Upload LoRA adapter (Fixed README metadata)
Browse files- README.md +2 -2
- adapter_config.json +7 -7
- adapter_model.safetensors +2 -2
- all_experiments_details.json +420 -420
README.md
CHANGED
|
@@ -39,8 +39,8 @@ while intermediate reasoning (Chain-of-Thought) is masked.
|
|
| 39 |
- Method: QLoRA (4-bit)
|
| 40 |
- Max sequence length: 512
|
| 41 |
- Epochs: 1
|
| 42 |
-
- Learning rate: 1e-
|
| 43 |
-
- LoRA: r=
|
| 44 |
|
| 45 |
## Usage
|
| 46 |
|
|
|
|
| 39 |
- Method: QLoRA (4-bit)
|
| 40 |
- Max sequence length: 512
|
| 41 |
- Epochs: 1
|
| 42 |
+
- Learning rate: 1e-06
|
| 43 |
+
- LoRA: r=80, alpha=160
|
| 44 |
|
| 45 |
## Usage
|
| 46 |
|
adapter_config.json
CHANGED
|
@@ -20,7 +20,7 @@
|
|
| 20 |
"layers_pattern": null,
|
| 21 |
"layers_to_transform": null,
|
| 22 |
"loftq_config": {},
|
| 23 |
-
"lora_alpha":
|
| 24 |
"lora_bias": false,
|
| 25 |
"lora_dropout": 0.0,
|
| 26 |
"megatron_config": null,
|
|
@@ -29,17 +29,17 @@
|
|
| 29 |
"peft_type": "LORA",
|
| 30 |
"peft_version": "0.18.1",
|
| 31 |
"qalora_group_size": 16,
|
| 32 |
-
"r":
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
-
"up_proj",
|
| 37 |
"v_proj",
|
| 38 |
-
"
|
| 39 |
-
"gate_proj",
|
| 40 |
-
"k_proj",
|
| 41 |
"down_proj",
|
| 42 |
-
"q_proj"
|
|
|
|
|
|
|
|
|
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
| 45 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 20 |
"layers_pattern": null,
|
| 21 |
"layers_to_transform": null,
|
| 22 |
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 160,
|
| 24 |
"lora_bias": false,
|
| 25 |
"lora_dropout": 0.0,
|
| 26 |
"megatron_config": null,
|
|
|
|
| 29 |
"peft_type": "LORA",
|
| 30 |
"peft_version": "0.18.1",
|
| 31 |
"qalora_group_size": 16,
|
| 32 |
+
"r": 80,
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
|
|
|
| 36 |
"v_proj",
|
| 37 |
+
"up_proj",
|
|
|
|
|
|
|
| 38 |
"down_proj",
|
| 39 |
+
"q_proj",
|
| 40 |
+
"k_proj",
|
| 41 |
+
"gate_proj",
|
| 42 |
+
"o_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
| 45 |
"task_type": "CAUSAL_LM",
|
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 sha256:32d951a8a18e0ae23940e93375640002eeb176d147e4512c2933c571edd1c141
|
| 3 |
+
size 660670872
|
all_experiments_details.json
CHANGED
|
@@ -13,8 +13,8 @@
|
|
| 13 |
"SEED": 3407,
|
| 14 |
"VAL_RATIO": 0.05,
|
| 15 |
"MAX_SEQ_LEN": 512,
|
| 16 |
-
"LORA_R":
|
| 17 |
-
"LORA_ALPHA":
|
| 18 |
"LORA_DROPOUT": 0.0,
|
| 19 |
"LORA_TARGET_MODULES": [
|
| 20 |
"q_proj",
|
|
@@ -29,7 +29,7 @@
|
|
| 29 |
"PER_DEVICE_TRAIN_BS": 2,
|
| 30 |
"PER_DEVICE_EVAL_BS": 2,
|
| 31 |
"GRAD_ACCUM": 8,
|
| 32 |
-
"LR":
|
| 33 |
"WARMUP_RATIO": 0.1,
|
| 34 |
"WEIGHT_DECAY": 0.05,
|
| 35 |
"MAX_STEPS": -1,
|
|
@@ -53,945 +53,945 @@
|
|
| 53 |
},
|
| 54 |
"history": [
|
| 55 |
{
|
| 56 |
-
"loss": 1.
|
| 57 |
-
"grad_norm":
|
| 58 |
-
"learning_rate": 8.
|
| 59 |
"epoch": 0.009169054441260744,
|
| 60 |
"step": 10
|
| 61 |
},
|
| 62 |
{
|
| 63 |
-
"loss": 1.
|
| 64 |
-
"grad_norm":
|
| 65 |
-
"learning_rate": 1.
|
| 66 |
"epoch": 0.01833810888252149,
|
| 67 |
"step": 20
|
| 68 |
},
|
| 69 |
{
|
| 70 |
-
"loss": 1.
|
| 71 |
-
"grad_norm":
|
| 72 |
-
"learning_rate": 2.636363636363636e-
|
| 73 |
"epoch": 0.027507163323782235,
|
| 74 |
"step": 30
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"loss": 1.
|
| 78 |
-
"grad_norm":
|
| 79 |
-
"learning_rate": 3.
|
| 80 |
"epoch": 0.03667621776504298,
|
| 81 |
"step": 40
|
| 82 |
},
|
| 83 |
{
|
| 84 |
-
"loss":
|
| 85 |
-
"grad_norm":
|
| 86 |
-
"learning_rate": 4.
|
| 87 |
"epoch": 0.045845272206303724,
|
| 88 |
"step": 50
|
| 89 |
},
|
| 90 |
{
|
| 91 |
-
"eval_loss": 1.
|
| 92 |
-
"eval_runtime":
|
| 93 |
-
"eval_samples_per_second": 5.
|
| 94 |
-
"eval_steps_per_second": 2.
|
| 95 |
"epoch": 0.045845272206303724,
|
| 96 |
"step": 50
|
| 97 |
},
|
| 98 |
{
|
| 99 |
-
"loss": 1.
|
| 100 |
-
"grad_norm":
|
| 101 |
-
"learning_rate": 5.
|
| 102 |
"epoch": 0.05501432664756447,
|
| 103 |
"step": 60
|
| 104 |
},
|
| 105 |
{
|
| 106 |
-
"loss": 1.
|
| 107 |
-
"grad_norm":
|
| 108 |
-
"learning_rate": 6.272727272727273e-
|
| 109 |
"epoch": 0.06418338108882521,
|
| 110 |
"step": 70
|
| 111 |
},
|
| 112 |
{
|
| 113 |
-
"loss": 1.
|
| 114 |
-
"grad_norm":
|
| 115 |
-
"learning_rate": 7.
|
| 116 |
"epoch": 0.07335243553008595,
|
| 117 |
"step": 80
|
| 118 |
},
|
| 119 |
{
|
| 120 |
-
"loss":
|
| 121 |
-
"grad_norm":
|
| 122 |
-
"learning_rate": 8.
|
| 123 |
"epoch": 0.0825214899713467,
|
| 124 |
"step": 90
|
| 125 |
},
|
| 126 |
{
|
| 127 |
-
"loss":
|
| 128 |
-
"grad_norm":
|
| 129 |
-
"learning_rate": 9e-
|
| 130 |
"epoch": 0.09169054441260745,
|
| 131 |
"step": 100
|
| 132 |
},
|
| 133 |
{
|
| 134 |
-
"eval_loss":
|
| 135 |
-
"eval_runtime":
|
| 136 |
-
"eval_samples_per_second":
|
| 137 |
-
"eval_steps_per_second":
|
| 138 |
"epoch": 0.09169054441260745,
|
| 139 |
"step": 100
|
| 140 |
},
|
| 141 |
{
|
| 142 |
-
"loss":
|
| 143 |
-
"grad_norm":
|
| 144 |
-
"learning_rate": 9.
|
| 145 |
"epoch": 0.1008595988538682,
|
| 146 |
"step": 110
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"loss":
|
| 150 |
-
"grad_norm":
|
| 151 |
-
"learning_rate": 9.
|
| 152 |
"epoch": 0.11002865329512894,
|
| 153 |
"step": 120
|
| 154 |
},
|
| 155 |
{
|
| 156 |
-
"loss":
|
| 157 |
-
"grad_norm":
|
| 158 |
-
"learning_rate": 9.990747162241872e-
|
| 159 |
"epoch": 0.11919770773638969,
|
| 160 |
"step": 130
|
| 161 |
},
|
| 162 |
{
|
| 163 |
-
"loss":
|
| 164 |
-
"grad_norm":
|
| 165 |
-
"learning_rate": 9.
|
| 166 |
"epoch": 0.12836676217765042,
|
| 167 |
"step": 140
|
| 168 |
},
|
| 169 |
{
|
| 170 |
-
"loss":
|
| 171 |
-
"grad_norm": 0.
|
| 172 |
-
"learning_rate": 9.96105368780285e-
|
| 173 |
"epoch": 0.13753581661891118,
|
| 174 |
"step": 150
|
| 175 |
},
|
| 176 |
{
|
| 177 |
-
"eval_loss":
|
| 178 |
-
"eval_runtime":
|
| 179 |
-
"eval_samples_per_second":
|
| 180 |
-
"eval_steps_per_second":
|
| 181 |
"epoch": 0.13753581661891118,
|
| 182 |
"step": 150
|
| 183 |
},
|
| 184 |
{
|
| 185 |
-
"loss":
|
| 186 |
-
"grad_norm": 0.
|
| 187 |
-
"learning_rate": 9.938566882624436e-
|
| 188 |
"epoch": 0.1467048710601719,
|
| 189 |
"step": 160
|
| 190 |
},
|
| 191 |
{
|
| 192 |
-
"loss":
|
| 193 |
-
"grad_norm": 0.
|
| 194 |
-
"learning_rate": 9.
|
| 195 |
"epoch": 0.15587392550143267,
|
| 196 |
"step": 170
|
| 197 |
},
|
| 198 |
{
|
| 199 |
-
"loss":
|
| 200 |
-
"grad_norm": 0.
|
| 201 |
-
"learning_rate": 9.
|
| 202 |
"epoch": 0.1650429799426934,
|
| 203 |
"step": 180
|
| 204 |
},
|
| 205 |
{
|
| 206 |
-
"loss":
|
| 207 |
-
"grad_norm": 0.
|
| 208 |
-
"learning_rate": 9.
|
| 209 |
"epoch": 0.17421203438395416,
|
| 210 |
"step": 190
|
| 211 |
},
|
| 212 |
{
|
| 213 |
-
"loss":
|
| 214 |
-
"grad_norm": 0.
|
| 215 |
-
"learning_rate": 9.
|
| 216 |
"epoch": 0.1833810888252149,
|
| 217 |
"step": 200
|
| 218 |
},
|
| 219 |
{
|
| 220 |
-
"eval_loss":
|
| 221 |
-
"eval_runtime":
|
| 222 |
-
"eval_samples_per_second":
|
| 223 |
-
"eval_steps_per_second":
|
| 224 |
"epoch": 0.1833810888252149,
|
| 225 |
"step": 200
|
| 226 |
},
|
| 227 |
{
|
| 228 |
-
"loss":
|
| 229 |
-
"grad_norm": 0.
