Model save
Browse files- README.md +3 -5
- all_results.json +6 -6
- train_results.json +6 -6
- trainer_state.json +1678 -1022
README.md
CHANGED
|
@@ -1,11 +1,9 @@
|
|
| 1 |
---
|
| 2 |
-
base_model: Qwen/Qwen2.5-
|
| 3 |
-
datasets: kguo2/scaffold_finetune
|
| 4 |
library_name: transformers
|
| 5 |
model_name: finetune_demo
|
| 6 |
tags:
|
| 7 |
- generated_from_trainer
|
| 8 |
-
- open-r1
|
| 9 |
- trl
|
| 10 |
- sft
|
| 11 |
licence: license
|
|
@@ -13,7 +11,7 @@ licence: license
|
|
| 13 |
|
| 14 |
# Model Card for finetune_demo
|
| 15 |
|
| 16 |
-
This model is a fine-tuned version of [Qwen/Qwen2.5-
|
| 17 |
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 18 |
|
| 19 |
## Quick start
|
|
@@ -29,7 +27,7 @@ print(output["generated_text"])
|
|
| 29 |
|
| 30 |
## Training procedure
|
| 31 |
|
| 32 |
-
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/nd/huggingface/runs/
|
| 33 |
|
| 34 |
|
| 35 |
This model was trained with SFT.
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model: Qwen/Qwen2.5-7B-Instruct
|
|
|
|
| 3 |
library_name: transformers
|
| 4 |
model_name: finetune_demo
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
|
|
|
| 7 |
- trl
|
| 8 |
- sft
|
| 9 |
licence: license
|
|
|
|
| 11 |
|
| 12 |
# Model Card for finetune_demo
|
| 13 |
|
| 14 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
|
| 15 |
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
|
| 17 |
## Quick start
|
|
|
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/nd/huggingface/runs/ijld17d7)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with SFT.
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 7.444201440207176e+17,
|
| 3 |
+
"train_loss": 0.06586769079210378,
|
| 4 |
+
"train_runtime": 7485.0094,
|
| 5 |
+
"train_samples": 28548,
|
| 6 |
+
"train_samples_per_second": 9.147,
|
| 7 |
+
"train_steps_per_second": 0.191
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"total_flos":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"total_flos": 7.444201440207176e+17,
|
| 3 |
+
"train_loss": 0.06586769079210378,
|
| 4 |
+
"train_runtime": 7485.0094,
|
| 5 |
+
"train_samples": 28548,
|
| 6 |
+
"train_samples_per_second": 9.147,
|
| 7 |
+
"train_steps_per_second": 0.191
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -4,1648 +4,2304 @@
|
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
"epoch": 3.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
-
"epoch": 0.
|
| 14 |
-
"grad_norm":
|
| 15 |
-
"learning_rate":
|
| 16 |
-
"loss": 1.
|
| 17 |
-
"num_tokens":
|
| 18 |
"step": 5
|
| 19 |
},
|
| 20 |
{
|
| 21 |
-
"epoch": 0.
|
| 22 |
-
"grad_norm": 6.
|
| 23 |
-
"learning_rate":
|
| 24 |
-
"loss": 0.
|
| 25 |
-
"num_tokens":
|
| 26 |
"step": 10
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
-
"grad_norm":
|
| 31 |
-
"learning_rate": 1.
|
| 32 |
-
"loss": 0.
|
| 33 |
-
"num_tokens":
|
| 34 |
"step": 15
|
| 35 |
},
|
| 36 |
{
|
| 37 |
-
"epoch": 0.
|
| 38 |
-
"grad_norm": 1.
|
| 39 |
-
"learning_rate": 1.
|
| 40 |
-
"loss": 0.
|
| 41 |
-
"num_tokens":
|
| 42 |
"step": 20
|
| 43 |
},
|
| 44 |
{
|
| 45 |
-
"epoch": 0.
|
| 46 |
-
"grad_norm":
|
| 47 |
-
"learning_rate":
|
| 48 |
-
"loss": 0.
|
| 49 |
-
"num_tokens":
|
| 50 |
"step": 25
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"grad_norm":
|
| 55 |
-
"learning_rate": 2.
|
| 56 |
-
"loss": 0.
|
| 57 |
-
"num_tokens":
|
| 58 |
"step": 30
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm": 1.
|
| 63 |
-
"learning_rate":
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"num_tokens":
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
-
"epoch": 0.
|
| 70 |
-
"grad_norm": 1.
|
| 71 |
-
"learning_rate":
|
| 72 |
-
"loss": 0.
|
| 73 |
-
"num_tokens":
|
| 74 |
"step": 40
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch": 0.
|
| 78 |
-
"grad_norm": 0.
|
| 79 |
-
"learning_rate":
|
| 80 |
-
"loss": 0.
|
| 81 |
-
"num_tokens":
|
| 82 |
"step": 45
|
| 83 |
},
|
| 84 |
{
|
| 85 |
-
"epoch": 0.
|
| 86 |
-
"grad_norm": 0.
|
| 87 |
-
"learning_rate":
|
| 88 |
-
"loss": 0.
|
| 89 |
-
"num_tokens":
|
| 90 |
"step": 50
|
| 91 |
},
|
| 92 |
{
|
| 93 |
-
"epoch": 0.
|
| 94 |
-
"grad_norm":
|
| 95 |
-
"learning_rate":
|
| 96 |
-
"loss": 0.
|
| 97 |
-
"num_tokens":
|
| 98 |
"step": 55
|
| 99 |
},
|
| 100 |
{
|
| 101 |
-
"epoch": 0.
|
| 102 |
-
"grad_norm":
|
| 103 |
-
"learning_rate": 4.
|
| 104 |
-
"loss": 0.
|
| 105 |
-
"num_tokens":
|
| 106 |
"step": 60
|
| 107 |
},
|
| 108 |
{
|
| 109 |
-
"epoch": 0.
|
| 110 |
-
"grad_norm": 0.
|
| 111 |
-
"learning_rate": 4.
|
| 112 |
-
"loss": 0.
|
| 113 |
-
"num_tokens":
|
| 114 |
"step": 65
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm": 0.
|
| 119 |
-
"learning_rate": 4.
|
| 120 |
-
"loss": 0.
|
| 121 |
-
"num_tokens":
|
| 122 |
"step": 70
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch": 0.
|
| 126 |
-
"grad_norm": 0.
|
| 127 |
-
"learning_rate": 4.
|
| 128 |
-
"loss": 0.
|
| 129 |
-
"num_tokens":
|
| 130 |
"step": 75
|
| 131 |
},
|
| 132 |
{
|
| 133 |
-
"epoch": 0.
|
| 134 |
-
"grad_norm": 0.
|
| 135 |
-
"learning_rate": 4.
|
| 136 |
-
"loss": 0.
|
| 137 |
-
"num_tokens":
|
| 138 |
"step": 80
|
| 139 |
},
|
| 140 |
{
|
| 141 |
-
"epoch": 0.
|
| 142 |
-
"grad_norm": 0.
|
| 143 |
-
"learning_rate": 4.
|
| 144 |
-
"loss": 0.
|
| 145 |
-
"num_tokens":
|
| 146 |
"step": 85
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch": 0.
|
| 150 |
-
"grad_norm": 0.
|
| 151 |
-
"learning_rate": 4.
|
| 152 |
-
"loss": 0.
|
| 153 |
-
"num_tokens":
|
| 154 |
"step": 90
|
| 155 |
},
|
| 156 |
{
|
| 157 |
-
"epoch": 0.
|
| 158 |
-
"grad_norm": 0.
|
| 159 |
-
"learning_rate": 4.
|
| 160 |
-
"loss": 0.
|
| 161 |
-
"num_tokens":
|
| 162 |
"step": 95
|
| 163 |
},
|
| 164 |
{
|
| 165 |
-
"epoch": 0.
|
| 166 |
-
"grad_norm": 0.
|
| 167 |
-
"learning_rate": 4.
|
| 168 |
"loss": 0.0638,
|
| 169 |
-
"num_tokens":
|
| 170 |
"step": 100
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm":
|
| 175 |
-
"learning_rate": 4.
|
| 176 |
-
"loss": 0.
|
| 177 |
-
"num_tokens":
|
| 178 |
"step": 105
|
| 179 |
},
|
| 180 |
{
|
| 181 |
-
"epoch": 0.
|
| 182 |
-
"grad_norm": 0.
|
| 183 |
-
"learning_rate": 4.
|
| 184 |
-
"loss": 0.
|
| 185 |
-
"num_tokens":
|
| 186 |
"step": 110
|
| 187 |
},
|
| 188 |
{
|
| 189 |
-
"epoch": 0.
|
| 190 |
-
"grad_norm": 0.
|
| 191 |
-
"learning_rate": 4.
|
| 192 |
-
"loss": 0.
|
| 193 |
-
"num_tokens":
|
| 194 |
"step": 115
|
| 195 |
},
|
| 196 |
{
|
| 197 |
-
"epoch": 0.
|
| 198 |
-
"grad_norm": 0.
|
| 199 |
-
"learning_rate": 4.
|
| 200 |
-
"loss": 0.
|
| 201 |
-
"num_tokens":
|
| 202 |
"step": 120
|
| 203 |
},
|
| 204 |
{
|
| 205 |
-
"epoch": 0.
|
| 206 |
-
"grad_norm": 0.
|
| 207 |
-
"learning_rate": 4.
|
| 208 |
-
"loss": 0.
|
| 209 |
-
"num_tokens":
|
| 210 |
"step": 125
|
| 211 |
},
|
| 212 |
{
|
| 213 |
-
"epoch": 0.
|
| 214 |
-
"grad_norm":
|
| 215 |
-
"learning_rate": 4.
|
| 216 |
-
"loss": 0.
|
| 217 |
-
"num_tokens":
|
| 218 |
"step": 130
|
| 219 |
},
|
| 220 |
{
|
| 221 |
-
"epoch": 0.
|
| 222 |
-
"grad_norm":
|
| 223 |
-
"learning_rate": 4.
|
| 224 |
-
"loss": 0.
|
| 225 |
-
"num_tokens":
|
| 226 |
"step": 135
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch": 0.
|
| 230 |
-
"grad_norm": 0.
|
| 231 |
-
"learning_rate": 4.
|
| 232 |
-
"loss": 0.
|
| 233 |
-
"num_tokens":
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"grad_norm": 0.
|
| 239 |
-
"learning_rate": 4.
|
| 240 |
-
"loss": 0.
|
| 241 |
-
"num_tokens":
|
| 242 |
"step": 145
|
| 243 |
},
|
| 244 |
{
|
| 245 |
-
"epoch": 0.
|
| 246 |
-
"grad_norm": 0.
|
| 247 |
-
"learning_rate": 4.
|
| 248 |
-
"loss": 0.
|
| 249 |
-
"num_tokens":
|
| 250 |
"step": 150
|
| 251 |
},
|
| 252 |
{
|
| 253 |
-
"epoch": 0.
|
| 254 |
-
"grad_norm": 0.
|
| 255 |
-
"learning_rate": 4.
|
| 256 |
-
"loss": 0.
|
| 257 |
-
"num_tokens":
|
| 258 |
"step": 155
|
| 259 |
},
|
| 260 |
{
|
| 261 |
-
"epoch": 0.
|
| 262 |
-
"grad_norm": 0.
|
| 263 |
-
"learning_rate": 4.
|
| 264 |
-
"loss": 0.
|
| 265 |
-
"num_tokens":
|
| 266 |
"step": 160
|
| 267 |
},
|
| 268 |
{
|
| 269 |
-
"epoch": 0.
|
| 270 |
-
"grad_norm": 0.
|
| 271 |
-
"learning_rate": 4.
|
| 272 |
-
"loss": 0.
|
| 273 |
-
"num_tokens":
|
| 274 |
"step": 165
|
| 275 |
},
|
| 276 |
{
|
| 277 |
-
"epoch": 0.
|
| 278 |
-
"grad_norm": 0.
|
| 279 |
-
"learning_rate": 4.
|
| 280 |
-
"loss": 0.
|
| 281 |
-
"num_tokens":
|
| 282 |
"step": 170
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch": 0.
|
| 286 |
-
"grad_norm": 0.
|
| 287 |
-
"learning_rate": 4.
|
| 288 |
-
"loss": 0.
|
| 289 |
-
"num_tokens":
|
| 290 |
"step": 175
|
| 291 |
},
|
| 292 |
{
|
| 293 |
-
"epoch": 0.
|
| 294 |
-
"grad_norm": 0.
|
| 295 |
-
"learning_rate": 4.
