Upload folder using huggingface_hub
Browse files- adapter_config.json +8 -8
- adapter_model.safetensors +2 -2
- optimizer.pt +2 -2
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +405 -405
- training_args.bin +1 -1
adapter_config.json
CHANGED
|
@@ -12,26 +12,26 @@
|
|
| 12 |
"layers_pattern": null,
|
| 13 |
"layers_to_transform": null,
|
| 14 |
"loftq_config": {},
|
| 15 |
-
"lora_alpha":
|
| 16 |
"lora_bias": false,
|
| 17 |
-
"lora_dropout": 0.
|
| 18 |
"megatron_config": null,
|
| 19 |
"megatron_core": "megatron.core",
|
| 20 |
"modules_to_save": null,
|
| 21 |
"peft_type": "LORA",
|
| 22 |
-
"r":
|
| 23 |
"rank_pattern": {},
|
| 24 |
"revision": null,
|
| 25 |
"target_modules": [
|
| 26 |
-
"k_proj",
|
| 27 |
-
"down_proj",
|
| 28 |
-
"v_proj",
|
| 29 |
"q_proj",
|
| 30 |
"gate_proj",
|
|
|
|
|
|
|
|
|
|
| 31 |
"up_proj",
|
| 32 |
-
"
|
| 33 |
],
|
| 34 |
"task_type": "CAUSAL_LM",
|
| 35 |
"use_dora": false,
|
| 36 |
-
"use_rslora":
|
| 37 |
}
|
|
|
|
| 12 |
"layers_pattern": null,
|
| 13 |
"layers_to_transform": null,
|
| 14 |
"loftq_config": {},
|
| 15 |
+
"lora_alpha": 128,
|
| 16 |
"lora_bias": false,
|
| 17 |
+
"lora_dropout": 0.05,
|
| 18 |
"megatron_config": null,
|
| 19 |
"megatron_core": "megatron.core",
|
| 20 |
"modules_to_save": null,
|
| 21 |
"peft_type": "LORA",
|
| 22 |
+
"r": 64,
|
| 23 |
"rank_pattern": {},
|
| 24 |
"revision": null,
|
| 25 |
"target_modules": [
|
|
|
|
|
|
|
|
|
|
| 26 |
"q_proj",
|
| 27 |
"gate_proj",
|
| 28 |
+
"down_proj",
|
| 29 |
+
"o_proj",
|
| 30 |
+
"v_proj",
|
| 31 |
"up_proj",
|
| 32 |
+
"k_proj"
|
| 33 |
],
|
| 34 |
"task_type": "CAUSAL_LM",
|
| 35 |
"use_dora": false,
|
| 36 |
+
"use_rslora": true
|
| 37 |
}
|
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 sha256:4b790de14d9f71c03183a1b51d97230ee9607a04471d98d7b54f9e511dc6dfa3
|
| 3 |
+
size 479005064
|
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:aa0125972c513d92b32ef60a5e448f6970d678262e9ba35201e9b2a5c21a0afa
|
| 3 |
+
size 958300235
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac62f0360825474af4f175694e4e3357aa4f6dae99ac74f00ee8d3b5796ba657
|
| 3 |
size 14645
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5418e0fc9a3a6a50ea3a7b440dfb8b2fa26686b28c8f28256150a09922035962
|
| 3 |
size 1465
|
trainer_state.json
CHANGED
|
@@ -2,7 +2,7 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 500,
|
| 7 |
"global_step": 100,
|
| 8 |
"is_hyper_param_search": false,
|
|
@@ -10,711 +10,711 @@
|
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
-
"epoch": 0.
|
| 14 |
-
"grad_norm":
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
-
"loss":
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
-
"epoch": 0.
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate":
|
| 23 |
-
"loss":
|
| 24 |
"step": 2
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"epoch": 0.
|
| 28 |
-
"grad_norm":
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"loss":
|
| 31 |
"step": 3
|
| 32 |
},
|
| 33 |
{
|
| 34 |
-
"epoch":
|
| 35 |
-
"grad_norm":
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss": 2.
|
| 38 |
"step": 4
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch":
|
| 42 |
-
"grad_norm":
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss": 2.
|
| 45 |
"step": 5
|
| 46 |
},
|
| 47 |
{
|
| 48 |
-
"epoch":
|
| 49 |
-
"grad_norm":
|
| 50 |
-
"learning_rate":
|
| 51 |
-
"loss": 2.
|
| 52 |
"step": 6
|
| 53 |
},
|
| 54 |
{
|
| 55 |
-
"epoch":
|
| 56 |
-
"grad_norm":
|
| 57 |
-
"learning_rate":
|
| 58 |
-
"loss":
|
| 59 |
"step": 7
|
| 60 |
},
|
| 61 |
{
|
| 62 |
-
"epoch":
|
| 63 |
-
"grad_norm":
|
| 64 |
-
"learning_rate":
|
| 65 |
-
"loss": 2.
|
| 66 |
"step": 8
|
| 67 |
},
|
| 68 |
{
|
| 69 |
-
"epoch":
|
| 70 |
-
"grad_norm":
|
| 71 |
-
"learning_rate":
|
| 72 |
-
"loss": 2.
|
| 73 |
"step": 9
|
| 74 |
},
|
| 75 |
{
|
| 76 |
-
"epoch":
|
| 77 |
-
"grad_norm":
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss":
|
| 80 |
"step": 10
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch":
|
| 84 |
-
"grad_norm":
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss": 2.
|
| 87 |
"step": 11
|
| 88 |
},
|
| 89 |
{
|
| 90 |
-
"epoch":
|
| 91 |
-
"grad_norm":
|
| 92 |
-
"learning_rate":
|
| 93 |
-
"loss":
|
| 94 |
"step": 12
|
| 95 |
},
|
| 96 |
{
|
| 97 |
-
"epoch":
|
| 98 |
-
"grad_norm":
|
| 99 |
-
"learning_rate":
|
| 100 |
-
"loss": 2.
|
| 101 |
"step": 13
|
| 102 |
},
|
| 103 |
{
|
| 104 |
-
"epoch":
|
| 105 |
-
"grad_norm":
|
| 106 |
-
"learning_rate":
|
| 107 |
-
"loss": 2.
|
| 108 |
"step": 14
|
| 109 |
},
|
| 110 |
{
|
| 111 |
-
"epoch":
|
| 112 |
-
"grad_norm":
|
| 113 |
-
"learning_rate":
|
| 114 |
-
"loss":
|
| 115 |
"step": 15
|
| 116 |
},
|
| 117 |
{
|
| 118 |
-
"epoch":
|
| 119 |
-
"grad_norm":
|
| 120 |
-
"learning_rate": 0.
|
| 121 |
-
"loss":
|
| 122 |
"step": 16
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch":
|
| 126 |
-
"grad_norm":
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss":
|
| 129 |
"step": 17
|
| 130 |
},
|
| 131 |
{
|
| 132 |
-
"epoch":
|
| 133 |
-
"grad_norm":
|
| 134 |
-
"learning_rate":
|
| 135 |
-
"loss":
|
| 136 |
"step": 18
|
| 137 |
},
|
| 138 |
{
|
| 139 |
-
"epoch":
|
| 140 |
-
"grad_norm":
|
| 141 |
-
"learning_rate":
|
| 142 |
-
"loss":
|
| 143 |
"step": 19
|
| 144 |
},
|
| 145 |
{
|
| 146 |
-
"epoch":
|
| 147 |
-
"grad_norm":
|
| 148 |
-
"learning_rate":
|
| 149 |
-
"loss":
|
| 150 |
"step": 20
|
| 151 |
},
|
| 152 |
{
|
| 153 |
-
"epoch":
|
| 154 |
-
"grad_norm":
|
| 155 |
-
"learning_rate":
|
| 156 |
-
"loss":
|
| 157 |
"step": 21
|
| 158 |
},
|
| 159 |
{
|
| 160 |
-
"epoch":
|
| 161 |
-
"grad_norm":
|
| 162 |
-
"learning_rate":
|
| 163 |
-
"loss":
|
| 164 |
"step": 22
|
| 165 |
},
|
| 166 |
{
|
| 167 |
-
"epoch":
|
| 168 |
-
"grad_norm":
|
| 169 |
-
"learning_rate":
|
| 170 |
-
"loss": 1.