|
| 230 |
-
"learning_rate": 9.
|
| 231 |
"epoch": 0.19255014326647565,
|
| 232 |
"step": 210
|
| 233 |
},
|
| 234 |
{
|
| 235 |
-
"loss":
|
| 236 |
-
"grad_norm": 0.
|
| 237 |
-
"learning_rate": 9.
|
| 238 |
"epoch": 0.2017191977077364,
|
| 239 |
"step": 220
|
| 240 |
},
|
| 241 |
{
|
| 242 |
-
"loss":
|
| 243 |
-
"grad_norm": 0.
|
| 244 |
-
"learning_rate": 9.641298456400363e-
|
| 245 |
"epoch": 0.21088825214899715,
|
| 246 |
"step": 230
|
| 247 |
},
|
| 248 |
{
|
| 249 |
-
"loss":
|
| 250 |
-
"grad_norm": 0.
|
| 251 |
-
"learning_rate": 9.
|
| 252 |
"epoch": 0.22005730659025788,
|
| 253 |
"step": 240
|
| 254 |
},
|
| 255 |
{
|
| 256 |
-
"loss":
|
| 257 |
-
"grad_norm": 0.
|
| 258 |
-
"learning_rate": 9.
|
| 259 |
"epoch": 0.22922636103151864,
|
| 260 |
"step": 250
|
| 261 |
},
|
| 262 |
{
|
| 263 |
-
"eval_loss":
|
| 264 |
-
"eval_runtime":
|
| 265 |
-
"eval_samples_per_second":
|
| 266 |
-
"eval_steps_per_second":
|
| 267 |
"epoch": 0.22922636103151864,
|
| 268 |
"step": 250
|
| 269 |
},
|
| 270 |
{
|
| 271 |
-
"loss":
|
| 272 |
-
"grad_norm": 0.
|
| 273 |
-
"learning_rate": 9.
|
| 274 |
"epoch": 0.23839541547277937,
|
| 275 |
"step": 260
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"loss":
|
| 279 |
-
"grad_norm": 0.
|
| 280 |
-
"learning_rate": 9.
|
| 281 |
"epoch": 0.2475644699140401,
|
| 282 |
"step": 270
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"loss":
|
| 286 |
-
"grad_norm": 0.
|
| 287 |
-
"learning_rate": 9.
|
| 288 |
"epoch": 0.25673352435530084,
|
| 289 |
"step": 280
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"loss":
|
| 293 |
-
"grad_norm": 0.
|
| 294 |
-
"learning_rate": 9.20075003133193e-
|
| 295 |
"epoch": 0.2659025787965616,
|
| 296 |
"step": 290
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"loss":
|
| 300 |
-
"grad_norm": 0.
|
| 301 |
-
"learning_rate": 9.
|
| 302 |
"epoch": 0.27507163323782235,
|
| 303 |
"step": 300
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"eval_loss":
|
| 307 |
-
"eval_runtime":
|
| 308 |
-
"eval_samples_per_second": 5.
|
| 309 |
-
"eval_steps_per_second": 2.
|
| 310 |
"epoch": 0.27507163323782235,
|
| 311 |
"step": 300
|
| 312 |
},
|
| 313 |
{
|
| 314 |
-
"loss":
|
| 315 |
-
"grad_norm": 0.
|
| 316 |
-
"learning_rate": 9.018569855778383e-
|
| 317 |
"epoch": 0.2842406876790831,
|
| 318 |
"step": 310
|
| 319 |
},
|
| 320 |
{
|
| 321 |
-
"loss":
|
| 322 |
-
"grad_norm": 0.
|
| 323 |
-
"learning_rate": 8.
|
| 324 |
"epoch": 0.2934097421203438,
|
| 325 |
"step": 320
|
| 326 |
},
|
| 327 |
{
|
| 328 |
-
"loss":
|
| 329 |
-
"grad_norm": 0.
|
| 330 |
-
"learning_rate": 8.
|
| 331 |
"epoch": 0.3025787965616046,
|
| 332 |
"step": 330
|
| 333 |
},
|
| 334 |
{
|
| 335 |
-
"loss":
|
| 336 |
-
"grad_norm": 0.
|
| 337 |
-
"learning_rate": 8.
|
| 338 |
"epoch": 0.31174785100286534,
|
| 339 |
"step": 340
|
| 340 |
},
|
| 341 |
{
|
| 342 |
-
"loss":
|
| 343 |
-
"grad_norm": 0.
|
| 344 |
-
"learning_rate": 8.
|
| 345 |
"epoch": 0.3209169054441261,
|
| 346 |
"step": 350
|
| 347 |
},
|
| 348 |
{
|
| 349 |
-
"eval_loss":
|
| 350 |
-
"eval_runtime":
|
| 351 |
-
"eval_samples_per_second": 5.
|
| 352 |
-
"eval_steps_per_second": 2.
|
| 353 |
"epoch": 0.3209169054441261,
|
| 354 |
"step": 350
|
| 355 |
},
|
| 356 |
{
|
| 357 |
-
"loss":
|
| 358 |
-
"grad_norm": 0.
|
| 359 |
-
"learning_rate": 8.
|
| 360 |
"epoch": 0.3300859598853868,
|
| 361 |
"step": 360
|
| 362 |
},
|
| 363 |
{
|
| 364 |
-
"loss":
|
| 365 |
-
"grad_norm":
|
| 366 |
-
"learning_rate": 8.376475088911317e-
|
| 367 |
"epoch": 0.33925501432664756,
|
| 368 |
"step": 370
|
| 369 |
},
|
| 370 |
{
|
| 371 |
-
"loss":
|
| 372 |
-
"grad_norm": 0.
|
| 373 |
-
"learning_rate": 8.
|
| 374 |
"epoch": 0.3484240687679083,
|
| 375 |
"step": 380
|
| 376 |
},
|
| 377 |
{
|
| 378 |
-
"loss":
|
| 379 |
-
"grad_norm": 0.
|
| 380 |
-
"learning_rate": 8.133518187573862e-
|
| 381 |
"epoch": 0.35759312320916903,
|
| 382 |
"step": 390
|
| 383 |
},
|
| 384 |
{
|
| 385 |
-
"loss":
|
| 386 |
-
"grad_norm": 0.
|
| 387 |
-
"learning_rate": 8.007156582535131e-
|
| 388 |
"epoch": 0.3667621776504298,
|
| 389 |
"step": 400
|
| 390 |
},
|
| 391 |
{
|
| 392 |
-
"eval_loss":
|
| 393 |
-
"eval_runtime":
|
| 394 |
-
"eval_samples_per_second":
|
| 395 |
-
"eval_steps_per_second":
|
| 396 |
"epoch": 0.3667621776504298,
|
| 397 |
"step": 400
|
| 398 |
},
|
| 399 |
{
|
| 400 |
-
"loss":
|
| 401 |
-
"grad_norm": 0.
|
| 402 |
-
"learning_rate": 7.877711216888867e-
|
| 403 |
"epoch": 0.37593123209169055,
|
| 404 |
"step": 410
|
| 405 |
},
|
| 406 |
{
|
| 407 |
-
"loss":
|
| 408 |
-
"grad_norm": 0.
|
| 409 |
-
"learning_rate": 7.745314833479833e-
|
| 410 |
"epoch": 0.3851002865329513,
|
| 411 |
"step": 420
|
| 412 |
},
|
| 413 |
{
|
| 414 |
-
"loss":
|
| 415 |
-
"grad_norm": 0.
|
| 416 |
-
"learning_rate": 7.6101032013445e-
|
| 417 |
"epoch": 0.394269340974212,
|
| 418 |
"step": 430
|
| 419 |
},
|
| 420 |
{
|
| 421 |
-
"loss":
|
| 422 |
-
"grad_norm": 0.
|
| 423 |
-
"learning_rate": 7.
|
| 424 |
"epoch": 0.4034383954154728,
|
| 425 |
"step": 440
|
| 426 |
},
|
| 427 |
{
|
| 428 |
-
"loss":
|
| 429 |
-
"grad_norm": 0.
|
| 430 |
-
"learning_rate": 7.
|
| 431 |
"epoch": 0.41260744985673353,
|
| 432 |
"step": 450
|
| 433 |
},
|
| 434 |
{
|
| 435 |
-
"eval_loss":
|
| 436 |
-
"eval_runtime":
|
| 437 |
-
"eval_samples_per_second": 6.
|
| 438 |
-
"eval_steps_per_second": 3.
|
| 439 |
"epoch": 0.41260744985673353,
|
| 440 |
"step": 450
|
| 441 |
},
|
| 442 |
{
|
| 443 |
-
"loss":
|
| 444 |
-
"grad_norm": 0.
|
| 445 |
-
"learning_rate": 7.188976951463723e-
|
| 446 |
"epoch": 0.4217765042979943,
|
| 447 |
"step": 460
|
| 448 |
},
|
| 449 |
{
|
| 450 |
-
"loss":
|
| 451 |
-
"grad_norm": 0.
|
| 452 |
-
"learning_rate": 7.043917604508971e-
|
| 453 |
"epoch": 0.430945558739255,
|
| 454 |
"step": 470
|
| 455 |
},
|
| 456 |
{
|
| 457 |
-
"loss":
|
| 458 |
-
"grad_norm": 0.
|
| 459 |
-
"learning_rate": 6.896762273384178e-
|
| 460 |
"epoch": 0.44011461318051576,
|
| 461 |
"step": 480
|
| 462 |
},
|
| 463 |
{
|
| 464 |
-
"loss":
|
| 465 |
-
"grad_norm": 0.
|
| 466 |
-
"learning_rate": 6.747661862041585e-
|
| 467 |
"epoch": 0.4492836676217765,
|
| 468 |
"step": 490
|
| 469 |
},
|
| 470 |
{
|
| 471 |
-
"loss":
|
| 472 |
-
"grad_norm": 0.
|
| 473 |
-
"learning_rate": 6.
|
| 474 |
"epoch": 0.4584527220630373,
|
| 475 |
"step": 500
|
| 476 |
},
|
| 477 |
{
|
| 478 |
-
"eval_loss":
|
| 479 |
-
"eval_runtime":
|
| 480 |
-
"eval_samples_per_second": 6.
|
| 481 |
-
"eval_steps_per_second": 3.
|
| 482 |
"epoch": 0.4584527220630373,
|
| 483 |
"step": 500
|
| 484 |
},
|
| 485 |
{
|
| 486 |
-
"loss":
|
| 487 |
-
"grad_norm": 0.
|
| 488 |
-
"learning_rate": 6.444239230863504e-
|
| 489 |
"epoch": 0.467621776504298,
|
| 490 |
"step": 510
|
| 491 |
},
|
| 492 |
{
|
| 493 |
-
"loss":
|
| 494 |
-
"grad_norm": 0.
|
| 495 |
-
"learning_rate": 6.
|
| 496 |
"epoch": 0.47679083094555874,
|
| 497 |
"step": 520
|
| 498 |
},
|
| 499 |
{
|
| 500 |
-
"loss":
|
| 501 |
-
"grad_norm": 0.