|
| 296 |
-
"loss": 0.
|
| 297 |
-
"num_tokens":
|
| 298 |
"step": 180
|
| 299 |
},
|
| 300 |
{
|
| 301 |
-
"epoch": 0.
|
| 302 |
-
"grad_norm": 0.
|
| 303 |
-
"learning_rate": 4.
|
| 304 |
-
"loss": 0.
|
| 305 |
-
"num_tokens":
|
| 306 |
"step": 185
|
| 307 |
},
|
| 308 |
{
|
| 309 |
-
"epoch": 0.
|
| 310 |
-
"grad_norm": 0.
|
| 311 |
-
"learning_rate": 4.
|
| 312 |
-
"loss": 0.
|
| 313 |
-
"num_tokens":
|
| 314 |
"step": 190
|
| 315 |
},
|
| 316 |
{
|
| 317 |
-
"epoch": 0.
|
| 318 |
-
"grad_norm": 0.
|
| 319 |
-
"learning_rate": 4.
|
| 320 |
-
"loss": 0.
|
| 321 |
-
"num_tokens":
|
| 322 |
"step": 195
|
| 323 |
},
|
| 324 |
{
|
| 325 |
-
"epoch": 0.
|
| 326 |
-
"grad_norm": 0.
|
| 327 |
-
"learning_rate": 4.
|
| 328 |
-
"loss": 0.
|
| 329 |
-
"num_tokens":
|
| 330 |
"step": 200
|
| 331 |
},
|
| 332 |
{
|
| 333 |
-
"epoch": 0.
|
| 334 |
-
"grad_norm": 0.
|
| 335 |
-
"learning_rate": 4.
|
| 336 |
-
"loss": 0.
|
| 337 |
-
"num_tokens":
|
| 338 |
"step": 205
|
| 339 |
},
|
| 340 |
{
|
| 341 |
-
"epoch": 0.
|
| 342 |
-
"grad_norm": 0.
|
| 343 |
-
"learning_rate": 4.
|
| 344 |
-
"loss": 0.
|
| 345 |
-
"num_tokens":
|
| 346 |
"step": 210
|
| 347 |
},
|
| 348 |
{
|
| 349 |
-
"epoch": 0.
|
| 350 |
-
"grad_norm": 0.
|
| 351 |
-
"learning_rate": 4.
|
| 352 |
-
"loss": 0.
|
| 353 |
-
"num_tokens":
|
| 354 |
"step": 215
|
| 355 |
},
|
| 356 |
{
|
| 357 |
-
"epoch": 0.
|
| 358 |
-
"grad_norm": 0.
|
| 359 |
-
"learning_rate": 4.
|
| 360 |
-
"loss": 0.
|
| 361 |
-
"num_tokens":
|
| 362 |
"step": 220
|
| 363 |
},
|
| 364 |
{
|
| 365 |
-
"epoch": 0.
|
| 366 |
-
"grad_norm": 0.
|
| 367 |
-
"learning_rate": 4.
|
| 368 |
-
"loss": 0.
|
| 369 |
-
"num_tokens":
|
| 370 |
"step": 225
|
| 371 |
},
|
| 372 |
{
|
| 373 |
-
"epoch": 0.
|
| 374 |
-
"grad_norm": 0.
|
| 375 |
-
"learning_rate": 4.
|
| 376 |
-
"loss": 0.
|
| 377 |
-
"num_tokens":
|
| 378 |
"step": 230
|
| 379 |
},
|
| 380 |
{
|
| 381 |
-
"epoch": 0.
|
| 382 |
-
"grad_norm": 0.
|
| 383 |
-
"learning_rate": 4.
|
| 384 |
-
"loss": 0.
|
| 385 |
-
"num_tokens":
|
| 386 |
"step": 235
|
| 387 |
},
|
| 388 |
{
|
| 389 |
-
"epoch": 0.
|
| 390 |
-
"grad_norm": 0.
|
| 391 |
-
"learning_rate": 4.
|
| 392 |
-
"loss": 0.
|
| 393 |
-
"num_tokens":
|
| 394 |
"step": 240
|
| 395 |
},
|
| 396 |
{
|
| 397 |
-
"epoch": 0.
|
| 398 |
-
"grad_norm": 0.
|
| 399 |
-
"learning_rate": 4.
|
| 400 |
-
"loss": 0.
|
| 401 |
-
"num_tokens":
|
| 402 |
"step": 245
|
| 403 |
},
|
| 404 |
{
|
| 405 |
-
"epoch": 0.
|
| 406 |
-
"grad_norm": 0.
|
| 407 |
-
"learning_rate": 4.
|
| 408 |
-
"loss": 0.
|
| 409 |
-
"num_tokens":
|
| 410 |
"step": 250
|
| 411 |
},
|
| 412 |
{
|
| 413 |
-
"epoch": 0.
|
| 414 |
-
"grad_norm": 0.
|
| 415 |
-
"learning_rate": 4.
|
| 416 |
-
"loss": 0.
|
| 417 |
-
"num_tokens":
|
| 418 |
"step": 255
|
| 419 |
},
|
| 420 |
{
|
| 421 |
-
"epoch": 0.
|
| 422 |
-
"grad_norm": 0.
|
| 423 |
-
"learning_rate": 4.
|
| 424 |
-
"loss": 0.
|
| 425 |
-
"num_tokens":
|
| 426 |
"step": 260
|
| 427 |
},
|
| 428 |
{
|
| 429 |
-
"epoch": 0.
|
| 430 |
-
"grad_norm": 0.
|
| 431 |
-
"learning_rate": 4.
|
| 432 |
-
"loss": 0.
|
| 433 |
-
"num_tokens":
|
| 434 |
"step": 265
|
| 435 |
},
|
| 436 |
{
|
| 437 |
-
"epoch": 0.
|
| 438 |
-
"grad_norm": 0.
|
| 439 |
-
"learning_rate": 4.
|
| 440 |
-
"loss": 0.
|
| 441 |
-
"num_tokens":
|
| 442 |
"step": 270
|
| 443 |
},
|
| 444 |
{
|
| 445 |
-
"epoch": 0.
|
| 446 |
-
"grad_norm": 0.
|
| 447 |
-
"learning_rate": 4.
|
| 448 |
-
"loss": 0.
|
| 449 |
-
"num_tokens":
|
| 450 |
"step": 275
|
| 451 |
},
|
| 452 |
{
|
| 453 |
-
"epoch": 0.
|
| 454 |
-
"grad_norm": 0.
|
| 455 |
-
"learning_rate": 4.
|
| 456 |
-
"loss": 0.
|
| 457 |
-
"num_tokens":
|
| 458 |
"step": 280
|
| 459 |
},
|
| 460 |
{
|
| 461 |
-
"epoch": 0.
|
| 462 |
-
"grad_norm": 0.
|
| 463 |
-
"learning_rate": 4.
|
| 464 |
-
"loss": 0.
|
| 465 |
-
"num_tokens":
|
| 466 |
"step": 285
|
| 467 |
},
|
| 468 |
{
|
| 469 |
-
"epoch": 0.
|
| 470 |
-
"grad_norm": 0.
|
| 471 |
-
"learning_rate": 4.
|
| 472 |
-
"loss": 0.
|
| 473 |
-
"num_tokens":
|
| 474 |
"step": 290
|
| 475 |
},
|
| 476 |
{
|
| 477 |
-
"epoch": 0.
|
| 478 |
-
"grad_norm": 0.
|
| 479 |
-
"learning_rate": 4.
|
| 480 |
-
"loss": 0.
|
| 481 |
-
"num_tokens":
|
| 482 |
"step": 295
|
| 483 |
},
|
| 484 |
{
|
| 485 |
-
"epoch": 0.
|
| 486 |
-
"grad_norm": 0.
|
| 487 |
-
"learning_rate": 4.
|
| 488 |
-
"loss": 0.
|
| 489 |
-
"num_tokens":
|
| 490 |
"step": 300
|
| 491 |
},
|
| 492 |
{
|
| 493 |
-
"epoch": 0.
|
| 494 |
-
"grad_norm": 0.
|
| 495 |
-
"learning_rate": 4.
|
| 496 |
-
"loss": 0.
|
| 497 |
-
"num_tokens":
|
| 498 |
"step": 305
|
| 499 |
},
|
| 500 |
{
|
| 501 |
-
"epoch": 0.
|
| 502 |
-
"grad_norm": 0.
|
| 503 |
-
"learning_rate": 4.
|
| 504 |
-
"loss": 0.
|
| 505 |
-
"num_tokens":
|
| 506 |
"step": 310
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch": 0.
|
| 510 |
-
"grad_norm": 0.
|
| 511 |
-
"learning_rate": 4.
|
| 512 |
-
"loss": 0.
|
| 513 |
-
"num_tokens":
|
| 514 |
"step": 315
|
| 515 |
},
|
| 516 |
{
|
| 517 |
-
"epoch": 0.
|
| 518 |
-
"grad_norm": 0.
|
| 519 |
-
"learning_rate": 4.
|
| 520 |
-
"loss": 0.
|
| 521 |
-
"num_tokens":
|
| 522 |
"step": 320
|
| 523 |
},
|
| 524 |
{
|
| 525 |
-
"epoch": 0.
|
| 526 |
-
"grad_norm": 0.
|
| 527 |
-
"learning_rate": 4.
|
| 528 |
-
"loss": 0.
|
| 529 |
-
"num_tokens":
|
| 530 |
"step": 325
|
| 531 |
},
|
| 532 |
{
|
| 533 |
-
"epoch": 0.
|
| 534 |
-
"grad_norm": 0.
|
| 535 |
-
"learning_rate": 4.
|
| 536 |
-
"loss": 0.
|
| 537 |
-
"num_tokens":
|
| 538 |
"step": 330
|
| 539 |
},
|
| 540 |
{
|
| 541 |
-
"epoch": 0.
|
| 542 |
-
"grad_norm": 0.
|
| 543 |
-
"learning_rate": 4.
|
| 544 |
-
"loss": 0.
|
| 545 |
-
"num_tokens":
|
| 546 |
"step": 335
|
| 547 |
},
|
| 548 |
{
|
| 549 |
-
"epoch":
|
| 550 |
-
"grad_norm": 0.
|
| 551 |
-
"learning_rate": 4.
|
| 552 |
-
"loss": 0.
|
| 553 |
-
"num_tokens":
|
| 554 |
"step": 340
|
| 555 |
},
|
| 556 |
{
|
| 557 |
-
"epoch":
|
| 558 |
-
"grad_norm": 0.
|
| 559 |
-
"learning_rate": 4.
|
| 560 |
-
"loss": 0.
|
| 561 |
-
"num_tokens":
|
| 562 |
"step": 345
|
| 563 |
},
|
| 564 |
{
|
| 565 |
-
"epoch":
|
| 566 |
-
"grad_norm": 0.
|
| 567 |
-
"learning_rate": 4.
|
| 568 |
-
"loss": 0.
|
| 569 |
-
"num_tokens":
|
| 570 |
"step": 350
|
| 571 |
},
|
| 572 |
{
|
| 573 |
-
"epoch":
|
| 574 |
-
"grad_norm": 0.
|
| 575 |
-
"learning_rate":
|
| 576 |
-
"loss": 0.
|
| 577 |
-
"num_tokens":
|
| 578 |
"step": 355
|
| 579 |
},
|
| 580 |
{
|
| 581 |
-
"epoch":
|
| 582 |
-
"grad_norm": 0.
|
| 583 |
-
"learning_rate":
|
| 584 |
-
"loss": 0.
|
| 585 |
-
"num_tokens":
|
| 586 |
"step": 360
|
| 587 |
},
|
| 588 |
{
|
| 589 |
-
"epoch":
|
| 590 |
-
"grad_norm": 0.
|
| 591 |
-
"learning_rate":
|
| 592 |
-
"loss": 0.
|
| 593 |
-
"num_tokens":
|
| 594 |
"step": 365
|
| 595 |
},
|
| 596 |
{
|
| 597 |
-
"epoch":
|
| 598 |
-
"grad_norm": 0.
|
| 599 |
-
"learning_rate":
|
| 600 |
-
"loss": 0.
|
| 601 |
-
"num_tokens":
|
| 602 |
"step": 370
|
| 603 |
},
|
| 604 |
{
|
| 605 |
-
"epoch":
|
| 606 |
-
"grad_norm": 0.
|
| 607 |
-
"learning_rate":
|
| 608 |
-
"loss": 0.
|
| 609 |
-
"num_tokens":
|
| 610 |
"step": 375
|
| 611 |
},
|
| 612 |
{
|
| 613 |
-
"epoch":
|
| 614 |
-
"grad_norm": 0.
|
| 615 |
-
"learning_rate":
|
| 616 |
-
"loss": 0.
|
| 617 |
-
"num_tokens":
|
| 618 |
"step": 380
|
| 619 |
},
|
| 620 |
{
|
| 621 |
-
"epoch":
|
| 622 |
-
"grad_norm": 0.
|
| 623 |
-
"learning_rate":
|
| 624 |
-
"loss": 0.
|
| 625 |
-
"num_tokens":
|
| 626 |
"step": 385
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch":
|
| 630 |
-
"grad_norm": 0.
