
-
"loss": 0.4278,
|
| 3217 |
-
"step": 4720
|
| 3218 |
-
},
|
| 3219 |
-
{
|
| 3220 |
-
"epoch": 7.72,
|
| 3221 |
-
"learning_rate": 0.0002486678507992895,
|
| 3222 |
-
"loss": 0.4369,
|
| 3223 |
-
"step": 4730
|
| 3224 |
-
},
|
| 3225 |
-
{
|
| 3226 |
-
"epoch": 7.73,
|
| 3227 |
-
"learning_rate": 0.0002468916518650089,
|
| 3228 |
-
"loss": 0.4287,
|
| 3229 |
-
"step": 4740
|
| 3230 |
-
},
|
| 3231 |
-
{
|
| 3232 |
-
"epoch": 7.75,
|
| 3233 |
-
"learning_rate": 0.00024511545293072826,
|
| 3234 |
-
"loss": 0.4238,
|
| 3235 |
-
"step": 4750
|
| 3236 |
-
},
|
| 3237 |
-
{
|
| 3238 |
-
"epoch": 7.77,
|
| 3239 |
-
"learning_rate": 0.0002433392539964476,
|
| 3240 |
-
"loss": 0.4434,
|
| 3241 |
-
"step": 4760
|
| 3242 |
-
},
|
| 3243 |
-
{
|
| 3244 |
-
"epoch": 7.78,
|
| 3245 |
-
"learning_rate": 0.00024156305506216696,
|
| 3246 |
-
"loss": 0.428,
|
| 3247 |
-
"step": 4770
|
| 3248 |
-
},
|
| 3249 |
-
{
|
| 3250 |
-
"epoch": 7.8,
|
| 3251 |
-
"learning_rate": 0.00023978685612788633,
|
| 3252 |
-
"loss": 0.4409,
|
| 3253 |
-
"step": 4780
|
| 3254 |
-
},
|
| 3255 |
-
{
|
| 3256 |
-
"epoch": 7.81,
|
| 3257 |
-
"learning_rate": 0.00023801065719360567,
|
| 3258 |
-
"loss": 0.4214,
|
| 3259 |
-
"step": 4790
|
| 3260 |
-
},
|
| 3261 |
-
{
|
| 3262 |
-
"epoch": 7.83,
|
| 3263 |
-
"learning_rate": 0.00023623445825932503,
|
| 3264 |
-
"loss": 0.4228,
|
| 3265 |
-
"step": 4800
|
| 3266 |
-
},
|
| 3267 |
-
{
|
| 3268 |
-
"epoch": 7.83,
|
| 3269 |
-
"eval_loss": 0.434493750333786,
|
| 3270 |
-
"eval_runtime": 5.5781,
|
| 3271 |
-
"eval_samples_per_second": 209.927,
|
| 3272 |
-
"eval_steps_per_second": 13.266,
|
| 3273 |
-
"step": 4800
|
| 3274 |
-
},
|
| 3275 |
-
{
|
| 3276 |
-
"epoch": 7.85,
|
| 3277 |
-
"learning_rate": 0.00023445825932504443,
|
| 3278 |
-
"loss": 0.442,
|
| 3279 |
-
"step": 4810
|
| 3280 |
-
},
|
| 3281 |
-
{
|
| 3282 |
-
"epoch": 7.86,
|
| 3283 |
-
"learning_rate": 0.0002326820603907638,
|
| 3284 |
-
"loss": 0.4259,
|
| 3285 |
-
"step": 4820
|
| 3286 |
-
},
|
| 3287 |
-
{
|
| 3288 |
-
"epoch": 7.88,
|
| 3289 |
-
"learning_rate": 0.00023090586145648313,
|
| 3290 |
-
"loss": 0.4453,
|
| 3291 |
-
"step": 4830
|
| 3292 |
-
},
|
| 3293 |
-
{
|
| 3294 |
-
"epoch": 7.9,
|
| 3295 |
-
"learning_rate": 0.0002291296625222025,
|
| 3296 |
-
"loss": 0.4205,
|
| 3297 |
-
"step": 4840
|
| 3298 |
-
},
|
| 3299 |
-
{
|
| 3300 |
-
"epoch": 7.91,
|
| 3301 |
-
"learning_rate": 0.00022735346358792187,
|
| 3302 |
-
"loss": 0.445,
|
| 3303 |
-
"step": 4850
|
| 3304 |
-
},
|
| 3305 |
-
{
|
| 3306 |
-
"epoch": 7.93,
|
| 3307 |
-
"learning_rate": 0.0002255772646536412,
|
| 3308 |
-
"loss": 0.4142,
|
| 3309 |
-
"step": 4860
|
| 3310 |
-
},
|
| 3311 |
-
{
|
| 3312 |
-
"epoch": 7.94,
|
| 3313 |
-
"learning_rate": 0.00022380106571936057,
|
| 3314 |
-
"loss": 0.4717,
|
| 3315 |
-
"step": 4870
|
| 3316 |
-
},
|
| 3317 |
-
{
|
| 3318 |
-
"epoch": 7.96,
|
| 3319 |
-
"learning_rate": 0.00022202486678507994,
|
| 3320 |
-
"loss": 0.4014,
|
| 3321 |
-
"step": 4880
|
| 3322 |
-
},
|
| 3323 |
-
{
|
| 3324 |
-
"epoch": 7.98,
|
| 3325 |
-
"learning_rate": 0.0002202486678507993,
|
| 3326 |
-
"loss": 0.3956,
|
| 3327 |
-
"step": 4890
|
| 3328 |
-
},
|
| 3329 |
-
{
|
| 3330 |
-
"epoch": 7.99,
|
| 3331 |
-
"learning_rate": 0.00021847246891651864,
|
| 3332 |
-
"loss": 0.4197,
|
| 3333 |
-
"step": 4900
|
| 3334 |
-
},
|
| 3335 |
-
{
|
| 3336 |
-
"epoch": 7.99,
|
| 3337 |
-
"eval_loss": 0.42817065119743347,
|
| 3338 |
-
"eval_runtime": 5.657,
|
| 3339 |
-
"eval_samples_per_second": 206.999,
|
| 3340 |
-
"eval_steps_per_second": 13.081,
|
| 3341 |
-
"step": 4900
|
| 3342 |
-
},
|
| 3343 |
-
{
|
| 3344 |
-
"epoch": 8.01,
|
| 3345 |
-
"learning_rate": 0.000216696269982238,
|
| 3346 |
-
"loss": 0.3854,
|
| 3347 |
-
"step": 4910
|
| 3348 |
-
},
|
| 3349 |
-
{
|
| 3350 |
-
"epoch": 8.03,
|
| 3351 |
-
"learning_rate": 0.00021492007104795738,
|
| 3352 |
-
"loss": 0.4118,
|
| 3353 |
-
"step": 4920
|
| 3354 |
-
},
|
| 3355 |
-
{
|
| 3356 |
-
"epoch": 8.04,
|
| 3357 |
-
"learning_rate": 0.00021314387211367671,
|
| 3358 |
-
"loss": 0.4117,
|
| 3359 |
-
"step": 4930
|
| 3360 |
-
},
|
| 3361 |
-
{
|
| 3362 |
-
"epoch": 8.06,
|
| 3363 |
-
"learning_rate": 0.00021136767317939608,
|
| 3364 |
-
"loss": 0.4063,
|
| 3365 |
-
"step": 4940