|
| 171 |
"step": 23
|
| 172 |
},
|
| 173 |
{
|
| 174 |
-
"epoch":
|
| 175 |
-
"grad_norm":
|
| 176 |
-
"learning_rate":
|
| 177 |
-
"loss":
|
| 178 |
"step": 24
|
| 179 |
},
|
| 180 |
{
|
| 181 |
-
"epoch":
|
| 182 |
-
"grad_norm":
|
| 183 |
-
"learning_rate":
|
| 184 |
-
"loss": 1.
|
| 185 |
"step": 25
|
| 186 |
},
|
| 187 |
{
|
| 188 |
-
"epoch":
|
| 189 |
-
"grad_norm":
|
| 190 |
-
"learning_rate":
|
| 191 |
-
"loss":
|
| 192 |
"step": 26
|
| 193 |
},
|
| 194 |
{
|
| 195 |
-
"epoch":
|
| 196 |
-
"grad_norm":
|
| 197 |
-
"learning_rate":
|
| 198 |
-
"loss":
|
| 199 |
"step": 27
|
| 200 |
},
|
| 201 |
{
|
| 202 |
-
"epoch":
|
| 203 |
-
"grad_norm":
|
| 204 |
-
"learning_rate":
|
| 205 |
-
"loss":
|
| 206 |
"step": 28
|
| 207 |
},
|
| 208 |
{
|
| 209 |
-
"epoch":
|
| 210 |
-
"grad_norm":
|
| 211 |
-
"learning_rate":
|
| 212 |
-
"loss":
|
| 213 |
"step": 29
|
| 214 |
},
|
| 215 |
{
|
| 216 |
-
"epoch":
|
| 217 |
-
"grad_norm":
|
| 218 |
-
"learning_rate":
|
| 219 |
-
"loss":
|
| 220 |
"step": 30
|
| 221 |
},
|
| 222 |
{
|
| 223 |
-
"epoch":
|
| 224 |
-
"grad_norm":
|
| 225 |
-
"learning_rate":
|
| 226 |
-
"loss":
|
| 227 |
"step": 31
|
| 228 |
},
|
| 229 |
{
|
| 230 |
-
"epoch":
|
| 231 |
-
"grad_norm":
|
| 232 |
-
"learning_rate":
|
| 233 |
-
"loss":
|
| 234 |
"step": 32
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch":
|
| 238 |
-
"grad_norm":
|
| 239 |
-
"learning_rate":
|
| 240 |
-
"loss":
|
| 241 |
"step": 33
|
| 242 |
},
|
| 243 |
{
|
| 244 |
-
"epoch":
|
| 245 |
-
"grad_norm":
|
| 246 |
-
"learning_rate":
|
| 247 |
-
"loss":
|
| 248 |
"step": 34
|
| 249 |
},
|
| 250 |
{
|
| 251 |
-
"epoch":
|
| 252 |
-
"grad_norm":
|
| 253 |
-
"learning_rate":
|
| 254 |
-
"loss":
|
| 255 |
"step": 35
|
| 256 |
},
|
| 257 |
{
|
| 258 |
-
"epoch":
|
| 259 |
-
"grad_norm":
|
| 260 |
-
"learning_rate":
|
| 261 |
-
"loss":
|
| 262 |
"step": 36
|
| 263 |
},
|
| 264 |
{
|
| 265 |
-
"epoch":
|
| 266 |
-
"grad_norm":
|
| 267 |
-
"learning_rate":
|
| 268 |
-
"loss":
|
| 269 |
"step": 37
|
| 270 |
},
|
| 271 |
{
|
| 272 |
-
"epoch":
|
| 273 |
-
"grad_norm":
|
| 274 |
-
"learning_rate":
|
| 275 |
-
"loss":
|
| 276 |
"step": 38
|
| 277 |
},
|
| 278 |
{
|
| 279 |
-
"epoch":
|
| 280 |
-
"grad_norm":
|
| 281 |
-
"learning_rate":
|
| 282 |
-
"loss":
|
| 283 |
"step": 39
|
| 284 |
},
|
| 285 |
{
|
| 286 |
-
"epoch":
|
| 287 |
-
"grad_norm":
|
| 288 |
-
"learning_rate":
|
| 289 |
-
"loss":
|
| 290 |
"step": 40
|
| 291 |
},
|
| 292 |
{
|
| 293 |
-
"epoch":
|
| 294 |
-
"grad_norm":
|
| 295 |
-
"learning_rate":
|
| 296 |
-
"loss":
|
| 297 |
"step": 41
|
| 298 |
},
|
| 299 |
{
|
| 300 |
-
"epoch":
|
| 301 |
-
"grad_norm":
|
| 302 |
-
"learning_rate":
|
| 303 |
-
"loss":
|
| 304 |
"step": 42
|
| 305 |
},
|
| 306 |
{
|
| 307 |
-
"epoch":
|
| 308 |
-
"grad_norm":
|
| 309 |
-
"learning_rate":
|
| 310 |
-
"loss":
|
| 311 |
"step": 43
|
| 312 |
},
|
| 313 |
{
|
| 314 |
-
"epoch":
|
| 315 |
-
"grad_norm":
|
| 316 |
-
"learning_rate":
|
| 317 |
-
"loss":
|
| 318 |
"step": 44
|
| 319 |
},
|
| 320 |
{
|
| 321 |
-
"epoch":
|
| 322 |
-
"grad_norm":
|
| 323 |
-
"learning_rate":
|
| 324 |
-
"loss":
|
| 325 |
"step": 45
|
| 326 |
},
|
| 327 |
{
|
| 328 |
-
"epoch":
|
| 329 |
-
"grad_norm":
|
| 330 |
-
"learning_rate":
|
| 331 |
-
"loss":
|
| 332 |
"step": 46
|
| 333 |
},
|
| 334 |
{
|
| 335 |
-
"epoch":
|
| 336 |
-
"grad_norm":
|
| 337 |
-
"learning_rate":
|
| 338 |
-
"loss":
|
| 339 |
"step": 47
|
| 340 |
},
|
| 341 |
{
|
| 342 |
-
"epoch":
|
| 343 |
-
"grad_norm":
|
| 344 |
-
"learning_rate":
|
| 345 |
-
"loss":
|
| 346 |
"step": 48
|
| 347 |
},
|
| 348 |
{
|
| 349 |
-
"epoch":
|
| 350 |
-
"grad_norm":
|
| 351 |
-
"learning_rate":
|
| 352 |
-
"loss":
|
| 353 |
"step": 49
|
| 354 |
},
|
| 355 |
{
|
| 356 |
-
"epoch":
|
| 357 |
-
"grad_norm":
|
| 358 |
-
"learning_rate":
|
| 359 |
-
"loss":
|
| 360 |
"step": 50
|
| 361 |
},
|
| 362 |
{
|
| 363 |
-
"epoch":
|
| 364 |
-
"grad_norm":
|
| 365 |
-
"learning_rate":
|
| 366 |
-
"loss":
|
| 367 |
"step": 51
|
| 368 |
},
|
| 369 |
{
|
| 370 |
-
"epoch":
|
| 371 |
-
"grad_norm":
|
| 372 |
-
"learning_rate":
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 52
|
| 375 |
},
|
| 376 |
{
|
| 377 |
-
"epoch":
|
| 378 |
-
"grad_norm":
|
| 379 |
-
"learning_rate":
|
| 380 |
-
"loss":
|
| 381 |
"step": 53
|
| 382 |
},
|
| 383 |
{
|
| 384 |
-
"epoch":
|
| 385 |
-
"grad_norm":
|
| 386 |
-
"learning_rate":
|
| 387 |
-
"loss":
|
| 388 |
"step": 54
|
| 389 |
},
|
| 390 |
{
|
| 391 |
-
"epoch":
|
| 392 |
-
"grad_norm":
|
| 393 |
-
"learning_rate":
|
| 394 |
-
"loss":
|
| 395 |
"step": 55
|
| 396 |
},
|
| 397 |
{
|
| 398 |
-
"epoch":
|
| 399 |
-
"grad_norm":
|
| 400 |
-
"learning_rate":
|
| 401 |
-
"loss":
|
| 402 |
"step": 56
|
| 403 |
},
|
| 404 |
{
|
| 405 |
-
"epoch":
|
| 406 |
-
"grad_norm":
|
| 407 |
-
"learning_rate":
|
| 408 |
-
"loss":
|
| 409 |
"step": 57
|
| 410 |
},
|
| 411 |
{
|
| 412 |
-
"epoch":
|
| 413 |
-
"grad_norm":
|
| 414 |
-
"learning_rate":
|
| 415 |
-
"loss":
|
| 416 |
"step": 58
|
| 417 |
},
|
| 418 |
{
|
| 419 |
-
"epoch":
|
| 420 |
-
"grad_norm":
|
| 421 |
-
"learning_rate":
|
| 422 |
-
"loss":
|
| 423 |
"step": 59
|
| 424 |
},
|
| 425 |
{
|
| 426 |
-
"epoch":
|
| 427 |
-
"grad_norm":
|
| 428 |
-
"learning_rate":
|
| 429 |
-
"loss": 0.