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss": 0.
|
| 633 |
-
"num_tokens":
|
| 634 |
"step": 390
|
| 635 |
},
|
| 636 |
{
|
| 637 |
-
"epoch":
|
| 638 |
-
"grad_norm": 0.
|
| 639 |
-
"learning_rate":
|
| 640 |
-
"loss": 0.
|
| 641 |
-
"num_tokens":
|
| 642 |
"step": 395
|
| 643 |
},
|
| 644 |
{
|
| 645 |
-
"epoch":
|
| 646 |
-
"grad_norm": 0.
|
| 647 |
-
"learning_rate":
|
| 648 |
-
"loss": 0.
|
| 649 |
-
"num_tokens":
|
| 650 |
"step": 400
|
| 651 |
},
|
| 652 |
{
|
| 653 |
-
"epoch":
|
| 654 |
-
"grad_norm": 0.
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 0.
|
| 657 |
-
"num_tokens":
|
| 658 |
"step": 405
|
| 659 |
},
|
| 660 |
{
|
| 661 |
-
"epoch":
|
| 662 |
-
"grad_norm": 0.
|
| 663 |
-
"learning_rate":
|
| 664 |
-
"loss": 0.
|
| 665 |
-
"num_tokens":
|
| 666 |
"step": 410
|
| 667 |
},
|
| 668 |
{
|
| 669 |
-
"epoch":
|
| 670 |
-
"grad_norm": 0.
|
| 671 |
-
"learning_rate":
|
| 672 |
-
"loss": 0.
|
| 673 |
-
"num_tokens":
|
| 674 |
"step": 415
|
| 675 |
},
|
| 676 |
{
|
| 677 |
-
"epoch":
|
| 678 |
-
"grad_norm": 0.
|
| 679 |
-
"learning_rate":
|
| 680 |
-
"loss": 0.
|
| 681 |
-
"num_tokens":
|
| 682 |
"step": 420
|
| 683 |
},
|
| 684 |
{
|
| 685 |
-
"epoch":
|
| 686 |
-
"grad_norm": 0.
|
| 687 |
-
"learning_rate":
|
| 688 |
-
"loss": 0.
|
| 689 |
-
"num_tokens":
|
| 690 |
"step": 425
|
| 691 |
},
|
| 692 |
{
|
| 693 |
-
"epoch":
|
| 694 |
-
"grad_norm": 0.
|
| 695 |
-
"learning_rate":
|
| 696 |
-
"loss": 0.
|
| 697 |
-
"num_tokens":
|
| 698 |
"step": 430
|
| 699 |
},
|
| 700 |
{
|
| 701 |
-
"epoch":
|
| 702 |
-
"grad_norm": 0.
|
| 703 |
-
"learning_rate":
|
| 704 |
-
"loss": 0.
|
| 705 |
-
"num_tokens":
|
| 706 |
"step": 435
|
| 707 |
},
|
| 708 |
{
|
| 709 |
-
"epoch":
|
| 710 |
-
"grad_norm":
|
| 711 |
-
"learning_rate":
|
| 712 |
-
"loss": 0.
|
| 713 |
-
"num_tokens":
|
| 714 |
"step": 440
|
| 715 |
},
|
| 716 |
{
|
| 717 |
-
"epoch":
|
| 718 |
-
"grad_norm": 0.
|
| 719 |
-
"learning_rate":
|
| 720 |
-
"loss": 0.
|
| 721 |
-
"num_tokens":
|
| 722 |
"step": 445
|
| 723 |
},
|
| 724 |
{
|
| 725 |
-
"epoch":
|
| 726 |
-
"grad_norm": 0.
|
| 727 |
-
"learning_rate":
|
| 728 |
-
"loss": 0.
|
| 729 |
-
"num_tokens":
|
| 730 |
"step": 450
|
| 731 |
},
|
| 732 |
{
|
| 733 |
-
"epoch":
|
| 734 |
-
"grad_norm": 0.
|
| 735 |
-
"learning_rate":
|
| 736 |
-
"loss": 0.
|
| 737 |
-
"num_tokens":
|
| 738 |
"step": 455
|
| 739 |
},
|
| 740 |
{
|
| 741 |
-
"epoch":
|
| 742 |
-
"grad_norm": 0.
|
| 743 |
-
"learning_rate":
|
| 744 |
-
"loss": 0.
|
| 745 |
-
"num_tokens":
|
| 746 |
"step": 460
|
| 747 |
},
|
| 748 |
{
|
| 749 |
-
"epoch":
|
| 750 |
-
"grad_norm": 0.
|
| 751 |
-
"learning_rate":
|
| 752 |
-
"loss": 0.
|
| 753 |
-
"num_tokens":
|
| 754 |
"step": 465
|
| 755 |
},
|
| 756 |
{
|
| 757 |
-
"epoch":
|
| 758 |
-
"grad_norm": 0.
|
| 759 |
-
"learning_rate":
|
| 760 |
-
"loss": 0.
|
| 761 |
-
"num_tokens":
|
| 762 |
"step": 470
|
| 763 |
},
|
| 764 |
{
|
| 765 |
-
"epoch":
|
| 766 |
-
"grad_norm": 0.
|
| 767 |
-
"learning_rate":
|
| 768 |
-
"loss": 0.
|
| 769 |
-
"num_tokens":
|
| 770 |
"step": 475
|
| 771 |
},
|
| 772 |
{
|
| 773 |
-
"epoch": 1.
|
| 774 |
-
"grad_norm":
|
| 775 |
-
"learning_rate":
|
| 776 |
-
"loss": 0.
|
| 777 |
-
"num_tokens":
|
| 778 |
"step": 480
|
| 779 |
},
|
| 780 |
{
|
| 781 |
-
"epoch": 1.
|
| 782 |
-
"grad_norm": 0.
|
| 783 |
-
"learning_rate":
|
| 784 |
-
"loss": 0.
|
| 785 |
-
"num_tokens":
|
| 786 |
"step": 485
|
| 787 |
},
|
| 788 |
{
|
| 789 |
-
"epoch": 1.
|
| 790 |
-
"grad_norm": 0.
|
| 791 |
-
"learning_rate":
|
| 792 |
-
"loss": 0.
|
| 793 |
-
"num_tokens":
|
| 794 |
"step": 490
|
| 795 |
},
|
| 796 |
{
|
| 797 |
-
"epoch": 1.
|
| 798 |
-
"grad_norm": 0.
|
| 799 |
-
"learning_rate":
|
| 800 |
-
"loss": 0.
|
| 801 |
-
"num_tokens":
|
| 802 |
"step": 495
|
| 803 |
},
|
| 804 |
{
|
| 805 |
-
"epoch": 1.
|
| 806 |
-
"grad_norm": 0.
|
| 807 |
-
"learning_rate":
|
| 808 |
-
"loss": 0.
|
| 809 |
-
"num_tokens":
|
| 810 |
"step": 500
|
| 811 |
},
|
| 812 |
{
|
| 813 |
-
"epoch": 1.
|
| 814 |
-
"grad_norm": 0.
|
| 815 |
-
"learning_rate":
|
| 816 |
-
"loss": 0.
|
| 817 |
-
"num_tokens":
|
| 818 |
"step": 505
|
| 819 |
},
|
| 820 |
{
|
| 821 |
-
"epoch": 1.
|
| 822 |
-
"grad_norm": 0.
|
| 823 |
-
"learning_rate":
|
| 824 |
-
"loss": 0.
|
| 825 |
-
"num_tokens":
|
| 826 |
"step": 510
|
| 827 |
},
|
| 828 |
{
|
| 829 |
-
"epoch": 1.
|
| 830 |
-
"grad_norm": 0.
|
| 831 |
-
"learning_rate":
|
| 832 |
-
"loss": 0.
|
| 833 |
-
"num_tokens":
|
| 834 |
"step": 515
|
| 835 |
},
|
| 836 |
{
|
| 837 |
-
"epoch": 1.
|
| 838 |
-
"grad_norm":
|
| 839 |
-
"learning_rate":
|
| 840 |
-
"loss": 0.
|
| 841 |
-
"num_tokens":
|
| 842 |
"step": 520
|
| 843 |
},
|
| 844 |
{
|
| 845 |
-
"epoch": 1.
|
| 846 |
-
"grad_norm": 0.
|
| 847 |
-
"learning_rate":
|
| 848 |
-
"loss": 0.
|
| 849 |
-
"num_tokens":
|
| 850 |
"step": 525
|
| 851 |
},
|
| 852 |
{
|
| 853 |
-
"epoch": 1.
|
| 854 |
-
"grad_norm": 0.
|
| 855 |
-
"learning_rate":
|
| 856 |
-
"loss": 0.
|
| 857 |
-
"num_tokens":
|
| 858 |
"step": 530
|
| 859 |
},
|
| 860 |
{
|
| 861 |
-
"epoch": 1.
|
| 862 |
-
"grad_norm": 0.
|
| 863 |
-
"learning_rate":
|
| 864 |
-
"loss": 0.
|
| 865 |
-
"num_tokens":
|
| 866 |
"step": 535
|
| 867 |
},
|
| 868 |
{
|
| 869 |
-
"epoch": 1.
|
| 870 |
-
"grad_norm": 0.
|
| 871 |
-
"learning_rate":
|
| 872 |
-
"loss": 0.
|
| 873 |
-
"num_tokens":
|
| 874 |
"step": 540
|
| 875 |
},
|
| 876 |
{
|
| 877 |
-
"epoch": 1.
|
| 878 |
-
"grad_norm": 0.
|
| 879 |
-
"learning_rate":
|
| 880 |
-
"loss": 0.
|
| 881 |
-
"num_tokens":
|
| 882 |
"step": 545
|
| 883 |
},
|
| 884 |
{
|
| 885 |
-
"epoch": 1.
|
| 886 |
-
"grad_norm":
|
| 887 |
-
"learning_rate":
|
| 888 |
-
"loss": 0.
|
| 889 |
-
"num_tokens":
|
| 890 |
"step": 550
|
| 891 |
},
|
| 892 |
{
|
| 893 |
-
"epoch": 1.
|
| 894 |
-
"grad_norm": 0.
|
| 895 |
-
"learning_rate":
|
| 896 |
-
"loss": 0.
|
| 897 |
-
"num_tokens":
|
| 898 |
"step": 555
|
| 899 |
},
|
| 900 |
{
|
| 901 |
-
"epoch": 1.
|
| 902 |
-
"grad_norm": 0.
|
| 903 |
-
"learning_rate":
|
| 904 |
-
"loss": 0.
|
| 905 |
-
"num_tokens":
|
| 906 |
"step": 560
|
| 907 |
},
|
| 908 |
{
|
| 909 |
-
"epoch": 1.
|
| 910 |
-
"grad_norm": 0.
|
| 911 |
-
"learning_rate":
|
| 912 |
-
"loss": 0.
|
| 913 |
-
"num_tokens":
|
| 914 |
"step": 565
|
| 915 |
},
|
| 916 |
{
|
| 917 |
-
"epoch": 1.
|
| 918 |
-
"grad_norm": 0.
|
| 919 |
-
"learning_rate":
|
| 920 |
-
"loss": 0.
|
| 921 |
-
"num_tokens":
|
| 922 |
"step": 570
|
| 923 |
},
|
| 924 |
{
|
| 925 |
-
"epoch": 1.
|
| 926 |
-
"grad_norm": 0.
|
| 927 |
-
"learning_rate":
|
| 928 |
-
"loss": 0.
|
| 929 |
-
"num_tokens":
|
| 930 |
"step": 575
|
| 931 |
},
|
| 932 |
{
|
| 933 |
-
"epoch": 1.
|
| 934 |
-
"grad_norm": 0.
|
| 935 |
-
"learning_rate":
|
| 936 |
-
"loss": 0.
|
| 937 |
-
"num_tokens":
|
| 938 |
"step": 580
|
| 939 |
},
|
| 940 |
{
|
| 941 |
-
"epoch": 1.