|
| 502 |
-
"learning_rate": 6.
|
| 503 |
"epoch": 0.4859598853868195,
|
| 504 |
"step": 530
|
| 505 |
},
|
| 506 |
{
|
| 507 |
-
"loss":
|
| 508 |
-
"grad_norm": 0.
|
| 509 |
-
"learning_rate": 5.
|
| 510 |
"epoch": 0.4951289398280802,
|
| 511 |
"step": 540
|
| 512 |
},
|
| 513 |
{
|
| 514 |
-
"loss":
|
| 515 |
-
"grad_norm": 0.
|
| 516 |
-
"learning_rate": 5.820894745574025e-
|
| 517 |
"epoch": 0.504297994269341,
|
| 518 |
"step": 550
|
| 519 |
},
|
| 520 |
{
|
| 521 |
-
"eval_loss":
|
| 522 |
-
"eval_runtime":
|
| 523 |
-
"eval_samples_per_second": 6.
|
| 524 |
-
"eval_steps_per_second": 3.
|
| 525 |
"epoch": 0.504297994269341,
|
| 526 |
"step": 550
|
| 527 |
},
|
| 528 |
{
|
| 529 |
-
"loss":
|
| 530 |
-
"grad_norm": 0.
|
| 531 |
-
"learning_rate": 5.
|
| 532 |
"epoch": 0.5134670487106017,
|
| 533 |
"step": 560
|
| 534 |
},
|
| 535 |
{
|
| 536 |
-
"loss":
|
| 537 |
-
"grad_norm": 0.
|
| 538 |
-
"learning_rate": 5.503529129972792e-
|
| 539 |
"epoch": 0.5226361031518625,
|
| 540 |
"step": 570
|
| 541 |
},
|
| 542 |
{
|
| 543 |
-
"loss":
|
| 544 |
-
"grad_norm": 0.
|
| 545 |
-
"learning_rate": 5.34399025147273e-
|
| 546 |
"epoch": 0.5318051575931232,
|
| 547 |
"step": 580
|
| 548 |
},
|
| 549 |
{
|
| 550 |
-
"loss":
|
| 551 |
-
"grad_norm": 0.
|
| 552 |
-
"learning_rate": 5.
|
| 553 |
"epoch": 0.540974212034384,
|
| 554 |
"step": 590
|
| 555 |
},
|
| 556 |
{
|
| 557 |
-
"loss":
|
| 558 |
-
"grad_norm": 0.
|
| 559 |
-
"learning_rate": 5.
|
| 560 |
"epoch": 0.5501432664756447,
|
| 561 |
"step": 600
|
| 562 |
},
|
| 563 |
{
|
| 564 |
-
"eval_loss":
|
| 565 |
-
"eval_runtime":
|
| 566 |
-
"eval_samples_per_second":
|
| 567 |
-
"eval_steps_per_second":
|
| 568 |
"epoch": 0.5501432664756447,
|
| 569 |
"step": 600
|
| 570 |
},
|
| 571 |
{
|
| 572 |
-
"loss":
|
| 573 |
-
"grad_norm": 0.
|
| 574 |
-
"learning_rate": 4.
|
| 575 |
"epoch": 0.5593123209169054,
|
| 576 |
"step": 610
|
| 577 |
},
|
| 578 |
{
|
| 579 |
-
"loss":
|
| 580 |
-
"grad_norm": 0.
|
| 581 |
-
"learning_rate": 4.703947653810575e-
|
| 582 |
"epoch": 0.5684813753581662,
|
| 583 |
"step": 620
|
| 584 |
},
|
| 585 |
{
|
| 586 |
-
"loss":
|
| 587 |
-
"grad_norm": 0.
|
| 588 |
-
"learning_rate": 4.544285751384584e-
|
| 589 |
"epoch": 0.5776504297994269,
|
| 590 |
"step": 630
|
| 591 |
},
|
| 592 |
{
|
| 593 |
-
"loss":
|
| 594 |
-
"grad_norm": 0.
|
| 595 |
-
"learning_rate": 4.
|
| 596 |
"epoch": 0.5868194842406876,
|
| 597 |
"step": 640
|
| 598 |
},
|
| 599 |
{
|
| 600 |
-
"loss":
|
| 601 |
-
"grad_norm": 0.
|
| 602 |
-
"learning_rate": 4.
|
| 603 |
"epoch": 0.5959885386819485,
|
| 604 |
"step": 650
|
| 605 |
},
|
| 606 |
{
|
| 607 |
-
"eval_loss":
|
| 608 |
-
"eval_runtime":
|
| 609 |
-
"eval_samples_per_second":
|
| 610 |
-
"eval_steps_per_second":
|
| 611 |
"epoch": 0.5959885386819485,
|
| 612 |
"step": 650
|
| 613 |
},
|
| 614 |
{
|
| 615 |
-
"loss":
|
| 616 |
-
"grad_norm":
|
| 617 |
-
"learning_rate": 4.
|
| 618 |
"epoch": 0.6051575931232092,
|
| 619 |
"step": 660
|
| 620 |
},
|
| 621 |
{
|
| 622 |
-
"loss":
|
| 623 |
-
"grad_norm": 0.
|
| 624 |
-
"learning_rate": 3.
|
| 625 |
"epoch": 0.6143266475644699,
|
| 626 |
"step": 670
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"loss":
|
| 630 |
-
"grad_norm": 0.
|
| 631 |
-
"learning_rate": 3.
|
| 632 |
"epoch": 0.6234957020057307,
|
| 633 |
"step": 680
|
| 634 |
},
|
| 635 |
{
|
| 636 |
-
"loss":
|
| 637 |
-
"grad_norm": 0.
|
| 638 |
-
"learning_rate": 3.
|
| 639 |
"epoch": 0.6326647564469914,
|
| 640 |
"step": 690
|
| 641 |
},
|
| 642 |
{
|
| 643 |
-
"loss":
|
| 644 |
-
"grad_norm": 0.
|
| 645 |
-
"learning_rate": 3.448824896507292e-
|
| 646 |
"epoch": 0.6418338108882522,
|
| 647 |
"step": 700
|
| 648 |
},
|
| 649 |
{
|
| 650 |
-
"eval_loss":
|
| 651 |
-
"eval_runtime":
|
| 652 |
-
"eval_samples_per_second":
|
| 653 |
-
"eval_steps_per_second":
|
| 654 |
"epoch": 0.6418338108882522,
|
| 655 |
"step": 700
|
| 656 |
},
|
| 657 |
{
|
| 658 |
-
"loss":
|
| 659 |
-
"grad_norm": 0.
|
| 660 |
-
"learning_rate": 3.
|
| 661 |
"epoch": 0.6510028653295129,
|
| 662 |
"step": 710
|
| 663 |
},
|
| 664 |
{
|
| 665 |
-
"loss":
|
| 666 |
-
"grad_norm": 0.
|
| 667 |
-
"learning_rate": 3.
|
| 668 |
"epoch": 0.6601719197707736,
|
| 669 |
"step": 720
|
| 670 |
},
|
| 671 |
{
|
| 672 |
-
"loss":
|
| 673 |
-
"grad_norm": 0.
|
| 674 |
-
"learning_rate": 3.
|
| 675 |
"epoch": 0.6693409742120344,
|
| 676 |
"step": 730
|
| 677 |
},
|
| 678 |
{
|
| 679 |
-
"loss":
|
| 680 |
-
"grad_norm": 0.
|
| 681 |
-
"learning_rate": 2.
|
| 682 |
"epoch": 0.6785100286532951,
|
| 683 |
"step": 740
|
| 684 |
},
|
| 685 |
{
|
| 686 |
-
"loss":
|
| 687 |
-
"grad_norm": 0.
|
| 688 |
-
"learning_rate": 2.710808365197e-
|
| 689 |
"epoch": 0.6876790830945558,
|
| 690 |
"step": 750
|
| 691 |
},
|
| 692 |
{
|
| 693 |
-
"eval_loss":
|
| 694 |
-
"eval_runtime":
|
| 695 |
-
"eval_samples_per_second":
|
| 696 |
-
"eval_steps_per_second":
|
| 697 |
"epoch": 0.6876790830945558,
|
| 698 |
"step": 750
|
| 699 |
},
|
| 700 |
{
|
| 701 |
-
"loss":
|
| 702 |
-
"grad_norm": 0.
|
| 703 |
-
"learning_rate": 2.
|
| 704 |
"epoch": 0.6968481375358166,
|
| 705 |
"step": 760
|
| 706 |
},
|
| 707 |
{
|
| 708 |
-
"loss":
|
| 709 |
-
"grad_norm": 0.
|
| 710 |
-
"learning_rate": 2.
|
| 711 |
"epoch": 0.7060171919770774,
|
| 712 |
"step": 770
|
| 713 |
},
|
| 714 |
{
|
| 715 |
-
"loss":
|
| 716 |
-
"grad_norm": 0.
|
| 717 |
-
"learning_rate": 2.
|
| 718 |
"epoch": 0.7151862464183381,
|
| 719 |
"step": 780
|
| 720 |
},
|
| 721 |
{
|
| 722 |
-
"loss":
|
| 723 |
-
"grad_norm": 0.
|
| 724 |
-
"learning_rate": 2.
|
| 725 |
"epoch": 0.7243553008595989,
|
| 726 |
"step": 790
|
| 727 |
},
|
| 728 |
{
|
| 729 |
-
"loss":
|
| 730 |
-
"grad_norm": 0.
|
| 731 |
-
"learning_rate": 2.
|
| 732 |
"epoch": 0.7335243553008596,
|
| 733 |
"step": 800
|
| 734 |
},
|
| 735 |
{
|
| 736 |
-
"eval_loss":
|
| 737 |
-
"eval_runtime":
|
| 738 |
-
"eval_samples_per_second":
|
| 739 |
-
"eval_steps_per_second":
|
| 740 |
"epoch": 0.7335243553008596,
|
| 741 |
"step": 800
|
| 742 |
},
|
| 743 |
{
|
| 744 |
-
"loss":
|
| 745 |
-
"grad_norm": 0.
|
| 746 |
-
"learning_rate": 1.904058726480367e-
|
| 747 |
"epoch": 0.7426934097421204,
|
| 748 |
"step": 810
|
| 749 |
},
|
| 750 |
{
|
| 751 |
-
"loss": 0.
|
| 752 |
-
"grad_norm": 0.
|
| 753 |
-
"learning_rate": 1.
|
| 754 |
"epoch": 0.7518624641833811,
|
| 755 |
"step": 820
|
| 756 |
},
|
| 757 |
{
|
| 758 |
-
"loss":
|
| 759 |
-
"grad_norm": 0.
|
| 760 |
-
"learning_rate": 1.
|
| 761 |
"epoch": 0.7610315186246418,
|
| 762 |
"step": 830
|
| 763 |
},
|
| 764 |
{
|
| 765 |
-
"loss":
|
| 766 |
-
"grad_norm": 0.
|
| 767 |
-
"learning_rate": 1.