|
| 430 |
"step": 60
|
| 431 |
},
|
| 432 |
{
|
| 433 |
-
"epoch":
|
| 434 |
-
"grad_norm":
|
| 435 |
-
"learning_rate":
|
| 436 |
-
"loss": 0.
|
| 437 |
"step": 61
|
| 438 |
},
|
| 439 |
{
|
| 440 |
-
"epoch":
|
| 441 |
-
"grad_norm":
|
| 442 |
-
"learning_rate":
|
| 443 |
-
"loss": 0.
|
| 444 |
"step": 62
|
| 445 |
},
|
| 446 |
{
|
| 447 |
-
"epoch":
|
| 448 |
-
"grad_norm":
|
| 449 |
-
"learning_rate":
|
| 450 |
-
"loss": 0.
|
| 451 |
"step": 63
|
| 452 |
},
|
| 453 |
{
|
| 454 |
-
"epoch":
|
| 455 |
-
"grad_norm":
|
| 456 |
-
"learning_rate":
|
| 457 |
-
"loss": 0.
|
| 458 |
"step": 64
|
| 459 |
},
|
| 460 |
{
|
| 461 |
-
"epoch":
|
| 462 |
-
"grad_norm":
|
| 463 |
-
"learning_rate":
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 65
|
| 466 |
},
|
| 467 |
{
|
| 468 |
-
"epoch":
|
| 469 |
-
"grad_norm":
|
| 470 |
-
"learning_rate":
|
| 471 |
-
"loss": 0.
|
| 472 |
"step": 66
|
| 473 |
},
|
| 474 |
{
|
| 475 |
-
"epoch":
|
| 476 |
-
"grad_norm":
|
| 477 |
-
"learning_rate":
|
| 478 |
-
"loss": 0.
|
| 479 |
"step": 67
|
| 480 |
},
|
| 481 |
{
|
| 482 |
-
"epoch":
|
| 483 |
-
"grad_norm":
|
| 484 |
-
"learning_rate":
|
| 485 |
-
"loss":
|
| 486 |
"step": 68
|
| 487 |
},
|
| 488 |
{
|
| 489 |
-
"epoch":
|
| 490 |
-
"grad_norm":
|
| 491 |
-
"learning_rate":
|
| 492 |
-
"loss":
|
| 493 |
"step": 69
|
| 494 |
},
|
| 495 |
{
|
| 496 |
-
"epoch":
|
| 497 |
-
"grad_norm":
|
| 498 |
-
"learning_rate":
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 70
|
| 501 |
},
|
| 502 |
{
|
| 503 |
-
"epoch":
|
| 504 |
-
"grad_norm":
|
| 505 |
-
"learning_rate":
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 71
|
| 508 |
},
|
| 509 |
{
|
| 510 |
-
"epoch":
|
| 511 |
-
"grad_norm":
|
| 512 |
-
"learning_rate":
|
| 513 |
-
"loss": 0.
|
| 514 |
"step": 72
|
| 515 |
},
|
| 516 |
{
|
| 517 |
-
"epoch":
|
| 518 |
-
"grad_norm":
|
| 519 |
-
"learning_rate":
|
| 520 |
-
"loss": 0.
|
| 521 |
"step": 73
|
| 522 |
},
|
| 523 |
{
|
| 524 |
-
"epoch":
|
| 525 |
-
"grad_norm":
|
| 526 |
-
"learning_rate":
|
| 527 |
-
"loss": 0.
|
| 528 |
"step": 74
|
| 529 |
},
|
| 530 |
{
|
| 531 |
-
"epoch":
|
| 532 |
-
"grad_norm":
|
| 533 |
-
"learning_rate":
|
| 534 |
-
"loss": 0.
|
| 535 |
"step": 75
|
| 536 |
},
|
| 537 |
{
|
| 538 |
-
"epoch":
|
| 539 |
-
"grad_norm":
|
| 540 |
-
"learning_rate":
|
| 541 |
-
"loss": 0.
|
| 542 |
"step": 76
|
| 543 |
},
|
| 544 |
{
|
| 545 |
-
"epoch":
|
| 546 |
-
"grad_norm":
|
| 547 |
-
"learning_rate":
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 77
|
| 550 |
},
|
| 551 |
{
|
| 552 |
-
"epoch":
|
| 553 |
-
"grad_norm":
|
| 554 |
-
"learning_rate": 8.
|
| 555 |
-
"loss": 0.
|
| 556 |
"step": 78
|
| 557 |
},
|
| 558 |
{
|
| 559 |
-
"epoch":
|
| 560 |
-
"grad_norm":
|
| 561 |
-
"learning_rate":
|
| 562 |
-
"loss": 0.
|
| 563 |
"step": 79
|
| 564 |
},
|
| 565 |
{
|
| 566 |
-
"epoch":
|
| 567 |
-
"grad_norm":
|
| 568 |
-
"learning_rate":
|
| 569 |
-
"loss": 0.
|
| 570 |
"step": 80
|
| 571 |
},
|
| 572 |
{
|
| 573 |
-
"epoch":
|
| 574 |
-
"grad_norm":
|
| 575 |
-
"learning_rate":
|
| 576 |
-
"loss": 0.
|
| 577 |
"step": 81
|
| 578 |
},
|
| 579 |
{
|
| 580 |
-
"epoch":
|
| 581 |
-
"grad_norm":
|
| 582 |
-
"learning_rate":
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 82
|
| 585 |
},
|
| 586 |
{
|
| 587 |
-
"epoch":
|
| 588 |
-
"grad_norm":
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 83
|
| 592 |
},
|
| 593 |
{
|
| 594 |
-
"epoch":
|
| 595 |
-
"grad_norm":
|
| 596 |
-
"learning_rate":
|
| 597 |
-
"loss": 0.