|
| 942 |
-
"grad_norm": 0.
|
| 943 |
-
"learning_rate":
|
| 944 |
-
"loss": 0.
|
| 945 |
-
"num_tokens":
|
| 946 |
"step": 585
|
| 947 |
},
|
| 948 |
{
|
| 949 |
-
"epoch": 1.
|
| 950 |
-
"grad_norm": 0.
|
| 951 |
-
"learning_rate":
|
| 952 |
-
"loss": 0.
|
| 953 |
-
"num_tokens":
|
| 954 |
"step": 590
|
| 955 |
},
|
| 956 |
{
|
| 957 |
-
"epoch": 1.
|
| 958 |
-
"grad_norm": 0.
|
| 959 |
-
"learning_rate":
|
| 960 |
-
"loss": 0.
|
| 961 |
-
"num_tokens":
|
| 962 |
"step": 595
|
| 963 |
},
|
| 964 |
{
|
| 965 |
-
"epoch": 1.
|
| 966 |
-
"grad_norm": 0.
|
| 967 |
-
"learning_rate":
|
| 968 |
-
"loss": 0.
|
| 969 |
-
"num_tokens":
|
| 970 |
"step": 600
|
| 971 |
},
|
| 972 |
{
|
| 973 |
-
"epoch": 1.
|
| 974 |
-
"grad_norm": 0.
|
| 975 |
-
"learning_rate":
|
| 976 |
-
"loss": 0.
|
| 977 |
-
"num_tokens":
|
| 978 |
"step": 605
|
| 979 |
},
|
| 980 |
{
|
| 981 |
-
"epoch": 1.
|
| 982 |
-
"grad_norm": 0.
|
| 983 |
-
"learning_rate":
|
| 984 |
-
"loss": 0.
|
| 985 |
-
"num_tokens":
|
| 986 |
"step": 610
|
| 987 |
},
|
| 988 |
{
|
| 989 |
-
"epoch": 1.
|
| 990 |
-
"grad_norm": 0.
|
| 991 |
-
"learning_rate":
|
| 992 |
-
"loss": 0.
|
| 993 |
-
"num_tokens":
|
| 994 |
"step": 615
|
| 995 |
},
|
| 996 |
{
|
| 997 |
-
"epoch": 1.
|
| 998 |
-
"grad_norm": 0.
|
| 999 |
-
"learning_rate":
|
| 1000 |
-
"loss": 0.
|
| 1001 |
-
"num_tokens":
|
| 1002 |
"step": 620
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
-
"epoch": 1.
|
| 1006 |
-
"grad_norm": 0.
|
| 1007 |
-
"learning_rate":
|
| 1008 |
-
"loss": 0.
|
| 1009 |
-
"num_tokens":
|
| 1010 |
"step": 625
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
-
"epoch": 1.
|
| 1014 |
-
"grad_norm": 0.
|
| 1015 |
-
"learning_rate":
|
| 1016 |
-
"loss": 0.
|
| 1017 |
-
"num_tokens":
|
| 1018 |
"step": 630
|
| 1019 |
},
|
| 1020 |
{
|
| 1021 |
-
"epoch": 1.
|
| 1022 |
-
"grad_norm": 0.
|
| 1023 |
-
"learning_rate":
|
| 1024 |
-
"loss": 0.
|
| 1025 |
-
"num_tokens":
|
| 1026 |
"step": 635
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
-
"epoch": 1.
|
| 1030 |
-
"grad_norm": 0.
|
| 1031 |
-
"learning_rate":
|
| 1032 |
-
"loss": 0.
|
| 1033 |
-
"num_tokens":
|
| 1034 |
"step": 640
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
-
"epoch": 1.
|
| 1038 |
-
"grad_norm": 0.
|
| 1039 |
-
"learning_rate":
|
| 1040 |
-
"loss": 0.
|
| 1041 |
-
"num_tokens":
|
| 1042 |
"step": 645
|
| 1043 |
},
|
| 1044 |
{
|
| 1045 |
-
"epoch": 1.
|
| 1046 |
-
"grad_norm": 0.
|
| 1047 |
-
"learning_rate":
|
| 1048 |
-
"loss": 0.
|
| 1049 |
-
"num_tokens":
|
| 1050 |
"step": 650
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
-
"epoch": 1.
|
| 1054 |
-
"grad_norm": 0.
|
| 1055 |
-
"learning_rate":
|
| 1056 |
-
"loss": 0.
|
| 1057 |
-
"num_tokens":
|
| 1058 |
"step": 655
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
-
"epoch": 1.
|
| 1062 |
-
"grad_norm": 0.
|
| 1063 |
-
"learning_rate":
|
| 1064 |
-
"loss": 0.
|
| 1065 |
-
"num_tokens":
|
| 1066 |
"step": 660
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
-
"epoch": 1.
|
| 1070 |
-
"grad_norm": 0.
|
| 1071 |
-
"learning_rate":
|
| 1072 |
-
"loss": 0.
|
| 1073 |
-
"num_tokens":
|
| 1074 |
"step": 665
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
-
"epoch": 1.
|
| 1078 |
-
"grad_norm": 0.
|
| 1079 |
-
"learning_rate":
|
| 1080 |
-
"loss": 0.
|
| 1081 |
-
"num_tokens":
|
| 1082 |
"step": 670
|
| 1083 |
},
|
| 1084 |
{
|
| 1085 |
-
"epoch": 1.
|
| 1086 |
-
"grad_norm": 0.
|
| 1087 |
-
"learning_rate":
|
| 1088 |
-
"loss": 0.
|
| 1089 |
-
"num_tokens":
|
| 1090 |
"step": 675
|
| 1091 |
},
|
| 1092 |
{
|
| 1093 |
-
"epoch":
|
| 1094 |
-
"grad_norm": 0.
|
| 1095 |
-
"learning_rate":
|
| 1096 |
-
"loss": 0.
|
| 1097 |
-
"num_tokens":
|
| 1098 |
"step": 680
|
| 1099 |
},
|
| 1100 |
{
|
| 1101 |
-
"epoch":
|
| 1102 |
-
"grad_norm": 0.
|
| 1103 |
-
"learning_rate":
|
| 1104 |
-
"loss": 0.
|
| 1105 |
-
"num_tokens":
|
| 1106 |
"step": 685
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
-
"epoch":
|
| 1110 |
-
"grad_norm": 0.
|
| 1111 |
-
"learning_rate":
|
| 1112 |
-
"loss": 0.
|
| 1113 |
-
"num_tokens":
|
| 1114 |
"step": 690
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
-
"epoch":
|
| 1118 |
-
"grad_norm": 0.
|
| 1119 |
-
"learning_rate":
|
| 1120 |
-
"loss": 0.
|
| 1121 |
-
"num_tokens":
|
| 1122 |
"step": 695
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
-
"epoch":
|
| 1126 |
-
"grad_norm": 0.
|
| 1127 |
-
"learning_rate":
|
| 1128 |
-
"loss": 0.
|
| 1129 |
-
"num_tokens":
|
| 1130 |
"step": 700
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
-
"epoch":
|
| 1134 |
-
"grad_norm": 0.
|
| 1135 |
-
"learning_rate":
|
| 1136 |
-
"loss": 0.
|
| 1137 |
-
"num_tokens":
|
| 1138 |
"step": 705
|
| 1139 |
},
|
| 1140 |
{
|
| 1141 |
-
"epoch":
|
| 1142 |
-
"grad_norm": 0.
|
| 1143 |
-
"learning_rate":
|
| 1144 |
-
"loss": 0.
|
| 1145 |
-
"num_tokens":
|
| 1146 |
"step": 710
|
| 1147 |
},
|
| 1148 |
{
|
| 1149 |
-
"epoch":
|
| 1150 |
-
"grad_norm": 0.
|
| 1151 |
-
"learning_rate":
|
| 1152 |
-
"loss": 0.
|
| 1153 |
-
"num_tokens":
|
| 1154 |
"step": 715
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
-
"epoch":
|
| 1158 |
-
"grad_norm": 0.
|
| 1159 |
-
"learning_rate":
|
| 1160 |
-
"loss": 0.
|
| 1161 |
-
"num_tokens":
|
| 1162 |
"step": 720
|
| 1163 |
},
|
| 1164 |
{
|
| 1165 |
-
"epoch":
|
| 1166 |
-
"grad_norm": 0.
|
| 1167 |
-
"learning_rate":
|
| 1168 |
-
"loss": 0.
|
| 1169 |
-
"num_tokens":
|
| 1170 |
"step": 725
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
-
"epoch":
|
| 1174 |
-
"grad_norm": 0.
|
| 1175 |
-
"learning_rate":
|
| 1176 |
-
"loss": 0.
|
| 1177 |
-
"num_tokens":
|
| 1178 |
"step": 730
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
-
"epoch":
|
| 1182 |
-
"grad_norm": 0.
|
| 1183 |
-
"learning_rate":
|
| 1184 |
-
"loss": 0.
|
| 1185 |
-
"num_tokens":
|
| 1186 |
"step": 735
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
-
"epoch":
|
| 1190 |
-
"grad_norm": 0.
|
| 1191 |
-
"learning_rate":
|
| 1192 |
-
"loss": 0.
|
| 1193 |
-
"num_tokens":
|
| 1194 |
"step": 740
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
-
"epoch":
|
| 1198 |
-
"grad_norm": 0.
|
| 1199 |
-
"learning_rate":
|
| 1200 |
-
"loss": 0.
|
| 1201 |
-
"num_tokens":
|
| 1202 |
"step": 745
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
-
"epoch":
|
| 1206 |
-
"grad_norm": 0.
|
| 1207 |
-
"learning_rate":
|
| 1208 |
-
"loss": 0.
|
| 1209 |
-
"num_tokens":
|
| 1210 |
"step": 750
|
| 1211 |
},
|
| 1212 |
{
|
| 1213 |
-
"epoch":
|
| 1214 |
-
"grad_norm": 0.
|
| 1215 |
-
"learning_rate":
|
| 1216 |
-
"loss": 0.
|
| 1217 |
-
"num_tokens":
|
| 1218 |
"step": 755
|
| 1219 |
},
|
| 1220 |
{
|
| 1221 |
-
"epoch":
|
| 1222 |
-
"grad_norm": 0.
|
| 1223 |
-
"learning_rate":
|
| 1224 |
-
"loss": 0.
|
| 1225 |
-
"num_tokens":
|
| 1226 |
"step": 760
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
-
"epoch":
|
| 1230 |
-
"grad_norm": 0.
|
| 1231 |
-
"learning_rate":
|
| 1232 |
-
"loss": 0.
|
| 1233 |
-
"num_tokens":
|
| 1234 |
"step": 765
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
-
"epoch":
|
| 1238 |
-
"grad_norm": 0.
|
| 1239 |
-
"learning_rate":
|
| 1240 |
-
"loss": 0.
|
| 1241 |
-
"num_tokens":
|
| 1242 |
"step": 770
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
-
"epoch":
|
| 1246 |
-
"grad_norm": 0.
|
| 1247 |
-
"learning_rate":
|
| 1248 |
-
"loss": 0.
|
| 1249 |
-
"num_tokens":
|
| 1250 |
"step": 775
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
-
"epoch":
|
| 1254 |
-
"grad_norm": 0.
|
| 1255 |
-
"learning_rate":
|
| 1256 |
-
"loss": 0.
|
| 1257 |
-
"num_tokens":
|
| 1258 |
"step": 780
|
| 1259 |
},
|
| 1260 |
{
|
| 1261 |
-
"epoch":
|
| 1262 |
-
"grad_norm": 0.
|
| 1263 |
-
"learning_rate":
|
| 1264 |
-
"loss": 0.
|
| 1265 |
-
"num_tokens":
|
| 1266 |
"step": 785
|
| 1267 |
},
|
| 1268 |
{
|
| 1269 |
-
"epoch":
|
| 1270 |
-
"grad_norm": 0.
|
| 1271 |
-
"learning_rate":
|
| 1272 |
-
"loss": 0.
|
| 1273 |
-
"num_tokens":
|
| 1274 |
"step": 790
|
| 1275 |
},
|
| 1276 |
{
|
| 1277 |
-
"epoch":
|
| 1278 |
-
"grad_norm": 0.
|
| 1279 |
-
"learning_rate":
|
| 1280 |
-
"loss": 0.
|
| 1281 |
-
"num_tokens":
|
| 1282 |
"step": 795
|
| 1283 |
},
|
| 1284 |
{
|
| 1285 |
-
"epoch":
|
| 1286 |
-
"grad_norm": 0.