|
| 3366 |
-
},
|
| 3367 |
-
{
|
| 3368 |
-
"epoch": 8.08,
|
| 3369 |
-
"learning_rate": 0.00020959147424511545,
|
| 3370 |
-
"loss": 0.4431,
|
| 3371 |
-
"step": 4950
|
| 3372 |
-
},
|
| 3373 |
-
{
|
| 3374 |
-
"epoch": 8.09,
|
| 3375 |
-
"learning_rate": 0.00020781527531083484,
|
| 3376 |
-
"loss": 0.4293,
|
| 3377 |
-
"step": 4960
|
| 3378 |
-
},
|
| 3379 |
-
{
|
| 3380 |
-
"epoch": 8.11,
|
| 3381 |
-
"learning_rate": 0.00020603907637655418,
|
| 3382 |
-
"loss": 0.4156,
|
| 3383 |
-
"step": 4970
|
| 3384 |
-
},
|
| 3385 |
-
{
|
| 3386 |
-
"epoch": 8.12,
|
| 3387 |
-
"learning_rate": 0.00020426287744227355,
|
| 3388 |
-
"loss": 0.4639,
|
| 3389 |
-
"step": 4980
|
| 3390 |
-
},
|
| 3391 |
-
{
|
| 3392 |
-
"epoch": 8.14,
|
| 3393 |
-
"learning_rate": 0.0002024866785079929,
|
| 3394 |
-
"loss": 0.4234,
|
| 3395 |
-
"step": 4990
|
| 3396 |
-
},
|
| 3397 |
-
{
|
| 3398 |
-
"epoch": 8.16,
|
| 3399 |
-
"learning_rate": 0.00020071047957371228,
|
| 3400 |
-
"loss": 0.3902,
|
| 3401 |
-
"step": 5000
|
| 3402 |
-
},
|
| 3403 |
-
{
|
| 3404 |
-
"epoch": 8.16,
|
| 3405 |
-
"eval_loss": 0.42921850085258484,
|
| 3406 |
-
"eval_runtime": 5.61,
|
| 3407 |
-
"eval_samples_per_second": 208.734,
|
| 3408 |
-
"eval_steps_per_second": 13.191,
|
| 3409 |
-
"step": 5000
|
| 3410 |
-
},
|
| 3411 |
-
{
|
| 3412 |
-
"epoch": 8.17,
|
| 3413 |
-
"learning_rate": 0.00019893428063943162,
|
| 3414 |
-
"loss": 0.4156,
|
| 3415 |
-
"step": 5010
|
| 3416 |
-
},
|
| 3417 |
-
{
|
| 3418 |
-
"epoch": 8.19,
|
| 3419 |
-
"learning_rate": 0.00019715808170515098,
|
| 3420 |
-
"loss": 0.4171,
|
| 3421 |
-
"step": 5020
|
| 3422 |
-
},
|
| 3423 |
-
{
|
| 3424 |
-
"epoch": 8.21,
|
| 3425 |
-
"learning_rate": 0.00019538188277087035,
|
| 3426 |
-
"loss": 0.4158,
|
| 3427 |
-
"step": 5030
|
| 3428 |
-
},
|
| 3429 |
-
{
|
| 3430 |
-
"epoch": 8.22,
|
| 3431 |
-
"learning_rate": 0.0001936056838365897,
|
| 3432 |
-
"loss": 0.4195,
|
| 3433 |
-
"step": 5040
|
| 3434 |
-
},
|
| 3435 |
-
{
|
| 3436 |
-
"epoch": 8.24,
|
| 3437 |
-
"learning_rate": 0.00019182948490230906,
|
| 3438 |
-
"loss": 0.4205,
|
| 3439 |
-
"step": 5050
|
| 3440 |
-
},
|
| 3441 |
-
{
|
| 3442 |
-
"epoch": 8.25,
|
| 3443 |
-
"learning_rate": 0.00019005328596802842,
|
| 3444 |
-
"loss": 0.4069,
|
| 3445 |
-
"step": 5060
|
| 3446 |
-
},
|
| 3447 |
-
{
|
| 3448 |
-
"epoch": 8.27,
|
| 3449 |
-
"learning_rate": 0.0001882770870337478,
|
| 3450 |
-
"loss": 0.4316,
|
| 3451 |
-
"step": 5070
|
| 3452 |
-
},
|
| 3453 |
-
{
|
| 3454 |
-
"epoch": 8.29,
|
| 3455 |
-
"learning_rate": 0.00018650088809946713,
|
| 3456 |
-
"loss": 0.4286,
|
| 3457 |
-
"step": 5080
|
| 3458 |
-
},
|
| 3459 |
-
{
|
| 3460 |
-
"epoch": 8.3,
|
| 3461 |
-
"learning_rate": 0.0001847246891651865,
|
| 3462 |
-
"loss": 0.4452,
|
| 3463 |
-
"step": 5090
|
| 3464 |
-
},
|
| 3465 |
-
{
|
| 3466 |
-
"epoch": 8.32,
|
| 3467 |
-
"learning_rate": 0.00018294849023090586,
|
| 3468 |
-
"loss": 0.4166,
|
| 3469 |
-
"step": 5100
|
| 3470 |
-
},
|
| 3471 |
-
{
|
| 3472 |
-
"epoch": 8.32,
|
| 3473 |
-
"eval_loss": 0.43339869379997253,
|
| 3474 |
-
"eval_runtime": 5.806,
|
| 3475 |
-
"eval_samples_per_second": 201.687,
|
| 3476 |
-
"eval_steps_per_second": 12.745,
|
| 3477 |
-
"step": 5100
|
| 3478 |
-
},
|
| 3479 |
-
{
|
| 3480 |
-
"epoch": 8.34,
|
| 3481 |
-
"learning_rate": 0.0001811722912966252,
|
| 3482 |
-
"loss": 0.448,
|
| 3483 |
-
"step": 5110
|
| 3484 |
-
},
|
| 3485 |
-
{
|
| 3486 |
-
"epoch": 8.35,
|
| 3487 |
-
"learning_rate": 0.0001793960923623446,
|
| 3488 |
-
"loss": 0.4058,
|
| 3489 |
-
"step": 5120
|
| 3490 |
-
},
|
| 3491 |
-
{
|
| 3492 |
-
"epoch": 8.37,
|
| 3493 |
-
"learning_rate": 0.00017761989342806396,
|
| 3494 |
-
"loss": 0.3919,
|
| 3495 |
-
"step": 5130
|
| 3496 |
-
},
|
| 3497 |
-
{
|
| 3498 |
-
"epoch": 8.38,
|
| 3499 |
-
"learning_rate": 0.00017584369449378333,
|
| 3500 |
-
"loss": 0.4525,
|
| 3501 |
-
"step": 5140
|
| 3502 |
-
},
|
| 3503 |
-
{
|
| 3504 |
-
"epoch": 8.4,
|
| 3505 |
-
"learning_rate": 0.00017406749555950267,
|
| 3506 |
-
"loss": 0.4269,
|
| 3507 |
-
"step": 5150
|
| 3508 |
-
},
|
| 3509 |
-
{
|
| 3510 |
-
"epoch": 8.42,
|
| 3511 |
-
"learning_rate": 0.00017229129662522203,
|
| 3512 |
-
"loss": 0.4355,
|
| 3513 |
-
"step": 5160
|
| 3514 |
-
},
|
| 3515 |
-
{
|