|
| 598 |
"step": 84
|
| 599 |
},
|
| 600 |
{
|
| 601 |
-
"epoch":
|
| 602 |
-
"grad_norm":
|
| 603 |
-
"learning_rate":
|
| 604 |
-
"loss": 0.
|
| 605 |
"step": 85
|
| 606 |
},
|
| 607 |
{
|
| 608 |
-
"epoch":
|
| 609 |
-
"grad_norm":
|
| 610 |
-
"learning_rate":
|
| 611 |
-
"loss": 0.
|
| 612 |
"step": 86
|
| 613 |
},
|
| 614 |
{
|
| 615 |
-
"epoch":
|
| 616 |
-
"grad_norm":
|
| 617 |
-
"learning_rate":
|
| 618 |
-
"loss": 0.
|
| 619 |
"step": 87
|
| 620 |
},
|
| 621 |
{
|
| 622 |
-
"epoch":
|
| 623 |
-
"grad_norm":
|
| 624 |
-
"learning_rate":
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 88
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch":
|
| 630 |
-
"grad_norm":
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss": 0.
|
| 633 |
"step": 89
|
| 634 |
},
|
| 635 |
{
|
| 636 |
-
"epoch":
|
| 637 |
-
"grad_norm":
|
| 638 |
-
"learning_rate":
|
| 639 |
-
"loss": 0.
|
| 640 |
"step": 90
|
| 641 |
},
|
| 642 |
{
|
| 643 |
-
"epoch":
|
| 644 |
-
"grad_norm":
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss": 0.
|
| 647 |
"step": 91
|
| 648 |
},
|
| 649 |
{
|
| 650 |
-
"epoch":
|
| 651 |
-
"grad_norm":
|
| 652 |
-
"learning_rate":
|
| 653 |
-
"loss": 0.
|
| 654 |
"step": 92
|
| 655 |
},
|
| 656 |
{
|
| 657 |
-
"epoch":
|
| 658 |
-
"grad_norm":
|
| 659 |
-
"learning_rate":
|
| 660 |
-
"loss": 0.
|
| 661 |
"step": 93
|
| 662 |
},
|
| 663 |
{
|
| 664 |
-
"epoch":
|
| 665 |
-
"grad_norm":
|
| 666 |
-
"learning_rate": 8.
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 94
|
| 669 |
},
|
| 670 |
{
|
| 671 |
-
"epoch":
|
| 672 |
-
"grad_norm":
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss": 0.
|
| 675 |
"step": 95
|
| 676 |
},
|
| 677 |
{
|
| 678 |
-
"epoch":
|
| 679 |
-
"grad_norm":
|
| 680 |
-
"learning_rate":
|
| 681 |
-
"loss": 0.
|
| 682 |
"step": 96
|
| 683 |
},
|
| 684 |
{
|
| 685 |
-
"epoch":
|
| 686 |
-
"grad_norm":
|
| 687 |
-
"learning_rate":
|
| 688 |
-
"loss": 0.
|
| 689 |
"step": 97
|
| 690 |
},
|
| 691 |
{
|
| 692 |
-
"epoch":
|
| 693 |
-
"grad_norm":
|
| 694 |
-
"learning_rate":
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 98
|
| 697 |
},
|
| 698 |
{
|
| 699 |
-
"epoch":
|
| 700 |
-
"grad_norm":
|
| 701 |
-
"learning_rate":
|
| 702 |
-
"loss": 0.
|
| 703 |
"step": 99
|
| 704 |
},
|
| 705 |
{
|
| 706 |
-
"epoch":
|
| 707 |
-
"grad_norm":
|
| 708 |
-
"learning_rate":
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 100
|
| 711 |
}
|
| 712 |
],
|
| 713 |
"logging_steps": 1,
|
| 714 |
-
"max_steps":
|
| 715 |
"num_input_tokens_seen": 0,
|
| 716 |
-
"num_train_epochs":
|
| 717 |
-
"save_steps":
|
| 718 |
"stateful_callbacks": {
|
| 719 |
"TrainerControl": {
|
| 720 |
"args": {
|
|
@@ -722,12 +722,12 @@
|
|
| 722 |
"should_evaluate": false,
|
| 723 |
"should_log": false,
|
| 724 |
"should_save": true,
|
| 725 |
-
"should_training_stop":
|
| 726 |
},
|
| 727 |
"attributes": {}
|
| 728 |
}
|
| 729 |
},
|
| 730 |
-
"total_flos":
|
| 731 |
"train_batch_size": 2,
|
| 732 |
"trial_name": null,
|
| 733 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
"global_step": 100,
|
| 8 |
"is_hyper_param_search": false,
|
|
|
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"epoch": 0.05063291139240506,
|
| 14 |
+
"grad_norm": 29.89462661743164,
|
| 15 |
"learning_rate": 0.0,
|
| 16 |
+
"loss": 3.0471,
|
| 17 |
"step": 1
|
| 18 |
},
|
| 19 |
{
|
| 20 |
+
"epoch": 0.10126582278481013,
|
| 21 |
+
"grad_norm": 48.531124114990234,
|
| 22 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 23 |
+
"loss": 3.2897,
|
| 24 |
"step": 2
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.1518987341772152,
|
| 28 |
+
"grad_norm": 7.42547082901001,
|
| 29 |
+
"learning_rate": 6.666666666666667e-05,
|
| 30 |
+
"loss": 3.0122,
|
| 31 |
"step": 3
|
| 32 |
},
|
| 33 |
{
|
| 34 |
+
"epoch": 0.20253164556962025,
|
| 35 |
+
"grad_norm": 6.992433547973633,
|
| 36 |
+
"learning_rate": 0.0001,
|
| 37 |
+
"loss": 2.608,
|
| 38 |
"step": 4
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.25316455696202533,
|
| 42 |
+
"grad_norm": 5.709129810333252,
|
| 43 |
+
"learning_rate": 0.00013333333333333334,
|
| 44 |
+
"loss": 2.2012,
|
| 45 |
"step": 5
|
| 46 |
},
|
| 47 |
{
|
| 48 |
+
"epoch": 0.3037974683544304,
|
| 49 |
+
"grad_norm": 7.335264205932617,
|
| 50 |
+
"learning_rate": 0.00016666666666666666,
|
| 51 |
+
"loss": 2.85,
|
| 52 |
"step": 6
|
| 53 |
},
|
| 54 |
{
|
| 55 |
+
"epoch": 0.35443037974683544,
|
| 56 |
+
"grad_norm": 5.759677410125732,
|
| 57 |
+
"learning_rate": 0.0002,
|
| 58 |
+
"loss": 2.5905,
|
| 59 |
"step": 7
|
| 60 |
},
|
| 61 |
{
|
| 62 |
+
"epoch": 0.4050632911392405,
|
| 63 |
+
"grad_norm": 5.33920955657959,
|
| 64 |
+
"learning_rate": 0.00023333333333333333,
|
| 65 |
+
"loss": 2.4433,
|
| 66 |
"step": 8
|
| 67 |
},
|
| 68 |
{
|
| 69 |
+
"epoch": 0.45569620253164556,
|
| 70 |
+
"grad_norm": 4.677101135253906,
|
| 71 |
+
"learning_rate": 0.0002666666666666667,
|
| 72 |
+
"loss": 2.7977,
|
| 73 |
"step": 9
|
| 74 |
},
|
| 75 |
{
|
| 76 |
+