|
| 768 |
"epoch": 0.7702005730659026,
|
| 769 |
"step": 840
|
| 770 |
},
|
| 771 |
{
|
| 772 |
-
"loss":
|
| 773 |
-
"grad_norm":
|
| 774 |
-
"learning_rate": 1.
|
| 775 |
"epoch": 0.7793696275071633,
|
| 776 |
"step": 850
|
| 777 |
},
|
| 778 |
{
|
| 779 |
-
"eval_loss":
|
| 780 |
-
"eval_runtime":
|
| 781 |
-
"eval_samples_per_second": 5.
|
| 782 |
-
"eval_steps_per_second": 2.
|
| 783 |
"epoch": 0.7793696275071633,
|
| 784 |
"step": 850
|
| 785 |
},
|
| 786 |
{
|
| 787 |
-
"loss":
|
| 788 |
-
"grad_norm":
|
| 789 |
-
"learning_rate": 1.3176726040171e-
|
| 790 |
"epoch": 0.788538681948424,
|
| 791 |
"step": 860
|
| 792 |
},
|
| 793 |
{
|
| 794 |
-
"loss":
|
| 795 |
-
"grad_norm": 0.
|
| 796 |
-
"learning_rate": 1.
|
| 797 |
"epoch": 0.7977077363896848,
|
| 798 |
"step": 870
|
| 799 |
},
|
| 800 |
{
|
| 801 |
-
"loss":
|
| 802 |
-
"grad_norm": 0.
|
| 803 |
-
"learning_rate": 1.
|
| 804 |
"epoch": 0.8068767908309455,
|
| 805 |
"step": 880
|
| 806 |
},
|
| 807 |
{
|
| 808 |
-
"loss":
|
| 809 |
-
"grad_norm": 0.
|
| 810 |
-
"learning_rate": 1.
|
| 811 |
"epoch": 0.8160458452722062,
|
| 812 |
"step": 890
|
| 813 |
},
|
| 814 |
{
|
| 815 |
-
"loss":
|
| 816 |
-
"grad_norm": 0.
|
| 817 |
-
"learning_rate": 9.157528360620415e-
|
| 818 |
"epoch": 0.8252148997134671,
|
| 819 |
"step": 900
|
| 820 |
},
|
| 821 |
{
|
| 822 |
-
"eval_loss":
|
| 823 |
-
"eval_runtime":
|
| 824 |
-
"eval_samples_per_second":
|
| 825 |
-
"eval_steps_per_second":
|
| 826 |
"epoch": 0.8252148997134671,
|
| 827 |
"step": 900
|
| 828 |
},
|
| 829 |
{
|
| 830 |
-
"loss":
|
| 831 |
-
"grad_norm": 0.
|
| 832 |
-
"learning_rate": 8.
|
| 833 |
"epoch": 0.8343839541547278,
|
| 834 |
"step": 910
|
| 835 |
},
|
| 836 |
{
|
| 837 |
-
"loss":
|
| 838 |
-
"grad_norm": 0.
|
| 839 |
-
"learning_rate": 7.
|
| 840 |
"epoch": 0.8435530085959886,
|
| 841 |
"step": 920
|
| 842 |
},
|
| 843 |
{
|
| 844 |
-
"loss":
|
| 845 |
-
"grad_norm": 0.
|
| 846 |
-
"learning_rate": 6.
|
| 847 |
"epoch": 0.8527220630372493,
|
| 848 |
"step": 930
|
| 849 |
},
|
| 850 |
{
|
| 851 |
-
"loss":
|
| 852 |
-
"grad_norm": 0.
|
| 853 |
-
"learning_rate": 5.
|
| 854 |
"epoch": 0.86189111747851,
|
| 855 |
"step": 940
|
| 856 |
},
|
| 857 |
{
|
| 858 |
-
"loss":
|
| 859 |
-
"grad_norm": 0.
|
| 860 |
-
"learning_rate": 5.
|
| 861 |
"epoch": 0.8710601719197708,
|
| 862 |
"step": 950
|
| 863 |
},
|
| 864 |
{
|
| 865 |
-
"eval_loss":
|
| 866 |
-
"eval_runtime":
|
| 867 |
-
"eval_samples_per_second":
|
| 868 |
-
"eval_steps_per_second":
|
| 869 |
"epoch": 0.8710601719197708,
|
| 870 |
"step": 950
|
| 871 |
},
|
| 872 |
{
|
| 873 |
-
"loss":
|
| 874 |
-
"grad_norm": 0.
|
| 875 |
-
"learning_rate": 4.
|
| 876 |
"epoch": 0.8802292263610315,
|
| 877 |
"step": 960
|
| 878 |
},
|
| 879 |
{
|
| 880 |
-
"loss":
|
| 881 |
-
"grad_norm": 0.
|
| 882 |
-
"learning_rate": 3.
|
| 883 |
"epoch": 0.8893982808022922,
|
| 884 |
"step": 970
|
| 885 |
},
|
| 886 |
{
|
| 887 |
-
"loss":
|
| 888 |
-
"grad_norm": 0.
|
| 889 |
-
"learning_rate": 3.
|
| 890 |
"epoch": 0.898567335243553,
|
| 891 |
"step": 980
|
| 892 |
},
|
| 893 |
{
|
| 894 |
-
"loss":
|
| 895 |
-
"grad_norm": 0.
|
| 896 |
-
"learning_rate": 2.643851527335006e-
|
| 897 |
"epoch": 0.9077363896848137,
|
| 898 |
"step": 990
|
| 899 |
},
|
| 900 |
{
|
| 901 |
-
"loss":
|
| 902 |
-
"grad_norm": 0.
|
| 903 |
-
"learning_rate": 2.
|
| 904 |
"epoch": 0.9169054441260746,
|
| 905 |
"step": 1000
|
| 906 |
},
|
| 907 |
{
|
| 908 |
-
"eval_loss":
|
| 909 |
-
"eval_runtime":
|
| 910 |
-
"eval_samples_per_second":
|
| 911 |
-
"eval_steps_per_second":
|
| 912 |
"epoch": 0.9169054441260746,
|
| 913 |
"step": 1000
|
| 914 |
},
|
| 915 |
{
|
| 916 |
-
"loss":
|
| 917 |
-
"grad_norm": 0.
|
| 918 |
-
"learning_rate": 1.
|
| 919 |
"epoch": 0.9260744985673353,
|
| 920 |
"step": 1010
|
| 921 |
},
|
| 922 |
{
|
| 923 |
-
"loss":
|
| 924 |
-
"grad_norm": 0.
|
| 925 |
-
"learning_rate": 1.
|
| 926 |
"epoch": 0.935243553008596,
|
| 927 |
"step": 1020
|
| 928 |
},
|
| 929 |
{
|
| 930 |
-
"loss":
|
| 931 |
-
"grad_norm": 0.
|
| 932 |
-
"learning_rate": 9.82331110311857e-
|
| 933 |
"epoch": 0.9444126074498568,
|
| 934 |
"step": 1030
|
| 935 |
},
|
| 936 |
{
|
| 937 |
-
"loss":
|
| 938 |
-
"grad_norm": 0.
|
| 939 |
-
"learning_rate": 6.
|
| 940 |
"epoch": 0.9535816618911175,
|
| 941 |
"step": 1040
|
| 942 |
},
|
| 943 |
{
|
| 944 |
-
"loss":
|
| 945 |
-
"grad_norm": 0.
|
| 946 |
-
"learning_rate": 4.
|
| 947 |
"epoch": 0.9627507163323782,
|
| 948 |
"step": 1050
|
| 949 |
},
|
| 950 |
{
|
| 951 |
-
"eval_loss":
|
| 952 |
-
"eval_runtime":
|
| 953 |
-
"eval_samples_per_second": 5.
|
| 954 |
-
"eval_steps_per_second": 2.
|
| 955 |
"epoch": 0.9627507163323782,
|
| 956 |
"step": 1050
|
| 957 |
},
|
| 958 |
{
|
| 959 |
-
"loss":
|
| 960 |
-
"grad_norm": 0.
|
| 961 |
-
"learning_rate": 2.6231407347736546e-
|
| 962 |
"epoch": 0.971919770773639,
|
| 963 |
"step": 1060
|
| 964 |
},
|
| 965 |
{
|
| 966 |
-
"loss":
|
| 967 |
-
"grad_norm": 0.
|
| 968 |
-
"learning_rate": 1.
|
| 969 |
"epoch": 0.9810888252148997,
|
| 970 |
"step": 1070
|
| 971 |
},
|
| 972 |
{
|
| 973 |
-
"loss":
|
| 974 |
-
"grad_norm": 0.
|
| 975 |
-
"learning_rate": 3.
|
| 976 |
"epoch": 0.9902578796561604,
|
| 977 |
"step": 1080
|
| 978 |
},
|
| 979 |
{
|
| 980 |
-
"loss":
|
| 981 |
-
"grad_norm": 0.
|
| 982 |
-
"learning_rate": 1.
|
| 983 |
"epoch": 0.9994269340974212,
|
| 984 |
"step": 1090
|
| 985 |
},
|
| 986 |
{
|
| 987 |
-
"train_runtime":
|
| 988 |
-
"train_samples_per_second": 1.
|
| 989 |
-
"train_steps_per_second": 0.
|
| 990 |
-
"total_flos": 1.