| 3516 |
-
"epoch": 8.43,
|
| 3517 |
-
"learning_rate": 0.0001705150976909414,
|
| 3518 |
-
"loss": 0.4458,
|
| 3519 |
-
"step": 5170
|
| 3520 |
-
},
|
| 3521 |
-
{
|
| 3522 |
-
"epoch": 8.45,
|
| 3523 |
-
"learning_rate": 0.00016873889875666074,
|
| 3524 |
-
"loss": 0.3966,
|
| 3525 |
-
"step": 5180
|
| 3526 |
-
},
|
| 3527 |
-
{
|
| 3528 |
-
"epoch": 8.47,
|
| 3529 |
-
"learning_rate": 0.0001669626998223801,
|
| 3530 |
-
"loss": 0.4058,
|
| 3531 |
-
"step": 5190
|
| 3532 |
-
},
|
| 3533 |
-
{
|
| 3534 |
-
"epoch": 8.48,
|
| 3535 |
-
"learning_rate": 0.00016518650088809947,
|
| 3536 |
-
"loss": 0.4199,
|
| 3537 |
-
"step": 5200
|
| 3538 |
-
},
|
| 3539 |
-
{
|
| 3540 |
-
"epoch": 8.48,
|
| 3541 |
-
"eval_loss": 0.4262863099575043,
|
| 3542 |
-
"eval_runtime": 5.725,
|
| 3543 |
-
"eval_samples_per_second": 204.541,
|
| 3544 |
-
"eval_steps_per_second": 12.926,
|
| 3545 |
-
"step": 5200
|
| 3546 |
-
},
|
| 3547 |
-
{
|
| 3548 |
-
"epoch": 8.5,
|
| 3549 |
-
"learning_rate": 0.00016341030195381884,
|
| 3550 |
-
"loss": 0.4185,
|
| 3551 |
-
"step": 5210
|
| 3552 |
-
},
|
| 3553 |
-
{
|
| 3554 |
-
"epoch": 8.52,
|
| 3555 |
-
"learning_rate": 0.00016163410301953818,
|
| 3556 |
-
"loss": 0.4145,
|
| 3557 |
-
"step": 5220
|
| 3558 |
-
},
|
| 3559 |
-
{
|
| 3560 |
-
"epoch": 8.53,
|
| 3561 |
-
"learning_rate": 0.00015985790408525754,
|
| 3562 |
-
"loss": 0.4178,
|
| 3563 |
-
"step": 5230
|
| 3564 |
-
},
|
| 3565 |
-
{
|
| 3566 |
-
"epoch": 8.55,
|
| 3567 |
-
"learning_rate": 0.0001580817051509769,
|
| 3568 |
-
"loss": 0.4285,
|
| 3569 |
-
"step": 5240
|
| 3570 |
-
},
|
| 3571 |
-
{
|
| 3572 |
-
"epoch": 8.56,
|
| 3573 |
-
"learning_rate": 0.00015630550621669628,
|
| 3574 |
-
"loss": 0.4062,
|
| 3575 |
-
"step": 5250
|
| 3576 |
-
},
|
| 3577 |
-
{
|
| 3578 |
-
"epoch": 8.58,
|
| 3579 |
-
"learning_rate": 0.00015452930728241561,
|
| 3580 |
-
"loss": 0.4308,
|
| 3581 |
-
"step": 5260
|
| 3582 |
-
},
|
| 3583 |
-
{
|
| 3584 |
-
"epoch": 8.6,
|
| 3585 |
-
"learning_rate": 0.000152753108348135,
|
| 3586 |
-
"loss": 0.4488,
|
| 3587 |
-
"step": 5270
|
| 3588 |
-
},
|
| 3589 |
-
{
|
| 3590 |
-
"epoch": 8.61,
|
| 3591 |
-
"learning_rate": 0.00015097690941385437,
|
| 3592 |
-
"loss": 0.4087,
|
| 3593 |
-
"step": 5280
|
| 3594 |
-
},
|
| 3595 |
-
{
|
| 3596 |
-
"epoch": 8.63,
|
| 3597 |
-
"learning_rate": 0.00014920071047957371,
|
| 3598 |
-
"loss": 0.421,
|
| 3599 |
-
"step": 5290
|
| 3600 |
-
},
|
| 3601 |
-
{
|
| 3602 |
-
"epoch": 8.65,
|
| 3603 |
-
"learning_rate": 0.00014742451154529308,
|
| 3604 |
-
"loss": 0.4415,
|
| 3605 |
-
"step": 5300
|
| 3606 |
-
},
|
| 3607 |
-
{
|
| 3608 |
-
"epoch": 8.65,
|
| 3609 |
-
"eval_loss": 0.4254954755306244,
|
| 3610 |
-
"eval_runtime": 5.593,
|
| 3611 |
-
"eval_samples_per_second": 209.368,
|
| 3612 |
-
"eval_steps_per_second": 13.231,
|
| 3613 |
-
"step": 5300
|
| 3614 |
-
},
|
| 3615 |
-
{
|
| 3616 |
-
"epoch": 8.66,
|
| 3617 |
-
"learning_rate": 0.00014564831261101245,
|
| 3618 |
-
"loss": 0.4358,
|
| 3619 |
-
"step": 5310
|
| 3620 |
-
},
|
| 3621 |
-
{
|
| 3622 |
-
"epoch": 8.68,
|
| 3623 |
-
"learning_rate": 0.0001438721136767318,
|
| 3624 |
-
"loss": 0.404,
|
| 3625 |
-
"step": 5320
|
| 3626 |
-
},
|
| 3627 |
-
{
|
| 3628 |
-
"epoch": 8.69,
|
| 3629 |
-
"learning_rate": 0.00014209591474245115,
|
| 3630 |
-
"loss": 0.4172,
|
| 3631 |
-
"step": 5330
|
| 3632 |
-
},
|
| 3633 |
-
{
|
| 3634 |
-
"epoch": 8.71,
|
| 3635 |
-
"learning_rate": 0.00014031971580817052,
|
| 3636 |
-
"loss": 0.3918,
|
| 3637 |
-
"step": 5340
|
| 3638 |
-
},
|
| 3639 |
-
{
|
| 3640 |
-
"epoch": 8.73,
|
| 3641 |
-
"learning_rate": 0.00013854351687388988,
|
| 3642 |
-
"loss": 0.4313,
|
| 3643 |
-
"step": 5350
|
| 3644 |
-
},
|
| 3645 |
-
{
|
| 3646 |
-
"epoch": 8.74,
|
| 3647 |
-
"learning_rate": 0.00013676731793960922,
|
| 3648 |
-
"loss": 0.4326,
|
| 3649 |
-
"step": 5360
|
| 3650 |
-
},
|
| 3651 |
-
{
|
| 3652 |
-
"epoch": 8.76,
|
| 3653 |
-
"learning_rate": 0.0001349911190053286,
|
| 3654 |
-
"loss": 0.4369,
|
| 3655 |
-
"step": 5370
|
| 3656 |
-
},
|
| 3657 |
-
{
|
| 3658 |
-
"epoch": 8.78,
|
| 3659 |
-
"learning_rate": 0.00013321492007104796,
|
| 3660 |
-
"loss": 0.4099,
|
| 3661 |
-
"step": 5380
|
| 3662 |
-
},
|
| 3663 |
-
{
|
| 3664 |
-
"epoch": 8.79,
|
| 3665 |
-