|
| 1287 |
-
"learning_rate":
|
| 1288 |
-
"loss": 0.
|
| 1289 |
-
"num_tokens":
|
| 1290 |
"step": 800
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
-
"epoch":
|
| 1294 |
-
"grad_norm": 0.
|
| 1295 |
-
"learning_rate":
|
| 1296 |
-
"loss": 0.
|
| 1297 |
-
"num_tokens":
|
| 1298 |
"step": 805
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
-
"epoch":
|
| 1302 |
-
"grad_norm": 0.
|
| 1303 |
-
"learning_rate":
|
| 1304 |
-
"loss": 0.
|
| 1305 |
-
"num_tokens":
|
| 1306 |
"step": 810
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
-
"epoch":
|
| 1310 |
-
"grad_norm": 0.
|
| 1311 |
-
"learning_rate":
|
| 1312 |
-
"loss": 0.
|
| 1313 |
-
"num_tokens":
|
| 1314 |
"step": 815
|
| 1315 |
},
|
| 1316 |
{
|
| 1317 |
-
"epoch":
|
| 1318 |
-
"grad_norm": 0.
|
| 1319 |
-
"learning_rate":
|
| 1320 |
-
"loss": 0.
|
| 1321 |
-
"num_tokens":
|
| 1322 |
"step": 820
|
| 1323 |
},
|
| 1324 |
{
|
| 1325 |
-
"epoch":
|
| 1326 |
-
"grad_norm": 0.
|
| 1327 |
-
"learning_rate":
|
| 1328 |
-
"loss": 0.
|
| 1329 |
-
"num_tokens":
|
| 1330 |
"step": 825
|
| 1331 |
},
|
| 1332 |
{
|
| 1333 |
-
"epoch":
|
| 1334 |
-
"grad_norm": 0.
|
| 1335 |
-
"learning_rate":
|
| 1336 |
-
"loss": 0.
|
| 1337 |
-
"num_tokens":
|
| 1338 |
"step": 830
|
| 1339 |
},
|
| 1340 |
{
|
| 1341 |
-
"epoch":
|
| 1342 |
-
"grad_norm": 0.
|
| 1343 |
-
"learning_rate":
|
| 1344 |
-
"loss": 0.
|
| 1345 |
-
"num_tokens":
|
| 1346 |
"step": 835
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
-
"epoch":
|
| 1350 |
-
"grad_norm": 0.
|
| 1351 |
-
"learning_rate":
|
| 1352 |
-
"loss": 0.
|
| 1353 |
-
"num_tokens":
|
| 1354 |
"step": 840
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
-
"epoch":
|
| 1358 |
-
"grad_norm": 0.
|
| 1359 |
-
"learning_rate":
|
| 1360 |
-
"loss": 0.
|
| 1361 |
-
"num_tokens":
|
| 1362 |
"step": 845
|
| 1363 |
},
|
| 1364 |
{
|
| 1365 |
-
"epoch":
|
| 1366 |
-
"grad_norm": 0.
|
| 1367 |
-
"learning_rate":
|
| 1368 |
-
"loss": 0.
|
| 1369 |
-
"num_tokens":
|
| 1370 |
"step": 850
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
-
"epoch":
|
| 1374 |
-
"grad_norm": 0.
|
| 1375 |
-
"learning_rate":
|
| 1376 |
-
"loss": 0.
|
| 1377 |
-
"num_tokens":
|
| 1378 |
"step": 855
|
| 1379 |
},
|
| 1380 |
{
|
| 1381 |
-
"epoch":
|
| 1382 |
-
"grad_norm": 0.
|
| 1383 |
-
"learning_rate":
|
| 1384 |
-
"loss": 0.
|
| 1385 |
-
"num_tokens":
|
| 1386 |
"step": 860
|
| 1387 |
},
|
| 1388 |
{
|
| 1389 |
-
"epoch":
|
| 1390 |
-
"grad_norm": 0.
|
| 1391 |
-
"learning_rate":
|
| 1392 |
-
"loss": 0.
|
| 1393 |
-
"num_tokens":
|
| 1394 |
"step": 865
|
| 1395 |
},
|
| 1396 |
{
|
| 1397 |
-
"epoch":
|
| 1398 |
-
"grad_norm": 0.
|
| 1399 |
-
"learning_rate":
|
| 1400 |
-
"loss": 0.
|
| 1401 |
-
"num_tokens":
|
| 1402 |
"step": 870
|
| 1403 |
},
|
| 1404 |
{
|
| 1405 |
-
"epoch":
|
| 1406 |
-
"grad_norm": 0.
|
| 1407 |
-
"learning_rate":
|
| 1408 |
-
"loss": 0.
|
| 1409 |
-
"num_tokens":
|
| 1410 |
"step": 875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
-
"epoch":
|
| 1414 |
-
"grad_norm": 0.
|
| 1415 |
-
"learning_rate":
|
| 1416 |
-
"loss": 0.
|
| 1417 |
-
"num_tokens":
|
| 1418 |
"step": 880
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
-
"epoch":
|
| 1422 |
-
"grad_norm": 0.
|
| 1423 |
-
"learning_rate":
|
| 1424 |
-
"loss": 0.
|
| 1425 |
-
"num_tokens":
|
| 1426 |
"step": 885
|
| 1427 |
},
|
| 1428 |
{
|
| 1429 |
-
"epoch":
|
| 1430 |
-
"grad_norm": 0.
|
| 1431 |
-
"learning_rate":
|
| 1432 |
-
"loss": 0.
|
| 1433 |
-
"num_tokens":
|
| 1434 |
"step": 890
|
| 1435 |
},
|
| 1436 |
{
|
| 1437 |
-
"epoch":
|
| 1438 |
-
"grad_norm": 0.
|
| 1439 |
-
"learning_rate":
|
| 1440 |
-
"loss": 0.
|
| 1441 |
-
"num_tokens":
|
| 1442 |
"step": 895
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
-
"epoch":
|
| 1446 |
-
"grad_norm": 0.
|
| 1447 |
-
"learning_rate":
|
| 1448 |
-
"loss": 0.
|
| 1449 |
-
"num_tokens":
|
| 1450 |
"step": 900
|
| 1451 |
},
|
| 1452 |
{
|
| 1453 |
-
"epoch":
|
| 1454 |
-
"grad_norm": 0.
|
| 1455 |
-
"learning_rate":
|
| 1456 |
-
"loss": 0.
|
| 1457 |
-
"num_tokens":
|
| 1458 |
"step": 905
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
-
"epoch":
|
| 1462 |
-
"grad_norm": 0.
|
| 1463 |
-
"learning_rate":
|
| 1464 |
-
"loss": 0.
|
| 1465 |
-
"num_tokens":
|
| 1466 |
"step": 910
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
-
"epoch":
|
| 1470 |
-
"grad_norm": 0.
|
| 1471 |
-
"learning_rate":
|
| 1472 |
"loss": 0.0506,
|
| 1473 |
-
"num_tokens":
|
| 1474 |
"step": 915
|
| 1475 |
},
|
| 1476 |
{
|
| 1477 |
-
"epoch":
|
| 1478 |
-
"grad_norm": 0.
|
| 1479 |
-
"learning_rate":
|
| 1480 |
-
"loss": 0.
|
| 1481 |
-
"num_tokens":
|
| 1482 |
"step": 920
|
| 1483 |
},
|
| 1484 |
{
|
| 1485 |
-
"epoch":
|
| 1486 |
-
"grad_norm": 0.
|
| 1487 |
-
"learning_rate":
|
| 1488 |
-
"loss": 0.
|
| 1489 |
-
"num_tokens":
|
| 1490 |
"step": 925
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
-
"epoch":
|
| 1494 |
-
"grad_norm": 0.
|
| 1495 |
-
"learning_rate":
|
| 1496 |
-
"loss": 0.
|
| 1497 |
-
"num_tokens":
|
| 1498 |
"step": 930
|
| 1499 |
},
|
| 1500 |
{
|
| 1501 |
-
"epoch":
|
| 1502 |
-
"grad_norm": 0.
|
| 1503 |
-
"learning_rate":
|
| 1504 |
"loss": 0.0523,
|
| 1505 |
-
"num_tokens":
|
| 1506 |
"step": 935
|
| 1507 |
},
|
| 1508 |
{
|
| 1509 |
-
"epoch":
|
| 1510 |
-
"grad_norm": 0.
|
| 1511 |
-
"learning_rate":
|
| 1512 |
-
"loss": 0.
|
| 1513 |
-
"num_tokens":
|
| 1514 |
"step": 940
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
-
"epoch":
|
| 1518 |
-
"grad_norm": 0.
|
| 1519 |
-
"learning_rate":
|
| 1520 |
-
"loss": 0.
|
| 1521 |
-
"num_tokens":
|
| 1522 |
"step": 945
|
| 1523 |
},
|
| 1524 |
{
|
| 1525 |
-
"epoch":
|
| 1526 |
-
"grad_norm": 0.
|
| 1527 |
-
"learning_rate":
|
| 1528 |
-
"loss": 0.
|
| 1529 |
-
"num_tokens":
|
| 1530 |
"step": 950
|
| 1531 |
},
|
| 1532 |
{
|
| 1533 |
-
"epoch": 2.
|
| 1534 |
-
"grad_norm": 0.
|
| 1535 |
-
"learning_rate":
|
| 1536 |
-
"loss": 0.
|
| 1537 |
-
"num_tokens":
|
| 1538 |
"step": 955
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
-
"epoch": 2.
|
| 1542 |
-
"grad_norm": 0.
|
| 1543 |
-
"learning_rate":
|
| 1544 |
-
"loss": 0.
|
| 1545 |
-
"num_tokens":
|
| 1546 |
"step": 960
|
| 1547 |
},
|
| 1548 |
{
|
| 1549 |
-
"epoch": 2.
|
| 1550 |
-
"grad_norm": 0.
|
| 1551 |
-
"learning_rate":
|
| 1552 |
-
"loss": 0.
|
| 1553 |
-
"num_tokens":
|
| 1554 |
"step": 965
|
| 1555 |
},
|
| 1556 |
{
|
| 1557 |
-
"epoch": 2.
|
| 1558 |
-
"grad_norm": 0.
|
| 1559 |
-
"learning_rate":
|
| 1560 |
-
"loss": 0.
|
| 1561 |
-
"num_tokens":
|
| 1562 |
"step": 970
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
-
"epoch": 2.
|
| 1566 |
-
"grad_norm": 0.
|
| 1567 |
-
"learning_rate":
|
| 1568 |
-
"loss": 0.
|
| 1569 |
-
"num_tokens":
|
| 1570 |
"step": 975
|
| 1571 |
},
|
| 1572 |
{
|
| 1573 |
-
"epoch": 2.
|
| 1574 |
-
"grad_norm": 0.
|
| 1575 |
-
"learning_rate":
|
| 1576 |
-
"loss": 0.
|
| 1577 |
-
"num_tokens":
|
| 1578 |
"step": 980
|
| 1579 |
},
|
| 1580 |
{
|
| 1581 |
-
"epoch": 2.
|
| 1582 |
-
"grad_norm": 0.
|
| 1583 |
-
"learning_rate":
|
| 1584 |
-
"loss": 0.
|
| 1585 |
-
"num_tokens":
|
| 1586 |
"step": 985
|
| 1587 |
},
|
| 1588 |
{
|
| 1589 |
-
"epoch": 2.
|
| 1590 |
-
"grad_norm": 0.
|
| 1591 |
-
"learning_rate":
|
| 1592 |
-
"loss": 0.
|
| 1593 |
-
"num_tokens":
|
| 1594 |
"step": 990
|
| 1595 |
},
|
| 1596 |
{
|
| 1597 |
-
"epoch": 2.
|
| 1598 |
-
"grad_norm": 0.
|
| 1599 |
-
"learning_rate":
|
| 1600 |
-
"loss": 0.
|
| 1601 |
-
"num_tokens":
|
| 1602 |
"step": 995
|
| 1603 |
},
|
| 1604 |
{
|
| 1605 |
-
"epoch": 2.
|
| 1606 |
-
"grad_norm": 0.
|
| 1607 |
-
"learning_rate":
|
| 1608 |
-
"loss": 0.
|
| 1609 |
-
"num_tokens":
|
| 1610 |
"step": 1000
|
| 1611 |
},
|
| 1612 |
{
|
| 1613 |
-
"epoch": 2.
|
| 1614 |
-
"grad_norm": 0.
|
| 1615 |
-
"learning_rate":
|
| 1616 |
-
"loss": 0.