|
| 991 |
-
"train_loss":
|
| 992 |
"epoch": 1.0,
|
| 993 |
"step": 1091,
|
| 994 |
-
"total_runtime_sec":
|
| 995 |
}
|
| 996 |
]
|
| 997 |
}
|
|
|
|
| 13 |
"SEED": 3407,
|
| 14 |
"VAL_RATIO": 0.05,
|
| 15 |
"MAX_SEQ_LEN": 512,
|
| 16 |
+
"LORA_R": 80,
|
| 17 |
+
"LORA_ALPHA": 160,
|
| 18 |
"LORA_DROPOUT": 0.0,
|
| 19 |
"LORA_TARGET_MODULES": [
|
| 20 |
"q_proj",
|
|
|
|
| 29 |
"PER_DEVICE_TRAIN_BS": 2,
|
| 30 |
"PER_DEVICE_EVAL_BS": 2,
|
| 31 |
"GRAD_ACCUM": 8,
|
| 32 |
+
"LR": 1e-06,
|
| 33 |
"WARMUP_RATIO": 0.1,
|
| 34 |
"WEIGHT_DECAY": 0.05,
|
| 35 |
"MAX_STEPS": -1,
|
|
|
|
| 53 |
},
|
| 54 |
"history": [
|
| 55 |
{
|
| 56 |
+
"loss": 1.9238,
|
| 57 |
+
"grad_norm": 5.567883491516113,
|
| 58 |
+
"learning_rate": 8.181818181818182e-08,
|
| 59 |
"epoch": 0.009169054441260744,
|
| 60 |
"step": 10
|
| 61 |
},
|
| 62 |
{
|
| 63 |
+
"loss": 1.7942,
|
| 64 |
+
"grad_norm": 3.81577467918396,
|
| 65 |
+
"learning_rate": 1.7272727272727272e-07,
|
| 66 |
"epoch": 0.01833810888252149,
|
| 67 |
"step": 20
|
| 68 |
},
|
| 69 |
{
|
| 70 |
+
"loss": 1.9151,
|
| 71 |
+
"grad_norm": 4.943554878234863,
|
| 72 |
+
"learning_rate": 2.636363636363636e-07,
|
| 73 |
"epoch": 0.027507163323782235,
|
| 74 |
"step": 30
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"loss": 1.7679,
|
| 78 |
+
"grad_norm": 4.759264945983887,
|
| 79 |
+
"learning_rate": 3.545454545454545e-07,
|
| 80 |
"epoch": 0.03667621776504298,
|
| 81 |
"step": 40
|
| 82 |
},
|
| 83 |
{
|
| 84 |
+
"loss": 2.0977,
|
| 85 |
+
"grad_norm": 6.200092315673828,
|
| 86 |
+
"learning_rate": 4.4545454545454544e-07,
|
| 87 |
"epoch": 0.045845272206303724,
|
| 88 |
"step": 50
|
| 89 |
},
|
| 90 |
{
|
| 91 |
+
"eval_loss": 1.8969855308532715,
|
| 92 |
+
"eval_runtime": 165.3674,
|
| 93 |
+
"eval_samples_per_second": 5.612,
|
| 94 |
+
"eval_steps_per_second": 2.806,
|
| 95 |
"epoch": 0.045845272206303724,
|
| 96 |
"step": 50
|
| 97 |
},
|
| 98 |
{
|
| 99 |
+
"loss": 1.8792,
|
| 100 |
+
"grad_norm": 4.762349605560303,
|
| 101 |
+
"learning_rate": 5.363636363636363e-07,
|
| 102 |
"epoch": 0.05501432664756447,
|
| 103 |
"step": 60
|
| 104 |
},
|
| 105 |
{
|
| 106 |
+
"loss": 1.8231,
|
| 107 |
+
"grad_norm": 4.353812217712402,
|
| 108 |
+
"learning_rate": 6.272727272727273e-07,
|
| 109 |
"epoch": 0.06418338108882521,
|
| 110 |
"step": 70
|
| 111 |
},
|
| 112 |
{
|
| 113 |
+
"loss": 1.8868,
|
| 114 |
+
"grad_norm": 4.538625240325928,
|
| 115 |
+
"learning_rate": 7.181818181818181e-07,
|
| 116 |
"epoch": 0.07335243553008595,
|
| 117 |
"step": 80
|
| 118 |
},
|
| 119 |
{
|
| 120 |
+
"loss": 1.6794,
|
| 121 |
+
"grad_norm": 3.11306095123291,
|
| 122 |
+
"learning_rate": 8.09090909090909e-07,
|
| 123 |
"epoch": 0.0825214899713467,
|
| 124 |
"step": 90
|
| 125 |
},
|
| 126 |
{
|
| 127 |
+
"loss": 1.7012,
|
| 128 |
+
"grad_norm": 3.303393602371216,
|
| 129 |
+
"learning_rate": 9e-07,
|
| 130 |
"epoch": 0.09169054441260745,
|
| 131 |
"step": 100
|
| 132 |
},
|
| 133 |
{
|
| 134 |
+
"eval_loss": 1.6190643310546875,
|
| 135 |
+
"eval_runtime": 163.7929,
|
| 136 |
+
"eval_samples_per_second": 5.666,
|
| 137 |
+
"eval_steps_per_second": 2.833,
|
| 138 |
"epoch": 0.09169054441260745,
|
| 139 |
"step": 100
|
| 140 |
},
|
| 141 |
{
|
| 142 |
+
"loss": 1.6738,
|
| 143 |
+
"grad_norm": 2.3432679176330566,
|
| 144 |
+
"learning_rate": 9.909090909090909e-07,
|
| 145 |
"epoch": 0.1008595988538682,
|
| 146 |
"step": 110
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"loss": 1.5342,
|
| 150 |
+
"grad_norm": 1.411333441734314,
|
| 151 |
+
"learning_rate": 9.997923381619255e-07,
|
| 152 |
"epoch": 0.11002865329512894,
|
| 153 |
"step": 120
|
| 154 |
},
|
| 155 |
{
|
| 156 |
+
"loss": 1.4696,
|
| 157 |
+
"grad_norm": 1.3321274518966675,
|
| 158 |
+
"learning_rate": 9.990747162241872e-07,
|
| 159 |
"epoch": 0.11919770773638969,
|
| 160 |
"step": 130
|
| 161 |
},
|
| 162 |
{
|
| 163 |
+
"loss": 1.4552,
|
| 164 |
+
"grad_norm": 1.0566920042037964,
|
| 165 |
+
"learning_rate": 9.978453061876695e-07,
|
| 166 |
"epoch": 0.12836676217765042,
|
| 167 |
"step": 140
|
| 168 |
},
|
| 169 |
{
|
| 170 |
+
"loss": 1.5192,
|
| 171 |
+
"grad_norm": 0.9345868229866028,
|
| 172 |
+
"learning_rate": 9.96105368780285e-07,
|
| 173 |
"epoch": 0.13753581661891118,
|
| 174 |
"step": 150
|
| 175 |
},
|
| 176 |
{
|
| 177 |
+
"eval_loss": 1.3909412622451782,
|
| 178 |
+
"eval_runtime": 170.2558,
|
| 179 |
+
"eval_samples_per_second": 5.451,
|
| 180 |
+
"eval_steps_per_second": 2.725,
|
| 181 |
"epoch": 0.13753581661891118,
|
| 182 |
"step": 150
|
| 183 |
},
|
| 184 |
{
|
| 185 |
+
"loss": 1.4217,
|
| 186 |
+
"grad_norm": 0.8553086519241333,
|
| 187 |
+
"learning_rate": 9.938566882624436e-07,
|
| 188 |
"epoch": 0.1467048710601719,
|
| 189 |
"step": 160
|
| 190 |
},
|
| 191 |
{
|
| 192 |
+
"loss": 1.4101,
|
| 193 |
+
"grad_norm": 0.8422027230262756,
|
| 194 |
+
"learning_rate": 9.911015705973398e-07,
|
| 195 |
"epoch": 0.15587392550143267,
|
| 196 |
"step": 170
|
| 197 |
},
|
| 198 |
{
|
| 199 |
+
"loss": 1.4131,
|
| 200 |
+
"grad_norm": 0.6763940453529358,
|
| 201 |
+
"learning_rate": 9.878428410862482e-07,
|
| 202 |
"epoch": 0.1650429799426934,
|
| 203 |
"step": 180
|
| 204 |
},
|
| 205 |
{
|
| 206 |
+
"loss": 1.4201,
|
| 207 |
+
"grad_norm": 0.5860380530357361,
|
| 208 |
+
"learning_rate": 9.8408384147125e-07,
|
| 209 |
"epoch": 0.17421203438395416,
|
| 210 |
"step": 190
|
| 211 |
},
|
| 212 |
{
|
| 213 |
+
"loss": 1.2782,
|
| 214 |
+
"grad_norm": 0.7507234215736389,
|
| 215 |
+
"learning_rate": 9.79828426508364e-07,
|
| 216 |
"epoch": 0.1833810888252149,
|
| 217 |
"step": 200
|
| 218 |
},
|
| 219 |
{
|
| 220 |
+
"eval_loss": 1.3075143098831177,
|
| 221 |
+
"eval_runtime": 168.9052,
|
| 222 |
+
"eval_samples_per_second": 5.494,
|
| 223 |
+
"eval_steps_per_second": 2.747,
|
| 224 |
"epoch": 0.1833810888252149,
|
| 225 |
"step": 200
|
| 226 |
},
|
| 227 |
{
|
| 228 |
+
"loss": 1.3912,
|
| 229 |
+
"grad_norm": 0.6225572228431702,
|
| 230 |
+
"learning_rate": 9.750809600145952e-07,
|
| 231 |
"epoch": 0.19255014326647565,
|
| 232 |
"step": 210
|
| 233 |
},
|
| 234 |
{
|
| 235 |
+
"loss": 1.2743,
|
| 236 |
+
"grad_norm": 0.5334329009056091,
|
| 237 |
+
"learning_rate": 9.698463103929541e-07,
|
| 238 |
"epoch": 0.2017191977077364,
|
| 239 |
"step": 220
|
| 240 |
},
|
| 241 |
{
|
| 242 |
+
"loss": 1.2105,
|
| 243 |
+
"grad_norm": 0.730050265789032,
|
| 244 |
+
"learning_rate": 9.641298456400363e-07,
|
| 245 |
"epoch": 0.21088825214899715,
|
| 246 |
"step": 230
|
| 247 |
},
|
| 248 |
{
|
| 249 |
+
"loss": 1.3692,
|
| 250 |
+
"grad_norm": 0.646460235118866,
|
| 251 |
+
"learning_rate": 9.579374278412817e-07,
|
| 252 |
"epoch": 0.22005730659025788,
|
| 253 |
"step": 240
|
| 254 |
},
|
| 255 |
{
|
| 256 |
+
"loss": 1.2971,
|
| 257 |
+
"grad_norm": 0.5395148992538452,
|
| 258 |
+
"learning_rate": 9.512754071595603e-07,
|
| 259 |
"epoch": 0.22922636103151864,
|
| 260 |
"step": 250
|
| 261 |
},
|
| 262 |
{
|
| 263 |
+
"eval_loss": 1.2624306678771973,
|
| 264 |
+
"eval_runtime": 157.9571,
|
| 265 |
+
"eval_samples_per_second": 5.875,
|
| 266 |
+
"eval_steps_per_second": 2.938,
|
| 267 |
"epoch": 0.22922636103151864,
|
| 268 |
"step": 250
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"loss": 1.3346,
|
| 272 |
+
"grad_norm": 0.5858215093612671,
|
| 273 |
+
"learning_rate": 9.441506153232442e-07,
|
| 274 |
"epoch": 0.23839541547277937,
|
| 275 |
"step": 260