"learning_rate": 0.00013143872113676732,
|
| 3666 |
-
"loss": 0.4165,
|
| 3667 |
-
"step": 5390
|
| 3668 |
-
},
|
| 3669 |
-
{
|
| 3670 |
-
"epoch": 8.81,
|
| 3671 |
-
"learning_rate": 0.00012966252220248666,
|
| 3672 |
-
"loss": 0.4014,
|
| 3673 |
-
"step": 5400
|
| 3674 |
-
},
|
| 3675 |
-
{
|
| 3676 |
-
"epoch": 8.81,
|
| 3677 |
-
"eval_loss": 0.4244977533817291,
|
| 3678 |
-
"eval_runtime": 5.568,
|
| 3679 |
-
"eval_samples_per_second": 210.308,
|
| 3680 |
-
"eval_steps_per_second": 13.29,
|
| 3681 |
-
"step": 5400
|
| 3682 |
-
},
|
| 3683 |
-
{
|
| 3684 |
-
"epoch": 8.83,
|
| 3685 |
-
"learning_rate": 0.00012788632326820603,
|
| 3686 |
-
"loss": 0.4111,
|
| 3687 |
-
"step": 5410
|
| 3688 |
-
},
|
| 3689 |
-
{
|
| 3690 |
-
"epoch": 8.84,
|
| 3691 |
-
"learning_rate": 0.0001261101243339254,
|
| 3692 |
-
"loss": 0.4323,
|
| 3693 |
-
"step": 5420
|
| 3694 |
-
},
|
| 3695 |
-
{
|
| 3696 |
-
"epoch": 8.86,
|
| 3697 |
-
"learning_rate": 0.00012433392539964476,
|
| 3698 |
-
"loss": 0.4342,
|
| 3699 |
-
"step": 5430
|
| 3700 |
-
},
|
| 3701 |
-
{
|
| 3702 |
-
"epoch": 8.87,
|
| 3703 |
-
"learning_rate": 0.00012255772646536413,
|
| 3704 |
-
"loss": 0.4111,
|
| 3705 |
-
"step": 5440
|
| 3706 |
-
},
|
| 3707 |
-
{
|
| 3708 |
-
"epoch": 8.89,
|
| 3709 |
-
"learning_rate": 0.00012078152753108348,
|
| 3710 |
-
"loss": 0.4464,
|
| 3711 |
-
"step": 5450
|
| 3712 |
-
},
|
| 3713 |
-
{
|
| 3714 |
-
"epoch": 8.91,
|
| 3715 |
-
"learning_rate": 0.00011900532859680283,
|
| 3716 |
-
"loss": 0.4244,
|
| 3717 |
-
"step": 5460
|
| 3718 |
-
},
|
| 3719 |
-
{
|
| 3720 |
-
"epoch": 8.92,
|
| 3721 |
-
"learning_rate": 0.00011722912966252221,
|
| 3722 |
-
"loss": 0.4503,
|
| 3723 |
-
"step": 5470
|
| 3724 |
-
},
|
| 3725 |
-
{
|
| 3726 |
-
"epoch": 8.94,
|
| 3727 |
-
"learning_rate": 0.00011545293072824157,
|
| 3728 |
-
"loss": 0.4276,
|
| 3729 |
-
"step": 5480
|
| 3730 |
-
},
|
| 3731 |
-
{
|
| 3732 |
-
"epoch": 8.96,
|
| 3733 |
-
"learning_rate": 0.00011367673179396093,
|
| 3734 |
-
"loss": 0.4119,
|
| 3735 |
-
"step": 5490
|
| 3736 |
-
},
|
| 3737 |
-
{
|
| 3738 |
-
"epoch": 8.97,
|
| 3739 |
-
"learning_rate": 0.00011190053285968029,
|
| 3740 |
-
"loss": 0.4158,
|
| 3741 |
-
"step": 5500
|
| 3742 |
-
},
|
| 3743 |
-
{
|
| 3744 |
-
"epoch": 8.97,
|
| 3745 |
-
"eval_loss": 0.4268430769443512,
|
| 3746 |
-
"eval_runtime": 5.711,
|
| 3747 |
-
"eval_samples_per_second": 205.042,
|
| 3748 |
-
"eval_steps_per_second": 12.957,
|
| 3749 |
-
"step": 5500
|
| 3750 |
-
},
|
| 3751 |
-
{
|
| 3752 |
-
"epoch": 8.99,
|
| 3753 |
-
"learning_rate": 0.00011012433392539965,
|
| 3754 |
-
"loss": 0.4506,
|
| 3755 |
-
"step": 5510
|
| 3756 |
-
},
|
| 3757 |
-
{
|
| 3758 |
-
"epoch": 9.0,
|
| 3759 |
-
"learning_rate": 0.000108348134991119,
|
| 3760 |
-
"loss": 0.3832,
|
| 3761 |
-
"step": 5520
|
| 3762 |
-
},
|
| 3763 |
-
{
|
| 3764 |
-
"epoch": 9.02,
|
| 3765 |
-
"learning_rate": 0.00010657193605683836,
|
| 3766 |
-
"loss": 0.4018,
|
| 3767 |
-
"step": 5530
|
| 3768 |
-
},
|
| 3769 |
-
{
|
| 3770 |
-
"epoch": 9.04,
|
| 3771 |
-
"learning_rate": 0.00010479573712255772,
|
| 3772 |
-
"loss": 0.413,
|
| 3773 |
-
"step": 5540
|
| 3774 |
-
},
|
| 3775 |
-
{
|
| 3776 |
-
"epoch": 9.05,
|
| 3777 |
-
"learning_rate": 0.00010301953818827709,
|
| 3778 |
-
"loss": 0.4317,
|
| 3779 |
-
"step": 5550
|
| 3780 |
-
},
|
| 3781 |
-
{
|
| 3782 |
-
"epoch": 9.07,
|
| 3783 |
-
"learning_rate": 0.00010124333925399646,
|
| 3784 |
-
"loss": 0.4117,
|
| 3785 |
-
"step": 5560
|
| 3786 |
-
},
|
| 3787 |
-
{
|
| 3788 |
-
"epoch": 9.09,
|
| 3789 |
-
"learning_rate": 9.946714031971581e-05,
|
| 3790 |
-
"loss": 0.4068,
|
| 3791 |
-
"step": 5570
|
| 3792 |
-
},
|
| 3793 |
-
{
|
| 3794 |
-
"epoch": 9.1,
|
| 3795 |
-
"learning_rate": 9.769094138543518e-05,
|
| 3796 |
-
"loss": 0.3956,
|
| 3797 |
-
"step": 5580
|
| 3798 |
-
},
|
| 3799 |
-
{
|
| 3800 |
-
"epoch": 9.12,
|
| 3801 |
-
"learning_rate": 9.591474245115453e-05,
|
| 3802 |
-
"loss": 0.3948,
|
| 3803 |
-
"step": 5590
|
| 3804 |
-
},
|
| 3805 |
-
{
|
| 3806 |
-
"epoch": 9.14,
|
| 3807 |
-
"learning_rate": 9.41385435168739e-05,
|
| 3808 |
-
"loss": 0.3842,
|
| 3809 |
-
"step": 5600
|
| 3810 |
-
},
|
| 3811 |
-
{
|
| 3812 |
-
"epoch": 9.14,
|
| 3813 |
-
"eval_loss": 0.42232006788253784,
|
| 3814 |
-