"epoch": 0.5063291139240507,
|
| 77 |
+
"grad_norm": 5.277487277984619,
|
| 78 |
+
"learning_rate": 0.0003,
|
| 79 |
+
"loss": 2.6438,
|
| 80 |
"step": 10
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 0.5569620253164557,
|
| 84 |
+
"grad_norm": 5.3378729820251465,
|
| 85 |
+
"learning_rate": 0.0003333333333333333,
|
| 86 |
+
"loss": 2.6596,
|
| 87 |
"step": 11
|
| 88 |
},
|
| 89 |
{
|
| 90 |
+
"epoch": 0.6075949367088608,
|
| 91 |
+
"grad_norm": 4.5508222579956055,
|
| 92 |
+
"learning_rate": 0.00036666666666666667,
|
| 93 |
+
"loss": 2.8059,
|
| 94 |
"step": 12
|
| 95 |
},
|
| 96 |
{
|
| 97 |
+
"epoch": 0.6582278481012658,
|
| 98 |
+
"grad_norm": 4.640895843505859,
|
| 99 |
+
"learning_rate": 0.0004,
|
| 100 |
+
"loss": 2.4764,
|
| 101 |
"step": 13
|
| 102 |
},
|
| 103 |
{
|
| 104 |
+
"epoch": 0.7088607594936709,
|
| 105 |
+
"grad_norm": 4.381677627563477,
|
| 106 |
+
"learning_rate": 0.00043333333333333337,
|
| 107 |
+
"loss": 2.1133,
|
| 108 |
"step": 14
|
| 109 |
},
|
| 110 |
{
|
| 111 |
+
"epoch": 0.759493670886076,
|
| 112 |
+
"grad_norm": 4.960311412811279,
|
| 113 |
+
"learning_rate": 0.00046666666666666666,
|
| 114 |
+
"loss": 2.4421,
|
| 115 |
"step": 15
|
| 116 |
},
|
| 117 |
{
|
| 118 |
+
"epoch": 0.810126582278481,
|
| 119 |
+
"grad_norm": 4.729428768157959,
|
| 120 |
+
"learning_rate": 0.0005,
|
| 121 |
+
"loss": 2.8667,
|
| 122 |
"step": 16
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 0.8607594936708861,
|
| 126 |
+
"grad_norm": 12.342778205871582,
|
| 127 |
+
"learning_rate": 0.0004998292650357557,
|
| 128 |
+
"loss": 2.9229,
|
| 129 |
"step": 17
|
| 130 |
},
|
| 131 |
{
|
| 132 |
+
"epoch": 0.9113924050632911,
|
| 133 |
+
"grad_norm": 6.200201988220215,
|
| 134 |
+
"learning_rate": 0.0004993172933464471,
|
| 135 |
+
"loss": 3.1633,
|
| 136 |
"step": 18
|
| 137 |
},
|
| 138 |
{
|
| 139 |
+
"epoch": 0.9620253164556962,
|
| 140 |
+
"grad_norm": 7.932784557342529,
|
| 141 |
+
"learning_rate": 0.0004984647842238185,
|
| 142 |
+
"loss": 2.9541,
|
| 143 |
"step": 19
|
| 144 |
},
|
| 145 |
{
|
| 146 |
+
"epoch": 1.0,
|
| 147 |
+
"grad_norm": 6.065736770629883,
|
| 148 |
+
"learning_rate": 0.0004972729020927865,
|
| 149 |
+
"loss": 2.6617,
|
| 150 |
"step": 20
|
| 151 |
},
|
| 152 |
{
|
| 153 |
+
"epoch": 1.0506329113924051,
|
| 154 |
+
"grad_norm": 9.076287269592285,
|
| 155 |
+
"learning_rate": 0.0004957432749209755,
|
| 156 |
+
"loss": 2.0925,
|
| 157 |
"step": 21
|
| 158 |
},
|
| 159 |
{
|
| 160 |
+
"epoch": 1.1012658227848102,
|
| 161 |
+
"grad_norm": 6.06325626373291,
|
| 162 |
+
"learning_rate": 0.0004938779919951092,
|
| 163 |
+
"loss": 1.8076,
|
| 164 |
"step": 22
|
| 165 |
},
|
| 166 |
{
|
| 167 |
+
"epoch": 1.1518987341772151,
|
| 168 |
+
"grad_norm": 4.740463733673096,
|
| 169 |
+
"learning_rate": 0.0004916796010672969,
|
| 170 |
+
"loss": 1.8304,
|
| 171 |
"step": 23
|
| 172 |
},
|
| 173 |
{
|
| 174 |
+
"epoch": 1.2025316455696202,
|
| 175 |
+
"grad_norm": 7.353160858154297,
|
| 176 |
+
"learning_rate": 0.0004891511048751102,
|
| 177 |
+
"loss": 2.1726,
|
| 178 |
"step": 24
|
| 179 |
},
|
| 180 |
{
|
| 181 |
+
"epoch": 1.2531645569620253,
|
| 182 |
+
"grad_norm": 3.101592779159546,
|
| 183 |
+
"learning_rate": 0.00048629595704020493,
|
| 184 |
+
"loss": 1.4538,
|
| 185 |
"step": 25
|
| 186 |
},
|
| 187 |
{
|
| 188 |
+
"epoch": 1.3037974683544304,
|
| 189 |
+
"grad_norm": 4.622443675994873,
|
| 190 |
+
"learning_rate": 0.00048311805735108893,
|
| 191 |
+
"loss": 1.9408,
|
| 192 |
"step": 26
|
| 193 |
},
|
| 194 |
{
|
| 195 |
+
"epoch": 1.3544303797468356,
|
| 196 |
+
"grad_norm": 6.337324619293213,
|
| 197 |
+
"learning_rate": 0.0004796217464364808,
|
| 198 |
+
"loss": 2.0398,
|
| 199 |
"step": 27
|
| 200 |
},
|
| 201 |
{
|
| 202 |
+
"epoch": 1.4050632911392404,
|
| 203 |
+
"grad_norm": 4.56858491897583,
|
| 204 |
+
"learning_rate": 0.0004758117998365322,
|
| 205 |
+
"loss": 1.8656,
|
| 206 |
"step": 28
|
| 207 |
},
|
| 208 |
{
|
| 209 |
+
"epoch": 1.4556962025316456,
|
| 210 |
+
"grad_norm": 196.54071044921875,
|
| 211 |
+
"learning_rate": 0.00047169342148001547,
|
| 212 |
+
"loss": 1.8473,
|
| 213 |
"step": 29
|
| 214 |
},
|
| 215 |
{
|
| 216 |
+
"epoch": 1.5063291139240507,
|
| 217 |
+
"grad_norm": 14.125255584716797,
|
| 218 |
+
"learning_rate": 0.0004672722365763821,
|
| 219 |
+
"loss": 2.092,
|
| 220 |
"step": 30
|
| 221 |
},
|
| 222 |
{
|
| 223 |
+
"epoch": 1.5569620253164556,
|
| 224 |
+
"grad_norm": 5.323953628540039,
|
| 225 |
+
"learning_rate": 0.0004625542839324036,
|
| 226 |
+
"loss": 1.72,
|
| 227 |
"step": 31
|
| 228 |
},
|
| 229 |
{
|
| 230 |
+
"epoch": 1.6075949367088609,
|
| 231 |
+
"grad_norm": 4.4790873527526855,
|
| 232 |
+
"learning_rate": 0.00045754600770388763,
|
| 233 |
+
"loss": 1.7183,
|
| 234 |
"step": 32
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 1.6582278481012658,
|
| 238 |
+
"grad_norm": 9.435335159301758,
|
| 239 |
+
"learning_rate": 0.0004522542485937369,
|
| 240 |
+
"loss": 1.9387,
|
| 241 |
"step": 33
|
| 242 |