|
| 1617 |
-
"num_tokens":
|
| 1618 |
"step": 1005
|
| 1619 |
},
|
| 1620 |
{
|
| 1621 |
-
"epoch": 2.
|
| 1622 |
-
"grad_norm": 0.
|
| 1623 |
-
"learning_rate":
|
| 1624 |
-
"loss": 0.
|
| 1625 |
-
"num_tokens":
|
| 1626 |
"step": 1010
|
| 1627 |
},
|
| 1628 |
{
|
| 1629 |
-
"epoch": 2.
|
| 1630 |
-
"grad_norm": 0.
|
| 1631 |
-
"learning_rate":
|
| 1632 |
-
"loss": 0.
|
| 1633 |
-
"num_tokens":
|
| 1634 |
"step": 1015
|
| 1635 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1636 |
{
|
| 1637 |
"epoch": 3.0,
|
| 1638 |
-
"num_tokens":
|
| 1639 |
-
"step":
|
| 1640 |
-
"total_flos":
|
| 1641 |
-
"train_loss": 0.
|
| 1642 |
-
"train_runtime":
|
| 1643 |
-
"train_samples_per_second":
|
| 1644 |
-
"train_steps_per_second": 0.
|
| 1645 |
}
|
| 1646 |
],
|
| 1647 |
"logging_steps": 5,
|
| 1648 |
-
"max_steps":
|
| 1649 |
"num_input_tokens_seen": 0,
|
| 1650 |
"num_train_epochs": 3,
|
| 1651 |
"save_steps": 100,
|
|
@@ -1661,7 +2317,7 @@
|
|
| 1661 |
"attributes": {}
|
| 1662 |
}
|
| 1663 |
},
|
| 1664 |
-
"total_flos":
|
| 1665 |
"train_batch_size": 16,
|
| 1666 |
"trial_name": null,
|
| 1667 |
"trial_params": null
|
|
|
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
"epoch": 3.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 1428,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
+
"epoch": 0.01050420168067227,
|
| 14 |
+
"grad_norm": 21.68925666809082,
|
| 15 |
+
"learning_rate": 3.4722222222222224e-06,
|
| 16 |
+
"loss": 1.7424,
|
| 17 |
+
"num_tokens": 61440.0,
|
| 18 |
"step": 5
|
| 19 |
},
|
| 20 |
{
|
| 21 |
+
"epoch": 0.02100840336134454,
|
| 22 |
+
"grad_norm": 6.517210483551025,
|
| 23 |
+
"learning_rate": 6.944444444444445e-06,
|
| 24 |
+
"loss": 0.8531,
|
| 25 |
+
"num_tokens": 122880.0,
|
| 26 |
"step": 10
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.031512605042016806,
|
| 30 |
+
"grad_norm": 1.3076003789901733,
|
| 31 |
+
"learning_rate": 1.0416666666666668e-05,
|
| 32 |
+
"loss": 0.2121,
|
| 33 |
+
"num_tokens": 184320.0,
|
| 34 |
"step": 15
|
| 35 |
},
|
| 36 |
{
|
| 37 |
+
"epoch": 0.04201680672268908,
|
| 38 |
+
"grad_norm": 1.2455157041549683,
|
| 39 |
+
"learning_rate": 1.388888888888889e-05,
|
| 40 |
+
"loss": 0.113,
|
| 41 |
+
"num_tokens": 245760.0,
|
| 42 |
"step": 20
|
| 43 |
},
|
| 44 |
{
|
| 45 |
+
"epoch": 0.052521008403361345,
|
| 46 |
+
"grad_norm": 0.854950487613678,
|
| 47 |
+
"learning_rate": 1.736111111111111e-05,
|
| 48 |
+
"loss": 0.0885,
|
| 49 |
+
"num_tokens": 307200.0,
|
| 50 |
"step": 25
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.06302521008403361,
|
| 54 |
+
"grad_norm": 0.9929510951042175,
|
| 55 |
+
"learning_rate": 2.0833333333333336e-05,
|
| 56 |
+
"loss": 0.0771,
|
| 57 |
+
"num_tokens": 368640.0,
|
| 58 |
"step": 30
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.07352941176470588,
|
| 62 |
+
"grad_norm": 1.0237308740615845,
|
| 63 |
+
"learning_rate": 2.4305555555555558e-05,
|
| 64 |
+
"loss": 0.0805,
|
| 65 |
+
"num_tokens": 430080.0,
|
| 66 |
"step": 35
|
| 67 |
},
|
| 68 |
{
|
| 69 |
+
"epoch": 0.08403361344537816,
|
| 70 |
+
"grad_norm": 1.4065616130828857,
|
| 71 |
+
"learning_rate": 2.777777777777778e-05,
|
| 72 |
+
"loss": 0.0822,
|
| 73 |
+
"num_tokens": 491520.0,
|
| 74 |
"step": 40
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 0.09453781512605042,
|
| 78 |
+
"grad_norm": 0.7961967587471008,
|
| 79 |
+
"learning_rate": 3.125e-05,
|
| 80 |
+
"loss": 0.0721,
|
| 81 |
+
"num_tokens": 552960.0,
|
| 82 |
"step": 45
|
| 83 |
},
|
| 84 |
{
|
| 85 |
+
"epoch": 0.10504201680672269,
|
| 86 |
+
"grad_norm": 0.7718358039855957,
|
| 87 |
+
"learning_rate": 3.472222222222222e-05,
|
| 88 |
+
"loss": 0.0683,
|
| 89 |
+
"num_tokens": 614297.0,
|
| 90 |
"step": 50
|
| 91 |
},
|
| 92 |
{
|
| 93 |
+
"epoch": 0.11554621848739496,
|
| 94 |
+
"grad_norm": 0.6594786047935486,
|
| 95 |
+
"learning_rate": 3.8194444444444444e-05,
|
| 96 |
+
"loss": 0.0665,
|
| 97 |
+
"num_tokens": 675536.0,
|
| 98 |
"step": 55
|
| 99 |
},
|
| 100 |
{
|
| 101 |
+
"epoch": 0.12605042016806722,
|
| 102 |
+
"grad_norm": 0.637739896774292,
|
| 103 |
+
"learning_rate": 4.166666666666667e-05,
|
| 104 |
+
"loss": 0.0718,
|
| 105 |
+
"num_tokens": 736976.0,
|
| 106 |
"step": 60
|
| 107 |
},
|
| 108 |
{
|
| 109 |
+
"epoch": 0.13655462184873948,
|
| 110 |
+
"grad_norm": 0.4128863215446472,
|
| 111 |
+
"learning_rate": 4.5138888888888894e-05,
|
| 112 |
+
"loss": 0.069,
|
| 113 |
+
"num_tokens": 798416.0,
|
| 114 |
"step": 65
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.14705882352941177,
|
| 118 |
+
"grad_norm": 0.7012873888015747,
|
| 119 |
+
"learning_rate": 4.8611111111111115e-05,
|
| 120 |
+
"loss": 0.0651,
|
| 121 |
+
"num_tokens": 859856.0,
|
| 122 |
"step": 70
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 0.15756302521008403,
|
| 126 |
+
"grad_norm": 0.6367799639701843,
|
| 127 |
+
"learning_rate": 4.9999456532409905e-05,
|
| 128 |
+
"loss": 0.0777,
|
| 129 |
+
"num_tokens": 921296.0,
|
| 130 |
"step": 75
|
| 131 |
},
|
| 132 |
{
|
| 133 |
+
"epoch": 0.16806722689075632,
|
| 134 |
+
"grad_norm": 0.49970555305480957,
|
| 135 |
+
"learning_rate": 4.999613543665713e-05,
|
| 136 |
+
"loss": 0.0642,
|
| 137 |
+
"num_tokens": 982736.0,
|
| 138 |
"step": 80
|
| 139 |
},
|
| 140 |
{
|
| 141 |
+
"epoch": 0.17857142857142858,
|
| 142 |
+
"grad_norm": 0.49360784888267517,
|
| 143 |
+
"learning_rate": 4.998979561670338e-05,
|
| 144 |
+
"loss": 0.0651,
|
| 145 |
+
"num_tokens": 1044176.0,
|
| 146 |
"step": 85
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 0.18907563025210083,
|
| 150 |
+
"grad_norm": 0.432689905166626,
|
| 151 |
+
"learning_rate": 4.9980437923280036e-05,
|
| 152 |
+
"loss": 0.063,
|
| 153 |
+
"num_tokens": 1105611.0,
|
| 154 |
"step": 90
|
| 155 |
},
|
| 156 |
{
|
| 157 |
+
"epoch": 0.19957983193277312,
|
| 158 |
+
"grad_norm": 0.6416879296302795,
|
| 159 |
+
"learning_rate": 4.996806361208257e-05,
|
| 160 |
+
"loss": 0.0659,
|
| 161 |
+
"num_tokens": 1167051.0,
|
| 162 |
"step": 95
|
| 163 |
},
|
| 164 |
{
|
| 165 |
+
"epoch": 0.21008403361344538,
|
| 166 |
+
"grad_norm": 0.517444908618927,
|
| 167 |
+
"learning_rate": 4.995267434360207e-05,
|
| 168 |
"loss": 0.0638,
|
| 169 |
+
"num_tokens": 1228314.0,
|
| 170 |
"step": 100
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.22058823529411764,
|
| 174 |
+
"grad_norm": 107.99620819091797,
|
| 175 |
+
"learning_rate": 4.993427218290246e-05,
|
| 176 |
+
"loss": 0.6636,
|
| 177 |
+
"num_tokens": 1289602.0,
|
| 178 |
"step": 105
|
| 179 |
},
|
| 180 |
{
|
| 181 |
+
"epoch": 0.23109243697478993,
|
| 182 |
+
"grad_norm": 0.5925813317298889,
|
| 183 |
+
"learning_rate": 4.991285959934332e-05,
|
| 184 |
+
"loss": 0.1472,
|
| 185 |
+
"num_tokens": 1351042.0,
|
| 186 |
"step": 110
|
| 187 |
},
|
| 188 |
{
|
| 189 |
+
"epoch": 0.2415966386554622,
|
| 190 |
+
"grad_norm": 0.3515715003013611,
|
| 191 |
+
"learning_rate": 4.988843946624858e-05,
|
| 192 |
+
"loss": 0.0645,
|
| 193 |
+
"num_tokens": 1412482.0,
|
| 194 |
"step": 115
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 0.25210084033613445,
|
| 198 |
+
"grad_norm": 0.4023377001285553,
|
| 199 |
+
"learning_rate": 4.9861015060520935e-05,
|
| 200 |
+
"loss": 0.0632,
|
| 201 |
+
"num_tokens": 1473922.0,
|
| 202 |
"step": 120
|
| 203 |
},
|
| 204 |
{
|
| 205 |
+
"epoch": 0.26260504201680673,
|
| 206 |
+
"grad_norm": 0.37604016065597534,
|
| 207 |
+
"learning_rate": 4.9830590062202105e-05,
|
| 208 |
+
"loss": 0.0635,
|
| 209 |
+
"num_tokens": 1535362.0,
|
| 210 |
"step": 125
|
| 211 |
},
|
| 212 |
{
|
| 213 |
+
"epoch": 0.27310924369747897,
|
| 214 |
+
"grad_norm": 3.824808359146118,
|
| 215 |
+
"learning_rate": 4.9797168553979054e-05,
|
| 216 |
+
"loss": 0.0704,
|
| 217 |
+
"num_tokens": 1596802.0,
|
| 218 |
"step": 130
|
| 219 |
},
|
| 220 |
{
|
| 221 |
+
"epoch": 0.28361344537815125,
|
| 222 |
+
"grad_norm": 4.908388137817383,
|
| 223 |
+
"learning_rate": 4.976075502063613e-05,
|
| 224 |
+
"loss": 0.128,
|
| 225 |
+
"num_tokens": 1658080.0,
|
| 226 |
"step": 135
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.29411764705882354,
|
| 230 |
+
"grad_norm": 0.3574487566947937,
|
| 231 |