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"loss": 1.2195,
|
| 279 |
+
"grad_norm": 0.5456379055976868,
|
| 280 |
+
"learning_rate": 9.365703586204494e-07,
|
| 281 |
"epoch": 0.2475644699140401,
|
| 282 |
"step": 270
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"loss": 1.3763,
|
| 286 |
+
"grad_norm": 0.8651963472366333,
|
| 287 |
+
"learning_rate": 9.285424104066275e-07,
|
| 288 |
"epoch": 0.25673352435530084,
|
| 289 |
"step": 280
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"loss": 1.2256,
|
| 293 |
+
"grad_norm": 0.5938352942466736,
|
| 294 |
+
"learning_rate": 9.20075003133193e-07,
|
| 295 |
"epoch": 0.2659025787965616,
|
| 296 |
"step": 290
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"loss": 1.3137,
|
| 300 |
+
"grad_norm": 0.6355459690093994,
|
| 301 |
+
"learning_rate": 9.111768199053586e-07,
|
| 302 |
"epoch": 0.27507163323782235,
|
| 303 |
"step": 300
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"eval_loss": 1.2268821001052856,
|
| 307 |
+
"eval_runtime": 164.1638,
|
| 308 |
+
"eval_samples_per_second": 5.653,
|
| 309 |
+
"eval_steps_per_second": 2.826,
|
| 310 |
"epoch": 0.27507163323782235,
|
| 311 |
"step": 300
|
| 312 |
},
|
| 313 |
{
|
| 314 |
+
"loss": 1.4409,
|
| 315 |
+
"grad_norm": 0.9009450078010559,
|
| 316 |
+
"learning_rate": 9.018569855778383e-07,
|
| 317 |
"epoch": 0.2842406876790831,
|
| 318 |
"step": 310
|
| 319 |
},
|
| 320 |
{
|
| 321 |
+
"loss": 1.2566,
|
| 322 |
+
"grad_norm": 0.5912006497383118,
|
| 323 |
+
"learning_rate": 8.921250573975455e-07,
|
| 324 |
"epoch": 0.2934097421203438,
|
| 325 |
"step": 320
|
| 326 |
},
|
| 327 |
{
|
| 328 |
+
"loss": 1.1593,
|
| 329 |
+
"grad_norm": 0.6118663549423218,
|
| 330 |
+
"learning_rate": 8.81991015202887e-07,
|
| 331 |
"epoch": 0.3025787965616046,
|
| 332 |
"step": 330
|
| 333 |
},
|
| 334 |
{
|
| 335 |
+
"loss": 1.2555,
|
| 336 |
+
"grad_norm": 0.581721305847168,
|
| 337 |
+
"learning_rate": 8.714652511896993e-07,
|
| 338 |
"epoch": 0.31174785100286534,
|
| 339 |
"step": 340
|
| 340 |
},
|
| 341 |
{
|
| 342 |
+
"loss": 1.1209,
|
| 343 |
+
"grad_norm": 0.5230151414871216,
|
| 344 |
+
"learning_rate": 8.605585592543211e-07,
|
| 345 |
"epoch": 0.3209169054441261,
|
| 346 |
"step": 350
|
| 347 |
},
|
| 348 |
{
|
| 349 |
+
"eval_loss": 1.1941955089569092,
|
| 350 |
+
"eval_runtime": 162.9119,
|
| 351 |
+
"eval_samples_per_second": 5.696,
|
| 352 |
+
"eval_steps_per_second": 2.848,
|
| 353 |
"epoch": 0.3209169054441261,
|
| 354 |
"step": 350
|
| 355 |
},
|
| 356 |
{
|
| 357 |
+
"loss": 1.3564,
|
| 358 |
+
"grad_norm": 0.565862238407135,
|
| 359 |
+
"learning_rate": 8.492821239247363e-07,
|
| 360 |
"epoch": 0.3300859598853868,
|
| 361 |
"step": 360
|
| 362 |
},
|
| 363 |
{
|
| 364 |
+
"loss": 1.3997,
|
| 365 |
+
"grad_norm": 0.6057285666465759,
|
| 366 |
+
"learning_rate": 8.376475088911317e-07,
|
| 367 |
"epoch": 0.33925501432664756,
|
| 368 |
"step": 370
|
| 369 |
},
|
| 370 |
{
|
| 371 |
+
"loss": 1.0878,
|
| 372 |
+
"grad_norm": 0.7517871260643005,
|
| 373 |
+
"learning_rate": 8.256666451476336e-07,
|
| 374 |
"epoch": 0.3484240687679083,
|
| 375 |
"step": 380
|
| 376 |
},
|
| 377 |
{
|
| 378 |
+
"loss": 1.3009,
|
| 379 |
+
"grad_norm": 0.7168652415275574,
|
| 380 |
+
"learning_rate": 8.133518187573862e-07,
|
| 381 |
"epoch": 0.35759312320916903,
|
| 382 |
"step": 390
|
| 383 |
},
|
| 384 |
{
|
| 385 |
+
"loss": 1.1606,
|
| 386 |
+
"grad_norm": 0.7250906825065613,
|
| 387 |
+
"learning_rate": 8.007156582535131e-07,
|
| 388 |
"epoch": 0.3667621776504298,
|
| 389 |
"step": 400
|
| 390 |
},
|
| 391 |
{
|
| 392 |
+
"eval_loss": 1.166169285774231,
|
| 393 |
+
"eval_runtime": 158.6155,
|
| 394 |
+
"eval_samples_per_second": 5.851,
|
| 395 |
+
"eval_steps_per_second": 2.925,
|
| 396 |
"epoch": 0.3667621776504298,
|
| 397 |
"step": 400
|
| 398 |
},
|
| 399 |
{
|
| 400 |
+
"loss": 1.2593,
|
| 401 |
+
"grad_norm": 0.6665163636207581,
|
| 402 |
+
"learning_rate": 7.877711216888867e-07,
|
| 403 |
"epoch": 0.37593123209169055,
|
| 404 |
"step": 410
|
| 405 |
},
|
| 406 |
{
|
| 407 |
+
"loss": 1.1365,
|
| 408 |
+
"grad_norm": 0.6199079751968384,
|
| 409 |
+
"learning_rate": 7.745314833479833e-07,
|
| 410 |
"epoch": 0.3851002865329513,
|
| 411 |
"step": 420
|
| 412 |
},
|
| 413 |
{
|
| 414 |
+
"loss": 1.19,
|
| 415 |
+
"grad_norm": 0.6242042779922485,
|
| 416 |
+
"learning_rate": 7.6101032013445e-07,
|
| 417 |
"epoch": 0.394269340974212,
|
| 418 |
"step": 430
|
| 419 |
},
|
| 420 |
{
|
| 421 |
+
"loss": 1.1405,
|
| 422 |
+
"grad_norm": 0.657778263092041,
|
| 423 |
+
"learning_rate": 7.472214976483451e-07,
|
| 424 |
"epoch": 0.4034383954154728,
|
| 425 |
"step": 440
|
| 426 |
},
|
| 427 |
{
|
| 428 |
+
"loss": 1.0791,
|
| 429 |
+
"grad_norm": 0.5634785890579224,
|
| 430 |
+
"learning_rate": 7.331791559673269e-07,
|
| 431 |
"epoch": 0.41260744985673353,
|
| 432 |
"step": 450
|
| 433 |
},
|
| 434 |
{
|
| 435 |
+
"eval_loss": 1.1434489488601685,
|
| 436 |
+
"eval_runtime": 153.2068,
|
| 437 |
+
"eval_samples_per_second": 6.057,
|
| 438 |
+
"eval_steps_per_second": 3.029,
|
| 439 |
"epoch": 0.41260744985673353,
|
| 440 |
"step": 450
|
| 441 |
},
|
| 442 |
{
|
| 443 |
+
"loss": 1.1352,
|
| 444 |
+
"grad_norm": 0.7913809418678284,
|
| 445 |
+
"learning_rate": 7.188976951463723e-07,
|
| 446 |
"epoch": 0.4217765042979943,
|
| 447 |
"step": 460
|
| 448 |
},
|
| 449 |
{
|
| 450 |
+
"loss": 1.3036,
|
| 451 |
+
"grad_norm": 0.6191056966781616,
|
| 452 |
+
"learning_rate": 7.043917604508971e-07,
|
| 453 |
"epoch": 0.430945558739255,
|
| 454 |
"step": 470
|
| 455 |
},
|
| 456 |
{
|
| 457 |
+
"loss": 1.121,
|
| 458 |
+
"grad_norm": 0.6955880522727966,
|
| 459 |
+
"learning_rate": 6.896762273384178e-07,
|
| 460 |
"epoch": 0.44011461318051576,
|
| 461 |
"step": 480
|
| 462 |
},
|
| 463 |
{
|
| 464 |
+
"loss": 1.0997,
|
| 465 |
+
"grad_norm": 0.8450888395309448,
|
| 466 |
+
"learning_rate": 6.747661862041585e-07,
|
| 467 |
"epoch": 0.4492836676217765,
|
| 468 |
"step": 490
|
| 469 |
},
|
| 470 |
{
|
| 471 |
+
"loss": 1.3604,
|
| 472 |
+
"grad_norm": 0.8735764622688293,
|
| 473 |
+
"learning_rate": 6.596769269062443e-07,
|
| 474 |
"epoch": 0.4584527220630373,
|
| 475 |
"step": 500
|
| 476 |
},
|
| 477 |
{
|
| 478 |
+
"eval_loss": 1.1247467994689941,
|
| 479 |
+
"eval_runtime": 152.4716,
|
| 480 |
+
"eval_samples_per_second": 6.086,
|
| 481 |
+
"eval_steps_per_second": 3.043,
|
| 482 |
"epoch": 0.4584527220630373,
|
| 483 |
"step": 500
|
| 484 |
},
|
| 485 |
{
|
| 486 |
+
"loss": 1.171,
|
| 487 |
+
"grad_norm": 0.7228217124938965,
|
| 488 |
+
"learning_rate": 6.444239230863504e-07,
|
| 489 |
"epoch": 0.467621776504298,
|
| 490 |
"step": 510
|
| 491 |
},
|
| 492 |
{
|
| 493 |
+
"loss": 1.03,
|
| 494 |
+
"grad_norm": 0.6075210571289062,
|
| 495 |
+
"learning_rate": 6.290228163018867e-07,
|
| 496 |
"epoch": 0.47679083094555874,
|
| 497 |
"step": 520
|
| 498 |
},
|
| 499 |
{
|
| 500 |
+
"loss": 1.0345,
|
| 501 |
+
"grad_norm": 0.7692680954933167,
|
| 502 |
+
"learning_rate": 6.134893999859886e-07,
|
| 503 |
"epoch": 0.4859598853868195,
|
| 504 |
"step": 530
|
| 505 |
},
|
| 506 |
{
|
| 507 |
+
"loss": 1.2594,
|
| 508 |
+
"grad_norm": 0.7600648403167725,
|
| 509 |
+
"learning_rate": 5.978396032517639e-07,
|
| 510 |
"epoch": 0.4951289398280802,
|
| 511 |
"step": 540
|
| 512 |
},
|
| 513 |
{
|
| 514 |
+
"loss": 1.0254,
|
| 515 |
+
"grad_norm": 0.6178115010261536,
|
| 516 |
+