},
|
| 243 |
{
|
| 244 |
+
"epoch": 1.7088607594936709,
|
| 245 |
+
"grad_norm": 3.8727502822875977,
|
| 246 |
+
"learning_rate": 0.0004466862345083708,
|
| 247 |
+
"loss": 2.0358,
|
| 248 |
"step": 34
|
| 249 |
},
|
| 250 |
{
|
| 251 |
+
"epoch": 1.759493670886076,
|
| 252 |
+
"grad_norm": 6.095119953155518,
|
| 253 |
+
"learning_rate": 0.0004408495706852758,
|
| 254 |
+
"loss": 1.8897,
|
| 255 |
"step": 35
|
| 256 |
},
|
| 257 |
{
|
| 258 |
+
"epoch": 1.810126582278481,
|
| 259 |
+
"grad_norm": 6.482694625854492,
|
| 260 |
+
"learning_rate": 0.00043475222930516476,
|
| 261 |
+
"loss": 2.0688,
|
| 262 |
"step": 36
|
| 263 |
},
|
| 264 |
{
|
| 265 |
+
"epoch": 1.8607594936708862,
|
| 266 |
+
"grad_norm": 3.911402702331543,
|
| 267 |
+
"learning_rate": 0.0004284025386029381,
|
| 268 |
+
"loss": 1.8147,
|
| 269 |
"step": 37
|
| 270 |
},
|
| 271 |
{
|
| 272 |
+
"epoch": 1.9113924050632911,
|
| 273 |
+
"grad_norm": 4.3811445236206055,
|
| 274 |
+
"learning_rate": 0.00042180917149231567,
|
| 275 |
+
"loss": 1.8666,
|
| 276 |
"step": 38
|
| 277 |
},
|
| 278 |
{
|
| 279 |
+
"epoch": 1.9620253164556962,
|
| 280 |
+
"grad_norm": 4.967111587524414,
|
| 281 |
+
"learning_rate": 0.0004149811337196807,
|
| 282 |
+
"loss": 1.8184,
|
| 283 |
"step": 39
|
| 284 |
},
|
| 285 |
{
|
| 286 |
+
"epoch": 2.0,
|
| 287 |
+
"grad_norm": 4.769927024841309,
|
| 288 |
+
"learning_rate": 0.00040792775156331276,
|
| 289 |
+
"loss": 2.0886,
|
| 290 |
"step": 40
|
| 291 |
},
|
| 292 |
{
|
| 293 |
+
"epoch": 2.050632911392405,
|
| 294 |
+
"grad_norm": 3.3723819255828857,
|
| 295 |
+
"learning_rate": 0.0004006586590948141,
|
| 296 |
+
"loss": 1.1652,
|
| 297 |
"step": 41
|
| 298 |
},
|
| 299 |
{
|
| 300 |
+
"epoch": 2.1012658227848102,
|
| 301 |
+
"grad_norm": 3.5571470260620117,
|
| 302 |
+
"learning_rate": 0.0003931837850201263,
|
| 303 |
+
"loss": 1.0747,
|
| 304 |
"step": 42
|
| 305 |
},
|
| 306 |
{
|
| 307 |
+
"epoch": 2.151898734177215,
|
| 308 |
+
"grad_norm": 5.328244686126709,
|
| 309 |
+
"learning_rate": 0.00038551333911811237,
|
| 310 |
+
"loss": 1.1751,
|
| 311 |
"step": 43
|
| 312 |
},
|
| 313 |
{
|
| 314 |
+
"epoch": 2.2025316455696204,
|
| 315 |
+
"grad_norm": 5.772764682769775,
|
| 316 |
+
"learning_rate": 0.00037765779829522674,
|
| 317 |
+
"loss": 1.247,
|
| 318 |
"step": 44
|
| 319 |
},
|
| 320 |
{
|
| 321 |
+
"epoch": 2.2531645569620253,
|
| 322 |
+
"grad_norm": 4.74567174911499,
|
| 323 |
+
"learning_rate": 0.00036962789227532164,
|
| 324 |
+
"loss": 1.1088,
|
| 325 |
"step": 45
|
| 326 |
},
|
| 327 |
{
|
| 328 |
+
"epoch": 2.3037974683544302,
|
| 329 |
+
"grad_norm": 127.826416015625,
|
| 330 |
+
"learning_rate": 0.0003614345889441346,
|
| 331 |
+
"loss": 1.7475,
|
| 332 |
"step": 46
|
| 333 |
},
|
| 334 |
{
|
| 335 |
+
"epoch": 2.3544303797468356,
|
| 336 |
+
"grad_norm": 5.490947246551514,
|
| 337 |
+
"learning_rate": 0.0003530890793684759,
|
| 338 |
+
"loss": 1.2896,
|
| 339 |
"step": 47
|
| 340 |
},
|
| 341 |
{
|
| 342 |
+
"epoch": 2.4050632911392404,
|
| 343 |
+
"grad_norm": 11.671218872070312,
|
| 344 |
+
"learning_rate": 0.0003446027625105776,
|
| 345 |
+
"loss": 1.6716,
|
| 346 |
"step": 48
|
| 347 |
},
|
| 348 |
{
|
| 349 |
+
"epoch": 2.4556962025316453,
|
| 350 |
+
"grad_norm": 3.728645086288452,
|
| 351 |
+
"learning_rate": 0.00033598722965848206,
|
| 352 |
+
"loss": 1.461,
|
| 353 |
"step": 49
|
| 354 |
},
|
| 355 |
{
|
| 356 |
+
"epoch": 2.5063291139240507,
|
| 357 |
+
"grad_norm": 3.956665515899658,
|
| 358 |
+
"learning_rate": 0.00032725424859373687,
|
| 359 |
+
"loss": 1.1111,
|
| 360 |
"step": 50
|
| 361 |
},
|
| 362 |
{
|
| 363 |
+
"epoch": 2.5569620253164556,
|
| 364 |
+
"grad_norm": 4.517430305480957,
|
| 365 |
+
"learning_rate": 0.0003184157475180208,
|
| 366 |
+
"loss": 1.2541,
|
| 367 |
"step": 51
|
| 368 |
},
|
| 369 |
{
|
| 370 |
+
"epoch": 2.607594936708861,
|
| 371 |
+
"grad_norm": 4.479220390319824,
|
| 372 |
+
"learning_rate": 0.00030948379876065467,
|
| 373 |
+
"loss": 0.9342,
|
| 374 |
"step": 52
|
| 375 |
},
|
| 376 |
{
|
| 377 |
+
"epoch": 2.6582278481012658,
|
| 378 |
+
"grad_norm": 5.564835548400879,
|
| 379 |
+
"learning_rate": 0.00030047060228925254,
|
| 380 |
+
"loss": 1.5894,
|
| 381 |
"step": 53
|
| 382 |
},
|
| 383 |
{
|
| 384 |
+
"epoch": 2.708860759493671,
|
| 385 |
+
"grad_norm": 6.077727317810059,
|
| 386 |
+
"learning_rate": 0.0002913884690460325,
|
| 387 |
+
"loss": 1.5569,
|
| 388 |
"step": 54
|
| 389 |
},
|
| 390 |
{
|
| 391 |
+
"epoch": 2.759493670886076,
|
| 392 |
+
"grad_norm": 4.0990495681762695,
|
| 393 |
+
"learning_rate": 0.00028224980413255084,
|
| 394 |
+
"loss": 1.0699,
|
| 395 |
"step": 55
|
| 396 |
},
|
| 397 |
{
|
| 398 |
+
"epoch": 2.810126582278481,
|
| 399 |
+
"grad_norm": 3.1783435344696045,
|
| 400 |
+
"learning_rate": 0.0002730670898658255,
|
| 401 |
+
"loss": 1.1104,
|
| 402 |
"step": 56
|
| 403 |
},
|
| 404 |
{
|
| 405 |
+
"epoch": 2.8607594936708862,
|
| 406 |
+
"grad_norm": 4.6365814208984375,
|
| 407 |
+
"learning_rate": 0.0002638528687289925,