"eval_runtime": 5.687,
|
| 3815 |
-
"eval_samples_per_second": 205.907,
|
| 3816 |
-
"eval_steps_per_second": 13.012,
|
| 3817 |
-
"step": 5600
|
| 3818 |
-
},
|
| 3819 |
-
{
|
| 3820 |
-
"epoch": 9.15,
|
| 3821 |
-
"learning_rate": 9.236234458259325e-05,
|
| 3822 |
-
"loss": 0.4099,
|
| 3823 |
-
"step": 5610
|
| 3824 |
-
},
|
| 3825 |
-
{
|
| 3826 |
-
"epoch": 9.17,
|
| 3827 |
-
"learning_rate": 9.05861456483126e-05,
|
| 3828 |
-
"loss": 0.413,
|
| 3829 |
-
"step": 5620
|
| 3830 |
-
},
|
| 3831 |
-
{
|
| 3832 |
-
"epoch": 9.18,
|
| 3833 |
-
"learning_rate": 8.880994671403198e-05,
|
| 3834 |
-
"loss": 0.4015,
|
| 3835 |
-
"step": 5630
|
| 3836 |
-
},
|
| 3837 |
-
{
|
| 3838 |
-
"epoch": 9.2,
|
| 3839 |
-
"learning_rate": 8.703374777975133e-05,
|
| 3840 |
-
"loss": 0.4304,
|
| 3841 |
-
"step": 5640
|
| 3842 |
-
},
|
| 3843 |
-
{
|
| 3844 |
-
"epoch": 9.22,
|
| 3845 |
-
"learning_rate": 8.52575488454707e-05,
|
| 3846 |
-
"loss": 0.4239,
|
| 3847 |
-
"step": 5650
|
| 3848 |
-
},
|
| 3849 |
-
{
|
| 3850 |
-
"epoch": 9.23,
|
| 3851 |
-
"learning_rate": 8.348134991119005e-05,
|
| 3852 |
-
"loss": 0.3997,
|
| 3853 |
-
"step": 5660
|
| 3854 |
-
},
|
| 3855 |
-
{
|
| 3856 |
-
"epoch": 9.25,
|
| 3857 |
-
"learning_rate": 8.170515097690942e-05,
|
| 3858 |
-
"loss": 0.4935,
|
| 3859 |
-
"step": 5670
|
| 3860 |
-
},
|
| 3861 |
-
{
|
| 3862 |
-
"epoch": 9.27,
|
| 3863 |
-
"learning_rate": 7.992895204262877e-05,
|
| 3864 |
-
"loss": 0.4035,
|
| 3865 |
-
"step": 5680
|
| 3866 |
-
},
|
| 3867 |
-
{
|
| 3868 |
-
"epoch": 9.28,
|
| 3869 |
-
"learning_rate": 7.815275310834814e-05,
|
| 3870 |
-
"loss": 0.435,
|
| 3871 |
-
"step": 5690
|
| 3872 |
-
},
|
| 3873 |
-
{
|
| 3874 |
-
"epoch": 9.3,
|
| 3875 |
-
"learning_rate": 7.63765541740675e-05,
|
| 3876 |
-
"loss": 0.4267,
|
| 3877 |
-
"step": 5700
|
| 3878 |
-
},
|
| 3879 |
-
{
|
| 3880 |
-
"epoch": 9.3,
|
| 3881 |
-
"eval_loss": 0.4202769696712494,
|
| 3882 |
-
"eval_runtime": 5.77,
|
| 3883 |
-
"eval_samples_per_second": 202.946,
|
| 3884 |
-
"eval_steps_per_second": 12.825,
|
| 3885 |
-
"step": 5700
|
| 3886 |
-
},
|
| 3887 |
-
{
|
| 3888 |
-
"epoch": 9.31,
|
| 3889 |
-
"learning_rate": 7.460035523978686e-05,
|
| 3890 |
-
"loss": 0.4154,
|
| 3891 |
-
"step": 5710
|
| 3892 |
-
},
|
| 3893 |
-
{
|
| 3894 |
-
"epoch": 9.33,
|
| 3895 |
-
"learning_rate": 7.282415630550622e-05,
|
| 3896 |
-
"loss": 0.4107,
|
| 3897 |
-
"step": 5720
|
| 3898 |
-
},
|
| 3899 |
-
{
|
| 3900 |
-
"epoch": 9.35,
|
| 3901 |
-
"learning_rate": 7.104795737122558e-05,
|
| 3902 |
-
"loss": 0.424,
|
| 3903 |
-
"step": 5730
|
| 3904 |
-
},
|
| 3905 |
-
{
|
| 3906 |
-
"epoch": 9.36,
|
| 3907 |
-
"learning_rate": 6.927175843694494e-05,
|
| 3908 |
-
"loss": 0.3896,
|
| 3909 |
-
"step": 5740
|
| 3910 |
-
},
|
| 3911 |
-
{
|
| 3912 |
-
"epoch": 9.38,
|
| 3913 |
-
"learning_rate": 6.74955595026643e-05,
|
| 3914 |
-
"loss": 0.4094,
|
| 3915 |
-
"step": 5750
|
| 3916 |
-
},
|
| 3917 |
-
{
|
| 3918 |
-
"epoch": 9.4,
|
| 3919 |
-
"learning_rate": 6.571936056838366e-05,
|
| 3920 |
-
"loss": 0.4134,
|
| 3921 |
-
"step": 5760
|
| 3922 |
-
},
|
| 3923 |
-
{
|
| 3924 |
-
"epoch": 9.41,
|
| 3925 |
-
"learning_rate": 6.394316163410301e-05,
|
| 3926 |
-
"loss": 0.4232,
|
| 3927 |
-
"step": 5770
|
| 3928 |
-
},
|
| 3929 |
-
{
|
| 3930 |
-
"epoch": 9.43,
|
| 3931 |
-
"learning_rate": 6.216696269982238e-05,
|
| 3932 |
-
"loss": 0.4005,
|
| 3933 |
-
"step": 5780
|
| 3934 |
-
},
|
| 3935 |
-
{
|
| 3936 |
-
"epoch": 9.45,
|
| 3937 |
-
"learning_rate": 6.039076376554174e-05,
|
| 3938 |
-
"loss": 0.4138,
|
| 3939 |
-
"step": 5790
|
| 3940 |
-
},
|
| 3941 |
-
{
|
| 3942 |
-
"epoch": 9.46,
|
| 3943 |
-
"learning_rate": 5.861456483126111e-05,
|
| 3944 |
-
"loss": 0.423,
|
| 3945 |
-
"step": 5800
|
| 3946 |
-
},
|
| 3947 |
-
{
|
| 3948 |
-
"epoch": 9.46,
|
| 3949 |
-
"eval_loss": 0.4260061979293823,
|
| 3950 |
-
"eval_runtime": 6.015,
|
| 3951 |
-
"eval_samples_per_second": 194.679,
|
| 3952 |
-
"eval_steps_per_second": 12.303,
|
| 3953 |
-
"step": 5800
|
| 3954 |
-
},
|
| 3955 |
-
{
|
| 3956 |
-
"epoch": 9.48,
|
| 3957 |
-
"learning_rate": 5.6838365896980466e-05,
|
| 3958 |
-
"loss": 0.3959,
|
| 3959 |
-
"step": 5810
|
| 3960 |
-
},
|
| 3961 |
-
{
|
| 3962 |
-
"epoch": 9.49,
|
| 3963 |
-