"learning_rate": 5.820894745574025e-07,
|
| 517 |
"epoch": 0.504297994269341,
|
| 518 |
"step": 550
|
| 519 |
},
|
| 520 |
{
|
| 521 |
+
"eval_loss": 1.1089264154434204,
|
| 522 |
+
"eval_runtime": 151.8982,
|
| 523 |
+
"eval_samples_per_second": 6.109,
|
| 524 |
+
"eval_steps_per_second": 3.055,
|
| 525 |
"epoch": 0.504297994269341,
|
| 526 |
"step": 550
|
| 527 |
},
|
| 528 |
{
|
| 529 |
+
"loss": 1.1049,
|
| 530 |
+
"grad_norm": 0.5585054159164429,
|
| 531 |
+
"learning_rate": 5.662551652489008e-07,
|
| 532 |
"epoch": 0.5134670487106017,
|
| 533 |
"step": 560
|
| 534 |
},
|
| 535 |
{
|
| 536 |
+
"loss": 1.0898,
|
| 537 |
+
"grad_norm": 0.6844518780708313,
|
| 538 |
+
"learning_rate": 5.503529129972792e-07,
|
| 539 |
"epoch": 0.5226361031518625,
|
| 540 |
"step": 570
|
| 541 |
},
|
| 542 |
{
|
| 543 |
+
"loss": 1.1037,
|
| 544 |
+
"grad_norm": 0.8425552845001221,
|
| 545 |
+
"learning_rate": 5.34399025147273e-07,
|
| 546 |
"epoch": 0.5318051575931232,
|
| 547 |
"step": 580
|
| 548 |
},
|
| 549 |
{
|
| 550 |
+
"loss": 1.1019,
|
| 551 |
+
"grad_norm": 0.648064136505127,
|
| 552 |
+
"learning_rate": 5.18409861994576e-07,
|
| 553 |
"epoch": 0.540974212034384,
|
| 554 |
"step": 590
|
| 555 |
},
|
| 556 |
{
|
| 557 |
+
"loss": 1.1863,
|
| 558 |
+
"grad_norm": 0.5788621306419373,
|
| 559 |
+
"learning_rate": 5.024018200087854e-07,
|
| 560 |
"epoch": 0.5501432664756447,
|
| 561 |
"step": 600
|
| 562 |
},
|
| 563 |
{
|
| 564 |
+
"eval_loss": 1.095629096031189,
|
| 565 |
+
"eval_runtime": 150.8171,
|
| 566 |
+
"eval_samples_per_second": 6.153,
|
| 567 |
+
"eval_steps_per_second": 3.077,
|
| 568 |
"epoch": 0.5501432664756447,
|
| 569 |
"step": 600
|
| 570 |
},
|
| 571 |
{
|
| 572 |
+
"loss": 1.1025,
|
| 573 |
+
"grad_norm": 0.6422027349472046,
|
| 574 |
+
"learning_rate": 4.86391315019248e-07,
|
| 575 |
"epoch": 0.5593123209169054,
|
| 576 |
"step": 610
|
| 577 |
},
|
| 578 |
{
|
| 579 |
+
"loss": 1.0666,
|
| 580 |
+
"grad_norm": 0.6005454063415527,
|
| 581 |
+
"learning_rate": 4.703947653810575e-07,
|
| 582 |
"epoch": 0.5684813753581662,
|
| 583 |
"step": 620
|
| 584 |
},
|
| 585 |
{
|
| 586 |
+
"loss": 1.215,
|
| 587 |
+
"grad_norm": 0.6145904064178467,
|
| 588 |
+
"learning_rate": 4.544285751384584e-07,
|
| 589 |
"epoch": 0.5776504297994269,
|
| 590 |
"step": 630
|
| 591 |
},
|
| 592 |
{
|
| 593 |
+
"loss": 1.1613,
|
| 594 |
+
"grad_norm": 0.8756449818611145,
|
| 595 |
+
"learning_rate": 4.385091172029275e-07,
|
| 596 |
"epoch": 0.5868194842406876,
|
| 597 |
"step": 640
|
| 598 |
},
|
| 599 |
{
|
| 600 |
+
"loss": 1.1092,
|
| 601 |
+
"grad_norm": 0.7930067181587219,
|
| 602 |
+
"learning_rate": 4.2265271656318e-07,
|
| 603 |
"epoch": 0.5959885386819485,
|
| 604 |
"step": 650
|
| 605 |
},
|
| 606 |
{
|
| 607 |
+
"eval_loss": 1.0845845937728882,
|
| 608 |
+
"eval_runtime": 151.0566,
|
| 609 |
+
"eval_samples_per_second": 6.143,
|
| 610 |
+
"eval_steps_per_second": 3.072,
|
| 611 |
"epoch": 0.5959885386819485,
|
| 612 |
"step": 650
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"loss": 1.214,
|
| 616 |
+
"grad_norm": 1.1016592979431152,
|
| 617 |
+
"learning_rate": 4.068756335443198e-07,
|
| 618 |
"epoch": 0.6051575931232092,
|
| 619 |
"step": 660
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"loss": 1.3335,
|
| 623 |
+
"grad_norm": 0.7920063138008118,
|
| 624 |
+
"learning_rate": 3.9119404713330013e-07,
|
| 625 |
"epoch": 0.6143266475644699,
|
| 626 |
"step": 670
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"loss": 1.123,
|
| 630 |
+
"grad_norm": 0.792630136013031,
|
| 631 |
+
"learning_rate": 3.7562403838779467e-07,
|
| 632 |
"epoch": 0.6234957020057307,
|
| 633 |
"step": 680
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"loss": 1.2098,
|
| 637 |
+
"grad_norm": 0.8105105757713318,
|
| 638 |
+
"learning_rate": 3.601815739454928e-07,
|
| 639 |
"epoch": 0.6326647564469914,
|
| 640 |
"step": 690
|
| 641 |
},
|
| 642 |
{
|
| 643 |
+
"loss": 1.0302,
|
| 644 |
+
"grad_norm": 0.6204477548599243,
|
| 645 |
+
"learning_rate": 3.448824896507292e-07,
|
| 646 |
"epoch": 0.6418338108882522,
|
| 647 |
"step": 700
|
| 648 |
},
|
| 649 |
{
|
| 650 |
+
"eval_loss": 1.0753319263458252,
|
| 651 |
+
"eval_runtime": 159.1632,
|
| 652 |
+
"eval_samples_per_second": 5.83,
|
| 653 |
+
"eval_steps_per_second": 2.915,
|
| 654 |
"epoch": 0.6418338108882522,
|
| 655 |
"step": 700
|
| 656 |
},
|
| 657 |
{
|
| 658 |
+
"loss": 1.109,
|
| 659 |
+
"grad_norm": 0.6277522444725037,
|
| 660 |
+
"learning_rate": 3.297424743152381e-07,
|
| 661 |
"epoch": 0.6510028653295129,
|
| 662 |
"step": 710
|
| 663 |
},
|
| 664 |
{
|
| 665 |
+
"loss": 1.0808,
|
| 666 |
+
"grad_norm": 0.7879471182823181,
|
| 667 |
+
"learning_rate": 3.1477705362968696e-07,
|
| 668 |
"epoch": 0.6601719197707736,
|
| 669 |
"step": 720
|
| 670 |
},
|
| 671 |
{
|
| 672 |
+
"loss": 1.0842,
|
| 673 |
+
"grad_norm": 0.8374884128570557,
|
| 674 |
+
"learning_rate": 3.000015742424857e-07,
|
| 675 |
"epoch": 0.6693409742120344,
|
| 676 |
"step": 730
|
| 677 |
},
|
| 678 |
{
|
| 679 |
+
"loss": 1.2445,
|
| 680 |
+
"grad_norm": 0.7892112731933594,
|
| 681 |
+
"learning_rate": 2.85431188022199e-07,
|
| 682 |
"epoch": 0.6785100286532951,
|
| 683 |
"step": 740
|
| 684 |
},
|
| 685 |
{
|
| 686 |
+
"loss": 1.1506,
|
| 687 |
+
"grad_norm": 0.5540062785148621,
|
| 688 |
+
"learning_rate": 2.710808365197e-07,
|
| 689 |
"epoch": 0.6876790830945558,
|
| 690 |
"step": 750
|
| 691 |
},
|
| 692 |
{
|
| 693 |
+
"eval_loss": 1.0686043500900269,
|
| 694 |
+
"eval_runtime": 154.055,
|
| 695 |
+
"eval_samples_per_second": 6.024,
|
| 696 |
+
"eval_steps_per_second": 3.012,
|
| 697 |
"epoch": 0.6876790830945558,
|
| 698 |
"step": 750
|
| 699 |
},
|
| 700 |
{
|
| 701 |
+
"loss": 1.1023,
|
| 702 |
+
"grad_norm": 0.5218796133995056,
|
| 703 |
+
"learning_rate": 2.569652356460007e-07,
|
| 704 |
"epoch": 0.6968481375358166,
|
| 705 |
"step": 760
|
| 706 |
},
|
| 707 |
{
|
| 708 |
+
"loss": 1.1801,
|
| 709 |
+
"grad_norm": 0.7433627247810364,
|
| 710 |
+
"learning_rate": 2.430988605814691e-07,
|
| 711 |
"epoch": 0.7060171919770774,
|
| 712 |
"step": 770
|
| 713 |
},
|
| 714 |
{
|
| 715 |
+
"loss": 1.0276,
|
| 716 |
+
"grad_norm": 0.629487931728363,
|
| 717 |
+
"learning_rate": 2.294959309319086e-07,
|
| 718 |
"epoch": 0.7151862464183381,
|
| 719 |
"step": 780
|
| 720 |
},
|
| 721 |
{
|
| 722 |
+
"loss": 1.1164,
|
| 723 |
+
"grad_norm": 0.6667075157165527,
|
| 724 |
+
"learning_rate": 2.1617039614672378e-07,
|
| 725 |
"epoch": 0.7243553008595989,
|
| 726 |
"step": 790
|
| 727 |
},
|
| 728 |
{
|
| 729 |
+
"loss": 1.0869,
|
| 730 |
+
"grad_norm": 0.6930222511291504,
|
| 731 |
+
"learning_rate": 2.0313592121412464e-07,
|
| 732 |
"epoch": 0.7335243553008596,
|
| 733 |
"step": 800
|
| 734 |
},
|
| 735 |
{
|
| 736 |
+
"eval_loss": 1.0634101629257202,
|
| 737 |
+
"eval_runtime": 154.1494,
|
| 738 |
+
"eval_samples_per_second": 6.02,
|
| 739 |
+
"eval_steps_per_second": 3.01,
|
| 740 |
"epoch": 0.7335243553008596,
|
| 741 |
"step": 800
|
| 742 |
},
|
| 743 |
{
|
| 744 |
+
"loss": 1.0937,
|
| 745 |
+
"grad_norm": 0.7793363332748413,
|
| 746 |
+
"learning_rate": 1.904058726480367e-07,
|
| 747 |
"epoch": 0.7426934097421204,
|
| 748 |
"step": 810
|
| 749 |
},
|
| 750 |
{
|
| 751 |
+
"loss": 0.9728,
|
| 752 |
+
"grad_norm": 0.5570642948150635,
|
| 753 |
+
"learning_rate": 1.7799330478109026e-07,
|
| 754 |
"epoch": 0.7518624641833811,
|
| 755 |
"step": 820