|
| 408 |
+
"loss": 1.1447,
|
| 409 |
"step": 57
|
| 410 |
},
|
| 411 |
{
|
| 412 |
+
"epoch": 2.911392405063291,
|
| 413 |
+
"grad_norm": 5.0435285568237305,
|
| 414 |
+
"learning_rate": 0.0002546197262397825,
|
| 415 |
+
"loss": 1.1904,
|
| 416 |
"step": 58
|
| 417 |
},
|
| 418 |
{
|
| 419 |
+
"epoch": 2.962025316455696,
|
| 420 |
+
"grad_norm": 9.943424224853516,
|
| 421 |
+
"learning_rate": 0.0002453802737602176,
|
| 422 |
+
"loss": 1.2399,
|
| 423 |
"step": 59
|
| 424 |
},
|
| 425 |
{
|
| 426 |
+
"epoch": 3.0,
|
| 427 |
+
"grad_norm": 4.899236679077148,
|
| 428 |
+
"learning_rate": 0.00023614713127100752,
|
| 429 |
+
"loss": 0.9004,
|
| 430 |
"step": 60
|
| 431 |
},
|
| 432 |
{
|
| 433 |
+
"epoch": 3.050632911392405,
|
| 434 |
+
"grad_norm": 3.2913708686828613,
|
| 435 |
+
"learning_rate": 0.00022693291013417452,
|
| 436 |
+
"loss": 0.7823,
|
| 437 |
"step": 61
|
| 438 |
},
|
| 439 |
{
|
| 440 |
+
"epoch": 3.1012658227848102,
|
| 441 |
+
"grad_norm": 2.5191409587860107,
|
| 442 |
+
"learning_rate": 0.00021775019586744925,
|
| 443 |
+
"loss": 0.5764,
|
| 444 |
"step": 62
|
| 445 |
},
|
| 446 |
{
|
| 447 |
+
"epoch": 3.151898734177215,
|
| 448 |
+
"grad_norm": 3.1726913452148438,
|
| 449 |
+
"learning_rate": 0.0002086115309539675,
|
| 450 |
+
"loss": 0.6909,
|
| 451 |
"step": 63
|
| 452 |
},
|
| 453 |
{
|
| 454 |
+
"epoch": 3.2025316455696204,
|
| 455 |
+
"grad_norm": 2.4147472381591797,
|
| 456 |
+
"learning_rate": 0.0001995293977107475,
|
| 457 |
+
"loss": 0.4218,
|
| 458 |
"step": 64
|
| 459 |
},
|
| 460 |
{
|
| 461 |
+
"epoch": 3.2531645569620253,
|
| 462 |
+
"grad_norm": 3.543077230453491,
|
| 463 |
+
"learning_rate": 0.00019051620123934537,
|
| 464 |
+
"loss": 0.7115,
|
| 465 |
"step": 65
|
| 466 |
},
|
| 467 |
{
|
| 468 |
+
"epoch": 3.3037974683544302,
|
| 469 |
+
"grad_norm": 3.2347869873046875,
|
| 470 |
+
"learning_rate": 0.0001815842524819793,
|
| 471 |
+
"loss": 0.7353,
|
| 472 |
"step": 66
|
| 473 |
},
|
| 474 |
{
|
| 475 |
+
"epoch": 3.3544303797468356,
|
| 476 |
+
"grad_norm": 3.2037904262542725,
|
| 477 |
+
"learning_rate": 0.00017274575140626317,
|
| 478 |
+
"loss": 0.7768,
|
| 479 |
"step": 67
|
| 480 |
},
|
| 481 |
{
|
| 482 |
+
"epoch": 3.4050632911392404,
|
| 483 |
+
"grad_norm": 2.9766736030578613,
|
| 484 |
+
"learning_rate": 0.00016401277034151795,
|
| 485 |
+
"loss": 1.0119,
|
| 486 |
"step": 68
|
| 487 |
},
|
| 488 |
{
|
| 489 |
+
"epoch": 3.4556962025316453,
|
| 490 |
+
"grad_norm": 3.8264787197113037,
|
| 491 |
+
"learning_rate": 0.00015539723748942243,
|
| 492 |
+
"loss": 1.1998,
|
| 493 |
"step": 69
|
| 494 |
},
|
| 495 |
{
|
| 496 |
+
"epoch": 3.5063291139240507,
|
| 497 |
+
"grad_norm": 2.4378528594970703,
|
| 498 |
+
"learning_rate": 0.00014691092063152418,
|
| 499 |
+
"loss": 0.5007,
|
| 500 |
"step": 70
|
| 501 |
},
|
| 502 |
{
|
| 503 |
+
"epoch": 3.5569620253164556,
|
| 504 |
+
"grad_norm": 2.639159917831421,
|
| 505 |
+
"learning_rate": 0.00013856541105586545,
|
| 506 |
+
"loss": 0.5924,
|
| 507 |
"step": 71
|
| 508 |
},
|
| 509 |
{
|
| 510 |
+
"epoch": 3.607594936708861,
|
| 511 |
+
"grad_norm": 2.789632797241211,
|
| 512 |
+
"learning_rate": 0.0001303721077246784,
|
| 513 |
+
"loss": 0.5326,
|
| 514 |
"step": 72
|
| 515 |
},
|
| 516 |
{
|
| 517 |
+
"epoch": 3.6582278481012658,
|
| 518 |
+
"grad_norm": 3.450500965118408,
|
| 519 |
+
"learning_rate": 0.0001223422017047733,
|
| 520 |
+
"loss": 0.6301,
|
| 521 |
"step": 73
|
| 522 |
},
|
| 523 |
{
|
| 524 |
+
"epoch": 3.708860759493671,
|
| 525 |
+
"grad_norm": 3.5389089584350586,
|
| 526 |
+
"learning_rate": 0.00011448666088188764,
|
| 527 |
+
"loss": 0.9008,
|
| 528 |
"step": 74
|
| 529 |
},
|
| 530 |
{
|
| 531 |
+
"epoch": 3.759493670886076,
|
| 532 |
+
"grad_norm": 3.7785134315490723,
|
| 533 |
+
"learning_rate": 0.00010681621497987371,
|
| 534 |
+
"loss": 0.5551,
|
| 535 |
"step": 75
|
| 536 |
},
|
| 537 |
{
|
| 538 |
+
"epoch": 3.810126582278481,
|
| 539 |
+
"grad_norm": 2.698173761367798,
|
| 540 |
+
"learning_rate": 9.934134090518593e-05,
|
| 541 |
+
"loss": 0.5115,
|
| 542 |
"step": 76
|
| 543 |
},
|
| 544 |
{
|
| 545 |
+
"epoch": 3.8607594936708862,
|
| 546 |
+
"grad_norm": 2.663907766342163,
|
| 547 |
+
"learning_rate": 9.207224843668733e-05,
|
| 548 |
+
"loss": 0.3971,
|
| 549 |
"step": 77
|
| 550 |
},
|
| 551 |
{
|
| 552 |
+
"epoch": 3.911392405063291,
|
| 553 |
+
"grad_norm": 2.5929627418518066,
|
| 554 |
+
"learning_rate": 8.50188662803194e-05,
|
| 555 |
+
"loss": 0.4122,
|
| 556 |
"step": 78
|
| 557 |
},
|
| 558 |
{
|
| 559 |
+
"epoch": 3.962025316455696,
|
| 560 |
+
"grad_norm": 3.3559207916259766,
|
| 561 |
+
"learning_rate": 7.819082850768433e-05,
|
| 562 |
+
"loss": 0.8821,
|
| 563 |
"step": 79
|
| 564 |
},
|
| 565 |
{
|
| 566 |
+
"epoch": 4.0,
|
| 567 |
+
"grad_norm": 2.1461448669433594,
|
| 568 |
+
"learning_rate": 7.159746139706194e-05,
|
| 569 |
+
"loss": 0.3809,
|
| 570 |
"step": 80
|
| 571 |
},
|
| 572 |
{
|
| 573 |
+
"epoch": 4.050632911392405,
|
| 574 |
+