"learning_rate": 5.5062166962699826e-05,
|
| 3964 |
-
"loss": 0.4072,
|
| 3965 |
-
"step": 5820
|
| 3966 |
-
},
|
| 3967 |
-
{
|
| 3968 |
-
"epoch": 9.51,
|
| 3969 |
-
"learning_rate": 5.328596802841918e-05,
|
| 3970 |
-
"loss": 0.429,
|
| 3971 |
-
"step": 5830
|
| 3972 |
-
},
|
| 3973 |
-
{
|
| 3974 |
-
"epoch": 9.53,
|
| 3975 |
-
"learning_rate": 5.1509769094138545e-05,
|
| 3976 |
-
"loss": 0.4161,
|
| 3977 |
-
"step": 5840
|
| 3978 |
-
},
|
| 3979 |
-
{
|
| 3980 |
-
"epoch": 9.54,
|
| 3981 |
-
"learning_rate": 4.9733570159857905e-05,
|
| 3982 |
-
"loss": 0.4319,
|
| 3983 |
-
"step": 5850
|
| 3984 |
-
},
|
| 3985 |
-
{
|
| 3986 |
-
"epoch": 9.56,
|
| 3987 |
-
"learning_rate": 4.7957371225577264e-05,
|
| 3988 |
-
"loss": 0.4195,
|
| 3989 |
-
"step": 5860
|
| 3990 |
-
},
|
| 3991 |
-
{
|
| 3992 |
-
"epoch": 9.58,
|
| 3993 |
-
"learning_rate": 4.6181172291296624e-05,
|
| 3994 |
-
"loss": 0.409,
|
| 3995 |
-
"step": 5870
|
| 3996 |
-
},
|
| 3997 |
-
{
|
| 3998 |
-
"epoch": 9.59,
|
| 3999 |
-
"learning_rate": 4.440497335701599e-05,
|
| 4000 |
-
"loss": 0.4327,
|
| 4001 |
-
"step": 5880
|
| 4002 |
-
},
|
| 4003 |
-
{
|
| 4004 |
-
"epoch": 9.61,
|
| 4005 |
-
"learning_rate": 4.262877442273535e-05,
|
| 4006 |
-
"loss": 0.4337,
|
| 4007 |
-
"step": 5890
|
| 4008 |
-
},
|
| 4009 |
-
{
|
| 4010 |
-
"epoch": 9.62,
|
| 4011 |
-
"learning_rate": 4.085257548845471e-05,
|
| 4012 |
-
"loss": 0.4506,
|
| 4013 |
-
"step": 5900
|
| 4014 |
-
},
|
| 4015 |
-
{
|
| 4016 |
-
"epoch": 9.62,
|
| 4017 |
-
"eval_loss": 0.4249822795391083,
|
| 4018 |
-
"eval_runtime": 5.633,
|
| 4019 |
-
"eval_samples_per_second": 207.881,
|
| 4020 |
-
"eval_steps_per_second": 13.137,
|
| 4021 |
-
"step": 5900
|
| 4022 |
-
},
|
| 4023 |
-
{
|
| 4024 |
-
"epoch": 9.64,
|
| 4025 |
-
"learning_rate": 3.907637655417407e-05,
|
| 4026 |
-
"loss": 0.4067,
|
| 4027 |
-
"step": 5910
|
| 4028 |
-
},
|
| 4029 |
-
{
|
| 4030 |
-
"epoch": 9.66,
|
| 4031 |
-
"learning_rate": 3.730017761989343e-05,
|
| 4032 |
-
"loss": 0.3946,
|
| 4033 |
-
"step": 5920
|
| 4034 |
-
},
|
| 4035 |
-
{
|
| 4036 |
-
"epoch": 9.67,
|
| 4037 |
-
"learning_rate": 3.552397868561279e-05,
|
| 4038 |
-
"loss": 0.4301,
|
| 4039 |
-
"step": 5930
|
| 4040 |
-
},
|
| 4041 |
-
{
|
| 4042 |
-
"epoch": 9.69,
|
| 4043 |
-
"learning_rate": 3.374777975133215e-05,
|
| 4044 |
-
"loss": 0.4204,
|
| 4045 |
-
"step": 5940
|
| 4046 |
-
},
|
| 4047 |
-
{
|
| 4048 |
-
"epoch": 9.71,
|
| 4049 |
-
"learning_rate": 3.197158081705151e-05,
|
| 4050 |
-
"loss": 0.4298,
|
| 4051 |
-
"step": 5950
|
| 4052 |
-
},
|
| 4053 |
-
{
|
| 4054 |
-
"epoch": 9.72,
|
| 4055 |
-
"learning_rate": 3.019538188277087e-05,
|
| 4056 |
-
"loss": 0.4056,
|
| 4057 |
-
"step": 5960
|
| 4058 |
-
},
|
| 4059 |
-
{
|
| 4060 |
-
"epoch": 9.74,
|
| 4061 |
-
"learning_rate": 2.8419182948490233e-05,
|
| 4062 |
-
"loss": 0.4377,
|
| 4063 |
-
"step": 5970
|
| 4064 |
-
},
|
| 4065 |
-
{
|
| 4066 |
-
"epoch": 9.76,
|
| 4067 |
-
"learning_rate": 2.664298401420959e-05,
|
| 4068 |
-
"loss": 0.417,
|
| 4069 |
-
"step": 5980
|
| 4070 |
-
},
|
| 4071 |
-
{
|
| 4072 |
-
"epoch": 9.77,
|
| 4073 |
-
"learning_rate": 2.4866785079928952e-05,
|
| 4074 |
-
"loss": 0.4153,
|
| 4075 |
-
"step": 5990
|
| 4076 |
-
},
|
| 4077 |
-
{
|
| 4078 |
-
"epoch": 9.79,
|
| 4079 |
-
"learning_rate": 2.3090586145648312e-05,
|
| 4080 |
-
"loss": 0.4087,
|
| 4081 |
-
"step": 6000
|
| 4082 |
-
},
|
| 4083 |
-
{
|
| 4084 |
-
"epoch": 9.79,
|
| 4085 |
-
"eval_loss": 0.4207456707954407,
|
| 4086 |
-
"eval_runtime": 5.625,
|
| 4087 |
-
"eval_samples_per_second": 208.177,
|
| 4088 |
-
"eval_steps_per_second": 13.156,
|
| 4089 |
-
"step": 6000
|
| 4090 |
}
|
| 4091 |
],
|
| 4092 |
"logging_steps": 10,
|
| 4093 |
"max_steps": 6130,
|
| 4094 |
"num_train_epochs": 10,
|
| 4095 |
"save_steps": 500,
|
| 4096 |
-
"total_flos":
|
| 4097 |
"trial_name": null,
|
| 4098 |
"trial_params": null
|
| 4099 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.5469470024108887,
|
| 3 |
+
"best_model_checkpoint": "bart_lora_outputs\\checkpoint-500",
|
| 4 |
+
"epoch": 0.8156606851549756,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 11 |
{
|
| 12 |
"epoch": 0.02,
|
| 13 |
"learning_rate": 2e-05,
|
| 14 |
+
"loss": 2.9281,
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.03,