|
| 756 |
},
|
| 757 |
{
|
| 758 |
+
"loss": 1.2568,
|
| 759 |
+
"grad_norm": 0.9017201662063599,
|
| 760 |
+
"learning_rate": 1.65910946377743e-07,
|
| 761 |
"epoch": 0.7610315186246418,
|
| 762 |
"step": 830
|
| 763 |
},
|
| 764 |
{
|
| 765 |
+
"loss": 1.0049,
|
| 766 |
+
"grad_norm": 0.7178328633308411,
|
| 767 |
+
"learning_rate": 1.5417118758126408e-07,
|
| 768 |
"epoch": 0.7702005730659026,
|
| 769 |
"step": 840
|
| 770 |
},
|
| 771 |
{
|
| 772 |
+
"loss": 1.0576,
|
| 773 |
+
"grad_norm": 1.031610369682312,
|
| 774 |
+
"learning_rate": 1.4278606720796543e-07,
|
| 775 |
"epoch": 0.7793696275071633,
|
| 776 |
"step": 850
|
| 777 |
},
|
| 778 |
{
|
| 779 |
+
"eval_loss": 1.0599370002746582,
|
| 780 |
+
"eval_runtime": 154.7364,
|
| 781 |
+
"eval_samples_per_second": 5.997,
|
| 782 |
+
"eval_steps_per_second": 2.999,
|
| 783 |
"epoch": 0.7793696275071633,
|
| 784 |
"step": 850
|
| 785 |
},
|
| 786 |
{
|
| 787 |
+
"loss": 1.1797,
|
| 788 |
+
"grad_norm": 1.0518614053726196,
|
| 789 |
+
"learning_rate": 1.3176726040171e-07,
|
| 790 |
"epoch": 0.788538681948424,
|
| 791 |
"step": 860
|
| 792 |
},
|
| 793 |
{
|
| 794 |
+
"loss": 1.2085,
|
| 795 |
+
"grad_norm": 0.7290861010551453,
|
| 796 |
+
"learning_rate": 1.21126066661356e-07,
|
| 797 |
"epoch": 0.7977077363896848,
|
| 798 |
"step": 870
|
| 799 |
},
|
| 800 |
{
|
| 801 |
+
"loss": 1.1165,
|
| 802 |
+
"grad_norm": 0.6315222382545471,
|
| 803 |
+
"learning_rate": 1.108733982534159e-07,
|
| 804 |
"epoch": 0.8068767908309455,
|
| 805 |
"step": 880
|
| 806 |
},
|
| 807 |
{
|
| 808 |
+
"loss": 1.158,
|
| 809 |
+
"grad_norm": 0.685243546962738,
|
| 810 |
+
"learning_rate": 1.0101976902181225e-07,
|
| 811 |
"epoch": 0.8160458452722062,
|
| 812 |
"step": 890
|
| 813 |
},
|
| 814 |
{
|
| 815 |
+
"loss": 1.0584,
|
| 816 |
+
"grad_norm": 0.7780338525772095,
|
| 817 |
+
"learning_rate": 9.157528360620415e-08,
|
| 818 |
"epoch": 0.8252148997134671,
|
| 819 |
"step": 900
|
| 820 |
},
|
| 821 |
{
|
| 822 |
+
"eval_loss": 1.057593584060669,
|
| 823 |
+
"eval_runtime": 153.5904,
|
| 824 |
+
"eval_samples_per_second": 6.042,
|
| 825 |
+
"eval_steps_per_second": 3.021,
|
| 826 |
"epoch": 0.8252148997134671,
|
| 827 |
"step": 900
|
| 828 |
},
|
| 829 |
{
|
| 830 |
+
"loss": 1.1489,
|
| 831 |
+
"grad_norm": 0.6839588284492493,
|
| 832 |
+
"learning_rate": 8.254962707994373e-08,
|
| 833 |
"epoch": 0.8343839541547278,
|
| 834 |
"step": 910
|
| 835 |
},
|
| 836 |
{
|
| 837 |
+
"loss": 1.1096,
|
| 838 |
+
"grad_norm": 0.9299020171165466,
|
| 839 |
+
"learning_rate": 7.395205501828577e-08,
|
| 840 |
"epoch": 0.8435530085959886,
|
| 841 |
"step": 920
|
| 842 |
},
|
| 843 |
{
|
| 844 |
+
"loss": 1.1224,
|
| 845 |
+
"grad_norm": 0.791289746761322,
|
| 846 |
+
"learning_rate": 6.579138400703715e-08,
|
| 847 |
"epoch": 0.8527220630372493,
|
| 848 |
"step": 930
|
| 849 |
},
|
| 850 |
{
|
| 851 |
+
"loss": 1.038,
|
| 852 |
+
"grad_norm": 0.6159808039665222,
|
| 853 |
+
"learning_rate": 5.807598260137758e-08,
|
| 854 |
"epoch": 0.86189111747851,
|
| 855 |
"step": 940
|
| 856 |
},
|
| 857 |
{
|
| 858 |
+
"loss": 1.0708,
|
| 859 |
+
"grad_norm": 0.7773894667625427,
|
| 860 |
+
"learning_rate": 5.08137627441253e-08,
|
| 861 |
"epoch": 0.8710601719197708,
|
| 862 |
"step": 950
|
| 863 |
},
|
| 864 |
{
|
| 865 |
+
"eval_loss": 1.0561405420303345,
|
| 866 |
+
"eval_runtime": 154.3615,
|
| 867 |
+
"eval_samples_per_second": 6.012,
|
| 868 |
+
"eval_steps_per_second": 3.006,
|
| 869 |
"epoch": 0.8710601719197708,
|
| 870 |
"step": 950
|
| 871 |
},
|
| 872 |
{
|
| 873 |
+
"loss": 1.0656,
|
| 874 |
+
"grad_norm": 0.707645833492279,
|
| 875 |
+
"learning_rate": 4.401217165224563e-08,
|
| 876 |
"epoch": 0.8802292263610315,
|
| 877 |
"step": 960
|
| 878 |
},
|
| 879 |
{
|
| 880 |
+
"loss": 1.0164,
|
| 881 |
+
"grad_norm": 0.6336905360221863,
|
| 882 |
+
"learning_rate": 3.767818417992446e-08,
|
| 883 |
"epoch": 0.8893982808022922,
|
| 884 |
"step": 970
|
| 885 |
},
|
| 886 |
{
|
| 887 |
+
"loss": 1.0766,
|
| 888 |
+
"grad_norm": 0.8207520842552185,
|
| 889 |
+
"learning_rate": 3.181829566603772e-08,
|
| 890 |
"epoch": 0.898567335243553,
|
| 891 |
"step": 980
|
| 892 |
},
|
| 893 |
{
|
| 894 |
+
"loss": 1.0929,
|
| 895 |
+
"grad_norm": 0.6286782026290894,
|
| 896 |
+
"learning_rate": 2.643851527335006e-08,
|
| 897 |
"epoch": 0.9077363896848137,
|
| 898 |
"step": 990
|
| 899 |
},
|
| 900 |
{
|
| 901 |
+
"loss": 1.2541,
|
| 902 |
+
"grad_norm": 0.817637026309967,
|
| 903 |
+
"learning_rate": 2.1544359826275726e-08,
|
| 904 |
"epoch": 0.9169054441260746,
|
| 905 |
"step": 1000
|
| 906 |
},
|
| 907 |
{
|
| 908 |
+
"eval_loss": 1.0553829669952393,
|
| 909 |
+
"eval_runtime": 155.0837,
|
| 910 |
+
"eval_samples_per_second": 5.984,
|
| 911 |
+
"eval_steps_per_second": 2.992,
|
| 912 |
"epoch": 0.9169054441260746,
|
| 913 |
"step": 1000
|
| 914 |
},
|
| 915 |
{
|
| 916 |
+
"loss": 1.0413,
|
| 917 |
+
"grad_norm": 0.9485034942626953,
|
| 918 |
+
"learning_rate": 1.714084815351913e-08,
|
| 919 |
"epoch": 0.9260744985673353,
|
| 920 |
"step": 1010
|
| 921 |
},
|
| 922 |
{
|
| 923 |
+
"loss": 1.0516,
|
| 924 |
+
"grad_norm": 0.6737267971038818,
|
| 925 |
+
"learning_rate": 1.3232495941396637e-08,
|
| 926 |
"epoch": 0.935243553008596,
|
| 927 |
"step": 1020
|
| 928 |
},
|
| 929 |
{
|
| 930 |
+
"loss": 1.0517,
|
| 931 |
+
"grad_norm": 0.9414446353912354,
|
| 932 |
+
"learning_rate": 9.82331110311857e-09,
|
| 933 |
"epoch": 0.9444126074498568,
|
| 934 |
"step": 1030
|
| 935 |
},
|
| 936 |
{
|
| 937 |
+
"loss": 1.1016,
|
| 938 |
+
"grad_norm": 0.8654493689537048,
|
| 939 |
+
"learning_rate": 6.916789668778122e-09,
|
| 940 |
"epoch": 0.9535816618911175,
|
| 941 |
"step": 1040
|
| 942 |
},
|
| 943 |
{
|
| 944 |
+
"loss": 1.1055,
|
| 945 |
+
"grad_norm": 0.8262504935264587,
|
| 946 |
+
"learning_rate": 4.515912200264427e-09,
|
| 947 |
"epoch": 0.9627507163323782,
|
| 948 |
"step": 1050
|
| 949 |
},
|
| 950 |
{
|
| 951 |
+
"eval_loss": 1.0550979375839233,
|
| 952 |
+
"eval_runtime": 156.3052,
|
| 953 |
+
"eval_samples_per_second": 5.937,
|
| 954 |
+
"eval_steps_per_second": 2.969,
|
| 955 |
"epoch": 0.9627507163323782,
|
| 956 |
"step": 1050
|
| 957 |
},
|
| 958 |
{
|
| 959 |
+
"loss": 1.1321,
|
| 960 |
+
"grad_norm": 0.7707592844963074,
|
| 961 |
+
"learning_rate": 2.6231407347736546e-09,
|
| 962 |
"epoch": 0.971919770773639,
|
| 963 |
"step": 1060
|
| 964 |
},
|
| 965 |
{
|
| 966 |
+
"loss": 1.1039,
|
| 967 |
+
"grad_norm": 0.7415518760681152,
|
| 968 |
+
"learning_rate": 1.2404162600541113e-09,
|
| 969 |
"epoch": 0.9810888252148997,
|
| 970 |
"step": 1070
|
| 971 |
},
|
| 972 |
{
|
| 973 |
+
"loss": 1.0673,
|
| 974 |
+
"grad_norm": 0.6835209727287292,
|
| 975 |
+
"learning_rate": 3.6915672397436204e-10,
|
| 976 |
"epoch": 0.9902578796561604,
|
| 977 |
"step": 1080
|
| 978 |
},
|
| 979 |
{
|
| 980 |
+
"loss": 1.1213,
|
| 981 |
+
"grad_norm": 0.6404680013656616,
|
| 982 |
+
"learning_rate": 1.0255580454254786e-11,
|
| 983 |
"epoch": 0.9994269340974212,
|
| 984 |
"step": 1090
|
| 985 |
},
|
| 986 |
{
|
| 987 |
+
"train_runtime": 10741.972,
|
| 988 |
+
"train_samples_per_second": 1.624,
|
| 989 |
+
"train_steps_per_second": 0.102,
|
| 990 |
+
"total_flos": 1.5816368624117146e+17,
|
| 991 |
+
"train_loss": 1.2491859223840436,
|
| 992 |
"epoch": 1.0,
|
| 993 |
"step": 1091,
|
| 994 |
+
"total_runtime_sec": 10743.41768693924
|
| 995 |
}
|
| 996 |
]
|
| 997 |
}
|