"grad_norm": 1.290476679801941,
|
| 575 |
+
"learning_rate": 6.524777069483526e-05,
|
| 576 |
+
"loss": 0.1557,
|
| 577 |
"step": 81
|
| 578 |
},
|
| 579 |
{
|
| 580 |
+
"epoch": 4.10126582278481,
|
| 581 |
+
"grad_norm": 2.6128554344177246,
|
| 582 |
+
"learning_rate": 5.9150429314724254e-05,
|
| 583 |
+
"loss": 0.5096,
|
| 584 |
"step": 82
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 4.151898734177215,
|
| 588 |
+
"grad_norm": 1.4323923587799072,
|
| 589 |
+
"learning_rate": 5.3313765491629194e-05,
|
| 590 |
+
"loss": 0.1989,
|
| 591 |
"step": 83
|
| 592 |
},
|
| 593 |
{
|
| 594 |
+
"epoch": 4.2025316455696204,
|
| 595 |
+
"grad_norm": 1.4174575805664062,
|
| 596 |
+
"learning_rate": 4.7745751406263163e-05,
|
| 597 |
+
"loss": 0.1367,
|
| 598 |
"step": 84
|
| 599 |
},
|
| 600 |
{
|
| 601 |
+
"epoch": 4.253164556962025,
|
| 602 |
+
"grad_norm": 1.7367544174194336,
|
| 603 |
+
"learning_rate": 4.245399229611238e-05,
|
| 604 |
+
"loss": 0.3181,
|
| 605 |
"step": 85
|
| 606 |
},
|
| 607 |
{
|
| 608 |
+
"epoch": 4.30379746835443,
|
| 609 |
+
"grad_norm": 2.1157405376434326,
|
| 610 |
+
"learning_rate": 3.7445716067596506e-05,
|
| 611 |
+
"loss": 0.2611,
|
| 612 |
"step": 86
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 4.3544303797468356,
|
| 616 |
+
"grad_norm": 1.41007661819458,
|
| 617 |
+
"learning_rate": 3.2727763423617915e-05,
|
| 618 |
+
"loss": 0.1407,
|
| 619 |
"step": 87
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"epoch": 4.405063291139241,
|
| 623 |
+
"grad_norm": 1.5083343982696533,
|
| 624 |
+
"learning_rate": 2.8306578519984528e-05,
|
| 625 |
+
"loss": 0.1538,
|
| 626 |
"step": 88
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 4.455696202531645,
|
| 630 |
+
"grad_norm": 1.710657000541687,
|
| 631 |
+
"learning_rate": 2.4188200163467787e-05,
|
| 632 |
+
"loss": 0.1437,
|
| 633 |
"step": 89
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"epoch": 4.506329113924051,
|
| 637 |
+
"grad_norm": 2.136242628097534,
|
| 638 |
+
"learning_rate": 2.0378253563519245e-05,
|
| 639 |
+
"loss": 0.3269,
|
| 640 |
"step": 90
|
| 641 |
},
|
| 642 |
{
|
| 643 |
+
"epoch": 4.556962025316456,
|
| 644 |
+
"grad_norm": 1.920873999595642,
|
| 645 |
+
"learning_rate": 1.6881942648911074e-05,
|
| 646 |
+
"loss": 0.181,
|
| 647 |
"step": 91
|
| 648 |
},
|
| 649 |
{
|
| 650 |
+
"epoch": 4.6075949367088604,
|
| 651 |
+
"grad_norm": 2.06215500831604,
|
| 652 |
+
"learning_rate": 1.3704042959795133e-05,
|
| 653 |
+
"loss": 0.3806,
|
| 654 |
"step": 92
|
| 655 |
},
|
| 656 |
{
|
| 657 |
+
"epoch": 4.658227848101266,
|
| 658 |
+
"grad_norm": 1.393393635749817,
|
| 659 |
+
"learning_rate": 1.0848895124889818e-05,
|
| 660 |
+
"loss": 0.1904,
|
| 661 |
"step": 93
|
| 662 |
},
|
| 663 |
{
|
| 664 |
+
"epoch": 4.708860759493671,
|
| 665 |
+
"grad_norm": 1.8562750816345215,
|
| 666 |
+
"learning_rate": 8.320398932703144e-06,
|
| 667 |
+
"loss": 0.2278,
|
| 668 |
"step": 94
|
| 669 |
},
|
| 670 |
{
|
| 671 |
+
"epoch": 4.759493670886076,
|
| 672 |
+
"grad_norm": 1.4714932441711426,
|
| 673 |
+
"learning_rate": 6.12200800489085e-06,
|
| 674 |
+
"loss": 0.1514,
|
| 675 |
"step": 95
|
| 676 |
},
|
| 677 |
{
|
| 678 |
+
"epoch": 4.810126582278481,
|
| 679 |
+
"grad_norm": 1.8964180946350098,
|
| 680 |
+
"learning_rate": 4.256725079024554e-06,
|
| 681 |
+
"loss": 0.172,
|
| 682 |
"step": 96
|
| 683 |
},
|
| 684 |
{
|
| 685 |
+
"epoch": 4.860759493670886,
|
| 686 |
+
"grad_norm": 1.761176347732544,
|
| 687 |
+
"learning_rate": 2.7270979072135106e-06,
|
| 688 |
+
"loss": 0.273,
|
| 689 |
"step": 97
|
| 690 |
},
|
| 691 |
{
|
| 692 |
+
"epoch": 4.911392405063291,
|
| 693 |
+
"grad_norm": 1.7138495445251465,
|
| 694 |
+
"learning_rate": 1.5352157761815977e-06,
|
| 695 |
+
"loss": 0.1886,
|
| 696 |
"step": 98
|
| 697 |
},
|
| 698 |
{
|
| 699 |
+
"epoch": 4.962025316455696,
|
| 700 |
+
"grad_norm": 2.0607151985168457,
|
| 701 |
+
"learning_rate": 6.827066535529947e-07,
|
| 702 |
+
"loss": 0.3668,
|
| 703 |
"step": 99
|
| 704 |
},
|
| 705 |
{
|
| 706 |
+
"epoch": 5.0,
|
| 707 |
+
"grad_norm": 1.7871519327163696,
|
| 708 |
+
"learning_rate": 1.7073496424427348e-07,
|
| 709 |
+
"loss": 0.0795,
|
| 710 |
"step": 100
|
| 711 |
}
|
| 712 |
],
|
| 713 |
"logging_steps": 1,
|
| 714 |
+
"max_steps": 100,
|
| 715 |
"num_input_tokens_seen": 0,
|
| 716 |
+
"num_train_epochs": 5,
|
| 717 |
+
"save_steps": 10,
|
| 718 |
"stateful_callbacks": {
|
| 719 |
"TrainerControl": {
|
| 720 |
"args": {
|
|
|
|
| 722 |
"should_evaluate": false,
|
| 723 |
"should_log": false,
|
| 724 |
"should_save": true,
|
| 725 |
+
"should_training_stop": true
|
| 726 |
},
|
| 727 |
"attributes": {}
|
| 728 |
}
|
| 729 |
},
|
| 730 |
+
"total_flos": 2.957146343615693e+16,
|
| 731 |
"train_batch_size": 2,
|
| 732 |
"trial_name": null,
|
| 733 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 6033
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8836c073a978d1cd0236e05e757ad1e437f7979b9636472ff16fdc5c97d05961
|
| 3 |
size 6033
|