|
| 19 |
"learning_rate": 4e-05,
|
| 20 |
+
"loss": 2.8201,
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
"epoch": 0.05,
|
| 25 |
"learning_rate": 6e-05,
|
| 26 |
+
"loss": 2.579,
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.07,
|
| 31 |
"learning_rate": 8e-05,
|
| 32 |
+
"loss": 2.4427,
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.08,
|
| 37 |
"learning_rate": 0.0001,
|
| 38 |
+
"loss": 2.1681,
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.1,
|
| 43 |
"learning_rate": 0.00012,
|
| 44 |
+
"loss": 1.7104,
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.11,
|
| 49 |
"learning_rate": 0.00014000000000000001,
|
| 50 |
+
"loss": 1.4177,
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.13,
|
| 55 |
"learning_rate": 0.00016,
|
| 56 |
+
"loss": 1.2515,
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 0.15,
|
| 61 |
"learning_rate": 0.00017999999999999998,
|
| 62 |
+
"loss": 1.1238,
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.16,
|
| 67 |
"learning_rate": 0.0002,
|
| 68 |
+
"loss": 1.0489,
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.16,
|
| 73 |
+
"eval_loss": 0.8469381928443909,
|
| 74 |
+
"eval_runtime": 5.9601,
|
| 75 |
+
"eval_samples_per_second": 196.474,
|
| 76 |
+
"eval_steps_per_second": 24.664,
|
| 77 |
"step": 100
|
| 78 |
},
|
| 79 |
{
|
| 80 |
"epoch": 0.18,
|
| 81 |
"learning_rate": 0.00022,
|
| 82 |
+
"loss": 1.0079,
|
| 83 |
"step": 110
|
| 84 |
},
|
| 85 |
{
|
| 86 |
"epoch": 0.2,
|
| 87 |
"learning_rate": 0.00024,
|
| 88 |
+
"loss": 1.0077,
|
| 89 |
"step": 120
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 0.21,
|
| 93 |
"learning_rate": 0.00026000000000000003,
|
| 94 |
+
"loss": 0.9416,
|
| 95 |
"step": 130
|
| 96 |
},
|
| 97 |
{
|
| 98 |
"epoch": 0.23,
|
| 99 |
"learning_rate": 0.00028000000000000003,
|
| 100 |
+
"loss": 0.8882,
|
| 101 |
"step": 140
|
| 102 |
},
|
| 103 |
{
|
| 104 |
"epoch": 0.24,
|
| 105 |
"learning_rate": 0.0003,
|
| 106 |
+
"loss": 0.8595,
|
| 107 |
"step": 150
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.26,
|
| 111 |
"learning_rate": 0.00032,
|
| 112 |
+
"loss": 0.8853,
|
| 113 |
"step": 160
|
| 114 |
},
|
| 115 |
{
|
| 116 |
"epoch": 0.28,
|
| 117 |
"learning_rate": 0.00034,
|
| 118 |
+
"loss": 0.7678,
|
| 119 |
"step": 170
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 0.29,
|
| 123 |
"learning_rate": 0.00035999999999999997,
|
| 124 |
+
"loss": 0.8595,
|
| 125 |
"step": 180
|
| 126 |
},
|
| 127 |
{
|
| 128 |
"epoch": 0.31,
|
| 129 |
"learning_rate": 0.00038,
|
| 130 |
+
"loss": 0.8514,
|
| 131 |
"step": 190
|
| 132 |
},
|
| 133 |
{
|
| 134 |
"epoch": 0.33,
|
| 135 |
"learning_rate": 0.0004,
|
| 136 |
+
"loss": 0.8128,
|
| 137 |
"step": 200
|
| 138 |
},
|
| 139 |
{
|
| 140 |
"epoch": 0.33,
|
| 141 |
+
"eval_loss": 0.6850531697273254,
|
| 142 |
+
"eval_runtime": 6.0279,
|
| 143 |
+
"eval_samples_per_second": 194.264,
|
| 144 |
+
"eval_steps_per_second": 24.387,
|
| 145 |
"step": 200
|
| 146 |
},
|
| 147 |
{
|
| 148 |
"epoch": 0.34,
|
| 149 |
"learning_rate": 0.00042,
|
| 150 |
+
"loss": 0.7782,
|
| 151 |
"step": 210
|
| 152 |
},
|
| 153 |
{
|
| 154 |
"epoch": 0.36,
|
| 155 |
"learning_rate": 0.00044,
|
| 156 |
+
"loss": 0.8064,
|
| 157 |
"step": 220
|
| 158 |
},
|
| 159 |
{
|
| 160 |
"epoch": 0.38,
|
| 161 |
"learning_rate": 0.00046,
|
| 162 |
+
"loss": 0.7627,
|
| 163 |
"step": 230
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 0.39,
|
| 167 |
"learning_rate": 0.00048,
|
| 168 |
+
"loss": 0.7447,
|
| 169 |
"step": 240
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 0.41,
|
| 173 |
"learning_rate": 0.0005,
|
| 174 |
+
"loss": 0.7652,
|
| 175 |
"step": 250
|
| 176 |
},
|
| 177 |
{
|
| 178 |
"epoch": 0.42,
|
| 179 |
"learning_rate": 0.0005200000000000001,
|
| 180 |
+
"loss": 0.7568,
|
| 181 |
"step": 260
|
| 182 |
},
|
| 183 |
{
|
| 184 |
"epoch": 0.44,
|
| 185 |
"learning_rate": 0.00054,
|
| 186 |
+
"loss": 0.7291,
|
| 187 |
"step": 270
|
| 188 |
},
|
| 189 |
{
|
| 190 |
"epoch": 0.46,
|
| 191 |
"learning_rate": 0.0005600000000000001,
|
| 192 |
+
"loss": 0.7118,
|
| 193 |
"step": 280
|
| 194 |
},
|
| 195 |
{
|
| 196 |
"epoch": 0.47,
|
| 197 |
"learning_rate": 0.00058,
|
| 198 |
+
"loss": 0.7462,
|
| 199 |
"step": 290