+
"learning_rate": 4.97213543484532e-05,
|
| 232 |
+
"loss": 0.0682,
|
| 233 |
+
"num_tokens": 1719520.0,
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.30462184873949577,
|
| 238 |
+
"grad_norm": 0.35017549991607666,
|
| 239 |
+
"learning_rate": 4.9678971824550074e-05,
|
| 240 |
+
"loss": 0.0593,
|
| 241 |
+
"num_tokens": 1780960.0,
|
| 242 |
"step": 145
|
| 243 |
},
|
| 244 |
{
|
| 245 |
+
"epoch": 0.31512605042016806,
|
| 246 |
+
"grad_norm": 0.36095139384269714,
|
| 247 |
+
"learning_rate": 4.9633613136176925e-05,
|
| 248 |
+
"loss": 0.0605,
|
| 249 |
+
"num_tokens": 1842400.0,
|
| 250 |
"step": 150
|
| 251 |
},
|
| 252 |
{
|
| 253 |
+
"epoch": 0.32563025210084034,
|
| 254 |
+
"grad_norm": 0.3857307434082031,
|
| 255 |
+
"learning_rate": 4.95852843699512e-05,
|
| 256 |
+
"loss": 0.059,
|
| 257 |
+
"num_tokens": 1903840.0,
|
| 258 |
"step": 155
|
| 259 |
},
|
| 260 |
{
|
| 261 |
+
"epoch": 0.33613445378151263,
|
| 262 |
+
"grad_norm": 0.27103978395462036,
|
| 263 |
+
"learning_rate": 4.953399201104084e-05,
|
| 264 |
+
"loss": 0.0581,
|
| 265 |
+
"num_tokens": 1965280.0,
|
| 266 |
"step": 160
|
| 267 |
},
|
| 268 |
{
|
| 269 |
+
"epoch": 0.34663865546218486,
|
| 270 |
+
"grad_norm": 0.29975804686546326,
|
| 271 |
+
"learning_rate": 4.9479742942294035e-05,
|
| 272 |
+
"loss": 0.0578,
|
| 273 |
+
"num_tokens": 2026720.0,
|
| 274 |
"step": 165
|
| 275 |
},
|
| 276 |
{
|
| 277 |
+
"epoch": 0.35714285714285715,
|
| 278 |
+
"grad_norm": 0.3278215825557709,
|
| 279 |
+
"learning_rate": 4.9422544443315635e-05,
|
| 280 |
+
"loss": 0.056,
|
| 281 |
+
"num_tokens": 2088160.0,
|
| 282 |
"step": 170
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 0.36764705882352944,
|
| 286 |
+
"grad_norm": 0.28858262300491333,
|
| 287 |
+
"learning_rate": 4.936240418949032e-05,
|
| 288 |
+
"loss": 0.0582,
|
| 289 |
+
"num_tokens": 2149600.0,
|
| 290 |
"step": 175
|
| 291 |
},
|
| 292 |
{
|
| 293 |
+
"epoch": 0.37815126050420167,
|
| 294 |
+
"grad_norm": 0.36514896154403687,
|
| 295 |
+
"learning_rate": 4.929933025095261e-05,
|
| 296 |
+
"loss": 0.0577,
|
| 297 |
+
"num_tokens": 2211007.0,
|
| 298 |
"step": 180
|
| 299 |
},
|
| 300 |
{
|
| 301 |
+
"epoch": 0.38865546218487396,
|
| 302 |
+
"grad_norm": 0.37031087279319763,
|
| 303 |
+
"learning_rate": 4.9233331091504034e-05,
|
| 304 |
+
"loss": 0.0554,
|
| 305 |
+
"num_tokens": 2272203.0,
|
| 306 |
"step": 185
|
| 307 |
},
|
| 308 |
{
|
| 309 |
+
"epoch": 0.39915966386554624,
|
| 310 |
+
"grad_norm": 0.28869447112083435,
|
| 311 |
+
"learning_rate": 4.916441556747727e-05,
|
| 312 |
+
"loss": 0.0575,
|
| 313 |
+
"num_tokens": 2333494.0,
|
| 314 |
"step": 190
|
| 315 |
},
|
| 316 |
{
|
| 317 |
+
"epoch": 0.4096638655462185,
|
| 318 |
+
"grad_norm": 0.29715102910995483,
|
| 319 |
+
"learning_rate": 4.909259292654782e-05,
|
| 320 |
+
"loss": 0.0573,
|
| 321 |
+
"num_tokens": 2394934.0,
|
| 322 |
"step": 195
|
| 323 |
},
|
| 324 |
{
|
| 325 |
+
"epoch": 0.42016806722689076,
|
| 326 |
+
"grad_norm": 0.40769094228744507,
|
| 327 |
+
"learning_rate": 4.9017872806492995e-05,
|
| 328 |
+
"loss": 0.0583,
|
| 329 |
+
"num_tokens": 2456374.0,
|
| 330 |
"step": 200
|
| 331 |
},
|
| 332 |
{
|
| 333 |
+
"epoch": 0.43067226890756305,
|
| 334 |
+
"grad_norm": 0.3187924325466156,
|
| 335 |
+
"learning_rate": 4.8940265233898744e-05,
|
| 336 |
+
"loss": 0.0546,
|
| 337 |
+
"num_tokens": 2517814.0,
|
| 338 |
"step": 205
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 0.4411764705882353,
|
| 342 |
+
"grad_norm": 0.25656041502952576,
|
| 343 |
+
"learning_rate": 4.885978062281408e-05,
|
| 344 |
+
"loss": 0.057,
|
| 345 |
+
"num_tokens": 2579248.0,
|
| 346 |
"step": 210
|
| 347 |
},
|
| 348 |
{
|
| 349 |
+
"epoch": 0.45168067226890757,
|
| 350 |
+
"grad_norm": 0.2591699957847595,
|
| 351 |
+
"learning_rate": 4.877642977335371e-05,
|
| 352 |
+
"loss": 0.0573,
|
| 353 |
+
"num_tokens": 2640688.0,
|
| 354 |
"step": 215
|
| 355 |
},
|
| 356 |
{
|
| 357 |
+
"epoch": 0.46218487394957986,
|
| 358 |
+
"grad_norm": 0.3359907567501068,
|
| 359 |
+
"learning_rate": 4.869022387024879e-05,
|
| 360 |
+
"loss": 0.0538,
|
| 361 |
+
"num_tokens": 2702128.0,
|
| 362 |
"step": 220
|
| 363 |
},
|
| 364 |
{
|
| 365 |
+
"epoch": 0.4726890756302521,
|
| 366 |
+
"grad_norm": 0.3090253174304962,
|
| 367 |
+
"learning_rate": 4.8601174481346015e-05,
|
| 368 |
+
"loss": 0.057,
|
| 369 |
+
"num_tokens": 2763412.0,
|
| 370 |
"step": 225
|
| 371 |
},
|
| 372 |
{
|
| 373 |
+
"epoch": 0.4831932773109244,
|
| 374 |
+
"grad_norm": 0.2006455659866333,
|
| 375 |
+
"learning_rate": 4.8509293556055345e-05,
|
| 376 |
+
"loss": 0.0554,
|
| 377 |
+
"num_tokens": 2824852.0,
|
| 378 |
"step": 230
|
| 379 |
},
|
| 380 |
{
|
| 381 |
+
"epoch": 0.49369747899159666,
|
| 382 |
+
"grad_norm": 0.255356103181839,
|
| 383 |
+
"learning_rate": 4.84145934237466e-05,
|
| 384 |
+
"loss": 0.0557,
|
| 385 |
+
"num_tokens": 2886292.0,
|
| 386 |
"step": 235
|
| 387 |
},
|
| 388 |
{
|
| 389 |
+
"epoch": 0.5042016806722689,
|
| 390 |
+
"grad_norm": 0.21021538972854614,
|
| 391 |
+
"learning_rate": 4.8317086792094906e-05,
|
| 392 |
+
"loss": 0.0527,
|
| 393 |
+
"num_tokens": 2947732.0,
|
| 394 |
"step": 240
|
| 395 |
},
|
| 396 |
{
|
| 397 |
+
"epoch": 0.5147058823529411,
|
| 398 |
+
"grad_norm": 0.24062202870845795,
|
| 399 |
+
"learning_rate": 4.821678674537557e-05,
|
| 400 |
+
"loss": 0.0545,
|
| 401 |
+
"num_tokens": 3009172.0,
|
| 402 |
"step": 245
|
| 403 |
},
|
| 404 |
{
|
| 405 |
+
"epoch": 0.5252100840336135,
|
| 406 |
+
"grad_norm": 0.30908721685409546,
|
| 407 |
+
"learning_rate": 4.811370674270821e-05,
|
| 408 |
+
"loss": 0.0538,
|
| 409 |
+
"num_tokens": 3070612.0,
|
| 410 |
"step": 250
|
| 411 |
},
|
| 412 |
{
|
| 413 |
+
"epoch": 0.5357142857142857,
|
| 414 |
+
"grad_norm": 0.2805303931236267,
|
| 415 |
+
"learning_rate": 4.800786061625078e-05,
|
| 416 |
+
"loss": 0.0528,
|
| 417 |
+
"num_tokens": 3132052.0,
|
| 418 |
"step": 255
|
| 419 |
},
|
| 420 |
{
|
| 421 |
+
"epoch": 0.5462184873949579,
|
| 422 |
+
"grad_norm": 0.245064839720726,
|
| 423 |
+
"learning_rate": 4.789926256934345e-05,
|
| 424 |
+
"loss": 0.0566,
|
| 425 |
+
"num_tokens": 3193416.0,
|
| 426 |
"step": 260
|
| 427 |
},
|
| 428 |
{
|
| 429 |
+
"epoch": 0.5567226890756303,
|
| 430 |
+
"grad_norm": 0.2865758538246155,
|
| 431 |
+
"learning_rate": 4.778792717460259e-05,
|
| 432 |
+
"loss": 0.0542,
|
| 433 |
+
"num_tokens": 3254856.0,
|
| 434 |
"step": 265
|
| 435 |
},
|
| 436 |
{
|
| 437 |
+
"epoch": 0.5672268907563025,
|
| 438 |
+
"grad_norm": 0.2498483657836914,
|
| 439 |
+
"learning_rate": 4.7673869371965425e-05,
|
| 440 |
+
"loss": 0.0544,
|
| 441 |
+
"num_tokens": 3316296.0,
|
| 442 |
"step": 270
|
| 443 |
},
|
| 444 |
{
|
| 445 |
+
"epoch": 0.5777310924369747,
|
| 446 |
+
"grad_norm": 0.24384859204292297,
|
| 447 |
+
"learning_rate": 4.755710446668515e-05,
|
| 448 |
+
"loss": 0.0564,
|
| 449 |
+
"num_tokens": 3377345.0,
|
| 450 |
"step": 275
|
| 451 |
},
|
| 452 |
{
|
| 453 |
+
"epoch": 0.5882352941176471,
|
| 454 |
+
"grad_norm": 0.22256287932395935,
|
| 455 |
+
"learning_rate": 4.7437648127277216e-05,
|
| 456 |
+
"loss": 0.0543,
|
| 457 |
+
"num_tokens": 3438785.0,
|
| 458 |
"step": 280
|
| 459 |
},
|
| 460 |
{
|
| 461 |
+
"epoch": 0.5987394957983193,
|
| 462 |
+
"grad_norm": 0.7542266845703125,
|
| 463 |
+
"learning_rate": 4.7315516383416736e-05,
|
| 464 |
+
"loss": 0.0595,
|
| 465 |
+
"num_tokens": 3500225.0,
|
| 466 |
"step": 285
|
| 467 |
},
|
| 468 |
{
|
| 469 |
+
"epoch": 0.6092436974789915,
|
| 470 |
+
"grad_norm": 0.24494485557079315,
|
| 471 |
+
"learning_rate": 4.7190725623787545e-05,
|
| 472 |
+
"loss": 0.0565,
|
| 473 |
+
"num_tokens": 3561466.0,
|
| 474 |
"step": 290
|
| 475 |
},
|
| 476 |
{
|
| 477 |
+
"epoch": 0.6197478991596639,
|
| 478 |
+
"grad_norm": 0.26762306690216064,
|
| 479 |
+
"learning_rate": 4.706329259388298e-05,
|
| 480 |
+
"loss": 0.0557,
|
| 481 |
+
"num_tokens": 3622906.0,
|
| 482 |
"step": 295
|
| 483 |
},
|
| 484 |
{
|
| 485 |
+
"epoch": 0.6302521008403361,
|
| 486 |
+
"grad_norm": 0.285203754901886,
|
| 487 |
+
"learning_rate": 4.6933234393758844e-05,
|
| 488 |
+
"loss": 0.0537,
|
| 489 |
+