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.49,
|
| 203 |
"learning_rate": 0.0006,
|
| 204 |
+
"loss": 0.6866,
|
| 205 |
"step": 300
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"epoch": 0.49,
|
| 209 |
+
"eval_loss": 0.629724383354187,
|
| 210 |
+
"eval_runtime": 6.0378,
|
| 211 |
+
"eval_samples_per_second": 193.943,
|
| 212 |
+
"eval_steps_per_second": 24.346,
|
| 213 |
"step": 300
|
| 214 |
},
|
| 215 |
{
|
| 216 |
"epoch": 0.51,
|
| 217 |
"learning_rate": 0.00062,
|
| 218 |
+
"loss": 0.6995,
|
| 219 |
"step": 310
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.52,
|
| 223 |
"learning_rate": 0.00064,
|
| 224 |
+
"loss": 0.724,
|
| 225 |
"step": 320
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 0.54,
|
| 229 |
"learning_rate": 0.00066,
|
| 230 |
+
"loss": 0.6698,
|
| 231 |
"step": 330
|
| 232 |
},
|
| 233 |
{
|
| 234 |
"epoch": 0.55,
|
| 235 |
"learning_rate": 0.00068,
|
| 236 |
+
"loss": 0.6516,
|
| 237 |
"step": 340
|
| 238 |
},
|
| 239 |
{
|
| 240 |
"epoch": 0.57,
|
| 241 |
"learning_rate": 0.0007,
|
| 242 |
+
"loss": 0.6657,
|
| 243 |
"step": 350
|
| 244 |
},
|
| 245 |
{
|
| 246 |
"epoch": 0.59,
|
| 247 |
"learning_rate": 0.0007199999999999999,
|
| 248 |
+
"loss": 0.6765,
|
| 249 |
"step": 360
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.6,
|
| 253 |
"learning_rate": 0.00074,
|
| 254 |
+
"loss": 0.6596,
|
| 255 |
"step": 370
|
| 256 |
},
|
| 257 |
{
|
| 258 |
"epoch": 0.62,
|
| 259 |
"learning_rate": 0.00076,
|
| 260 |
+
"loss": 0.6884,
|
| 261 |
"step": 380
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"epoch": 0.64,
|
| 265 |
"learning_rate": 0.0007800000000000001,
|
| 266 |
+
"loss": 0.647,
|
| 267 |
"step": 390
|
| 268 |
},
|
| 269 |
{
|
| 270 |
"epoch": 0.65,
|
| 271 |
"learning_rate": 0.0008,
|
| 272 |
+
"loss": 0.713,
|
| 273 |
"step": 400
|
| 274 |
},
|
| 275 |
{
|
| 276 |
"epoch": 0.65,
|
| 277 |
+
"eval_loss": 0.5541791319847107,
|
| 278 |
+
"eval_runtime": 6.0716,
|
| 279 |
+
"eval_samples_per_second": 192.864,
|
| 280 |
+
"eval_steps_per_second": 24.211,
|
| 281 |
"step": 400
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 0.67,
|
| 285 |
"learning_rate": 0.00082,
|
| 286 |
+
"loss": 0.6593,
|
| 287 |
"step": 410
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 0.69,
|
| 291 |
"learning_rate": 0.00084,
|
| 292 |
+
"loss": 0.62,
|
| 293 |
"step": 420
|
| 294 |
},
|
| 295 |
{
|
| 296 |
"epoch": 0.7,
|
| 297 |
"learning_rate": 0.00086,
|
| 298 |
+
"loss": 0.6912,
|
| 299 |
"step": 430
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 0.72,
|
| 303 |
"learning_rate": 0.00088,
|
| 304 |
+
"loss": 0.6407,
|
| 305 |
"step": 440
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 0.73,
|
| 309 |
"learning_rate": 0.0009000000000000001,
|
| 310 |
+
"loss": 0.6444,
|
| 311 |
"step": 450
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 0.75,
|
| 315 |
"learning_rate": 0.00092,
|
| 316 |
+
"loss": 0.6591,
|
| 317 |
"step": 460
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"epoch": 0.77,
|
| 321 |
"learning_rate": 0.00094,
|
| 322 |
+
"loss": 0.6329,
|
| 323 |
"step": 470
|
| 324 |
},
|
| 325 |
{
|
| 326 |
"epoch": 0.78,
|
| 327 |
"learning_rate": 0.00096,
|
| 328 |
+
"loss": 0.6097,
|
| 329 |
"step": 480
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 0.8,
|
| 333 |
"learning_rate": 0.00098,
|
| 334 |
+
"loss": 0.6444,
|
| 335 |
"step": 490
|
| 336 |
},
|
| 337 |
{
|
| 338 |
"epoch": 0.82,
|
| 339 |
"learning_rate": 0.001,
|
| 340 |
+
"loss": 0.6106,
|
| 341 |
"step": 500
|
| 342 |
},
|
| 343 |
{
|
| 344 |
"epoch": 0.82,
|
| 345 |
+
"eval_loss": 0.5469470024108887,
|
| 346 |
+
"eval_runtime": 6.1353,
|
| 347 |
+
"eval_samples_per_second": 190.861,
|
| 348 |
+
"eval_steps_per_second": 23.96,
|
| 349 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 350 |
}
|
| 351 |
],
|
| 352 |
"logging_steps": 10,
|
| 353 |
"max_steps": 6130,
|
| 354 |
"num_train_epochs": 10,
|
| 355 |
"save_steps": 500,
|
| 356 |
+
"total_flos": 939361765588992.0,
|
| 357 |
"trial_name": null,
|
| 358 |
"trial_params": null
|
| 359 |
}
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4600
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c82f8daa53495bfef11faca7a8d954f8023465412fa138e9c80f9e8382c966a8
|
| 3 |
size 4600
|