"num_tokens": 3684184.0,
|
| 490 |
"step": 300
|
| 491 |
},
|
| 492 |
{
|
| 493 |
+
"epoch": 0.6407563025210085,
|
| 494 |
+
"grad_norm": 0.5487973690032959,
|
| 495 |
+
"learning_rate": 4.680056847573878e-05,
|
| 496 |
+
"loss": 0.0551,
|
| 497 |
+
"num_tokens": 3745624.0,
|
| 498 |
"step": 305
|
| 499 |
},
|
| 500 |
{
|
| 501 |
+
"epoch": 0.6512605042016807,
|
| 502 |
+
"grad_norm": 0.2598506808280945,
|
| 503 |
+
"learning_rate": 4.666531264207235e-05,
|
| 504 |
+
"loss": 0.0542,
|
| 505 |
+
"num_tokens": 3806907.0,
|
| 506 |
"step": 310
|
| 507 |
},
|
| 508 |
{
|
| 509 |
+
"epoch": 0.6617647058823529,
|
| 510 |
+
"grad_norm": 0.30089858174324036,
|
| 511 |
+
"learning_rate": 4.6527485042546204e-05,
|
| 512 |
+
"loss": 0.0576,
|
| 513 |
+
"num_tokens": 3868347.0,
|
| 514 |
"step": 315
|
| 515 |
},
|
| 516 |
{
|
| 517 |
+
"epoch": 0.6722689075630253,
|
| 518 |
+
"grad_norm": 0.24752771854400635,
|
| 519 |
+
"learning_rate": 4.638710417204855e-05,
|
| 520 |
+
"loss": 0.0538,
|
| 521 |
+
"num_tokens": 3929787.0,
|
| 522 |
"step": 320
|
| 523 |
},
|
| 524 |
{
|
| 525 |
+
"epoch": 0.6827731092436975,
|
| 526 |
+
"grad_norm": 0.2561896741390228,
|
| 527 |
+
"learning_rate": 4.6244188868087395e-05,
|
| 528 |
+
"loss": 0.0556,
|
| 529 |
+
"num_tokens": 3991227.0,
|
| 530 |
"step": 325
|
| 531 |
},
|
| 532 |
{
|
| 533 |
+
"epoch": 0.6932773109243697,
|
| 534 |
+
"grad_norm": 0.3156558871269226,
|
| 535 |
+
"learning_rate": 4.609875830826272e-05,
|
| 536 |
+
"loss": 0.0564,
|
| 537 |
+
"num_tokens": 4052667.0,
|
| 538 |
"step": 330
|
| 539 |
},
|
| 540 |
{
|
| 541 |
+
"epoch": 0.7037815126050421,
|
| 542 |
+
"grad_norm": 0.291674941778183,
|
| 543 |
+
"learning_rate": 4.59508320076931e-05,
|
| 544 |
+
"loss": 0.0558,
|
| 545 |
+
"num_tokens": 4114107.0,
|
| 546 |
"step": 335
|
| 547 |
},
|
| 548 |
{
|
| 549 |
+
"epoch": 0.7142857142857143,
|
| 550 |
+
"grad_norm": 0.29663559794425964,
|
| 551 |
+
"learning_rate": 4.580042981639698e-05,
|
| 552 |
+
"loss": 0.0545,
|
| 553 |
+
"num_tokens": 4175485.0,
|
| 554 |
"step": 340
|
| 555 |
},
|
| 556 |
{
|
| 557 |
+
"epoch": 0.7247899159663865,
|
| 558 |
+
"grad_norm": 0.2254744917154312,
|
| 559 |
+
"learning_rate": 4.5647571916629064e-05,
|
| 560 |
+
"loss": 0.0544,
|
| 561 |
+
"num_tokens": 4236925.0,
|
| 562 |
"step": 345
|
| 563 |
},
|
| 564 |
{
|
| 565 |
+
"epoch": 0.7352941176470589,
|
| 566 |
+
"grad_norm": 0.327118843793869,
|
| 567 |
+
"learning_rate": 4.549227882017202e-05,
|
| 568 |
+
"loss": 0.0556,
|
| 569 |
+
"num_tokens": 4298365.0,
|
| 570 |
"step": 350
|
| 571 |
},
|
| 572 |
{
|
| 573 |
+
"epoch": 0.7457983193277311,
|
| 574 |
+
"grad_norm": 0.24738788604736328,
|
| 575 |
+
"learning_rate": 4.533457136558408e-05,
|
| 576 |
+
"loss": 0.0533,
|
| 577 |
+
"num_tokens": 4359805.0,
|
| 578 |
"step": 355
|
| 579 |
},
|
| 580 |
{
|
| 581 |
+
"epoch": 0.7563025210084033,
|
| 582 |
+
"grad_norm": 0.17776817083358765,
|
| 583 |
+
"learning_rate": 4.5174470715402764e-05,
|
| 584 |
+
"loss": 0.0559,
|
| 585 |
+
"num_tokens": 4421245.0,
|
| 586 |
"step": 360
|
| 587 |
},
|
| 588 |
{
|
| 589 |
+
"epoch": 0.7668067226890757,
|
| 590 |
+
"grad_norm": 0.2214263677597046,
|
| 591 |
+
"learning_rate": 4.501199835330507e-05,
|
| 592 |
+
"loss": 0.0533,
|
| 593 |
+
"num_tokens": 4482685.0,
|
| 594 |
"step": 365
|
| 595 |
},
|
| 596 |
{
|
| 597 |
+
"epoch": 0.7773109243697479,
|
| 598 |
+
"grad_norm": 0.2129214107990265,
|
| 599 |
+
"learning_rate": 4.484717608122459e-05,
|
| 600 |
+
"loss": 0.0542,
|
| 601 |
+
"num_tokens": 4544125.0,
|
| 602 |
"step": 370
|
| 603 |
},
|
| 604 |
{
|
| 605 |
+
"epoch": 0.7878151260504201,
|
| 606 |
+
"grad_norm": 0.2668643295764923,
|
| 607 |
+
"learning_rate": 4.468002601642603e-05,
|
| 608 |
+
"loss": 0.052,
|
| 609 |
+
"num_tokens": 4605565.0,
|
| 610 |
"step": 375
|
| 611 |
},
|
| 612 |
{
|
| 613 |
+
"epoch": 0.7983193277310925,
|
| 614 |
+
"grad_norm": 0.4849870502948761,
|
| 615 |
+
"learning_rate": 4.4510570588537206e-05,
|
| 616 |
+
"loss": 0.057,
|
| 617 |
+
"num_tokens": 4666849.0,
|
| 618 |
"step": 380
|
| 619 |
},
|
| 620 |
{
|
| 621 |
+
"epoch": 0.8088235294117647,
|
| 622 |
+
"grad_norm": 0.2671234607696533,
|
| 623 |
+
"learning_rate": 4.433883253653936e-05,
|
| 624 |
+
"loss": 0.0533,
|
| 625 |
+
"num_tokens": 4728021.0,
|
| 626 |
"step": 385
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 0.819327731092437,
|
| 630 |
+
"grad_norm": 0.8898406028747559,
|
| 631 |
+
"learning_rate": 4.416483490571574e-05,
|
| 632 |
+
"loss": 0.0551,
|
| 633 |
+
"num_tokens": 4789461.0,
|
| 634 |
"step": 390
|
| 635 |
},
|
| 636 |
{
|
| 637 |
+
"epoch": 0.8298319327731093,
|
| 638 |
+
"grad_norm": 0.1947408765554428,
|
| 639 |
+
"learning_rate": 4.39886010445593e-05,
|
| 640 |
+
"loss": 0.0555,
|
| 641 |
+
"num_tokens": 4850862.0,
|
| 642 |
"step": 395
|
| 643 |
},
|
| 644 |
{
|
| 645 |
+
"epoch": 0.8403361344537815,
|
| 646 |
+
"grad_norm": 0.3551996648311615,
|
| 647 |
+
"learning_rate": 4.381015460163949e-05,
|
| 648 |
+
"loss": 0.0559,
|
| 649 |
+
"num_tokens": 4912302.0,
|
| 650 |
"step": 400
|
| 651 |
},
|
| 652 |
{
|
| 653 |
+
"epoch": 0.8508403361344538,
|
| 654 |
+
"grad_norm": 0.24865303933620453,
|
| 655 |
+
"learning_rate": 4.362951952242898e-05,
|
| 656 |
+
"loss": 0.0554,
|
| 657 |
+
"num_tokens": 4973742.0,
|
| 658 |
"step": 405
|
| 659 |
},
|
| 660 |
{
|
| 661 |
+
"epoch": 0.8613445378151261,
|
| 662 |
+
"grad_norm": 0.21480585634708405,
|
| 663 |
+
"learning_rate": 4.344672004609037e-05,
|
| 664 |
+
"loss": 0.0536,
|
| 665 |
+
"num_tokens": 5035005.0,
|
| 666 |
"step": 410
|
| 667 |
},
|
| 668 |
{
|
| 669 |
+
"epoch": 0.8718487394957983,
|
| 670 |
+
"grad_norm": 0.2362818419933319,
|
| 671 |
+
"learning_rate": 4.326178070222364e-05,
|
| 672 |
+
"loss": 0.0552,
|
| 673 |
+
"num_tokens": 5096371.0,
|
| 674 |
"step": 415
|
| 675 |
},
|
| 676 |
{
|
| 677 |
+
"epoch": 0.8823529411764706,
|
| 678 |
+
"grad_norm": 0.3188863694667816,
|
| 679 |
+
"learning_rate": 4.3074726307574516e-05,
|
| 680 |
+
"loss": 0.0579,
|
| 681 |
+
"num_tokens": 5157811.0,
|
| 682 |
"step": 420
|
| 683 |
},
|
| 684 |
{
|
| 685 |
+
"epoch": 0.8928571428571429,
|
| 686 |
+
"grad_norm": 0.2944329082965851,
|
| 687 |
+
"learning_rate": 4.2885581962704366e-05,
|
| 688 |
+
"loss": 0.0555,
|
| 689 |
+
"num_tokens": 5219251.0,
|
| 690 |
"step": 425
|
| 691 |
},
|
| 692 |
{
|
| 693 |
+
"epoch": 0.9033613445378151,
|
| 694 |
+
"grad_norm": 0.35381075739860535,
|
| 695 |
+
"learning_rate": 4.2694373048622e-05,
|
| 696 |
+
"loss": 0.0548,
|
| 697 |
+
"num_tokens": 5280691.0,
|
| 698 |
"step": 430
|
| 699 |
},
|
| 700 |
{
|
| 701 |
+
"epoch": 0.9138655462184874,
|
| 702 |
+
"grad_norm": 0.2967956066131592,
|
| 703 |
+
"learning_rate": 4.2501125223377754e-05,
|
| 704 |
+
"loss": 0.0542,
|
| 705 |
+
"num_tokens": 5342131.0,
|
| 706 |
"step": 435
|
| 707 |
},
|
| 708 |
{
|
| 709 |
+
"epoch": 0.9243697478991597,
|
| 710 |
+
"grad_norm": 1.3979747295379639,
|
| 711 |
+
"learning_rate": 4.230586441862062e-05,
|
| 712 |
+
"loss": 0.0529,
|
| 713 |
+
"num_tokens": 5403410.0,
|
| 714 |
"step": 440
|
| 715 |
},
|
| 716 |
{
|
| 717 |
+
"epoch": 0.9348739495798319,
|
| 718 |
+
"grad_norm": 0.2130478024482727,
|
| 719 |
+
"learning_rate": 4.210861683611837e-05,
|
| 720 |
+
"loss": 0.0546,
|
| 721 |
+
"num_tokens": 5464723.0,
|
| 722 |
"step": 445
|
| 723 |
},
|
| 724 |
{
|
| 725 |
+
"epoch": 0.9453781512605042,
|
| 726 |
+
"grad_norm": 0.23148652911186218,
|
| 727 |
+
"learning_rate": 4.1909408944241644e-05,
|
| 728 |
+
"loss": 0.0543,
|
| 729 |
+
"num_tokens": 5526163.0,
|
| 730 |
"step": 450
|
| 731 |
},
|
| 732 |
{
|
| 733 |
+
"epoch": 0.9558823529411765,
|
| 734 |
+
"grad_norm": 0.17318475246429443,
|
| 735 |
+
"learning_rate": 4.1708267474412215e-05,
|
| 736 |
+
"loss": 0.0543,
|
| 737 |
+
"num_tokens": 5587603.0,
|
| 738 |
"step": 455
|
| 739 |
},
|
| 740 |
{
|
| 741 |
+
"epoch": 0.9663865546218487,
|
| 742 |
+
"grad_norm": 0.28641510009765625,
|
| 743 |
+
"learning_rate": 4.1505219417515884e-05,
|
| 744 |
+
"loss": 0.0549,
|
| 745 |
+
"num_tokens": 5649043.0,
|
| 746 |
"step": 460
|
| 747 |
},
|
| 748 |
{
|
| 749 |
+
"epoch": 0.976890756302521,
|
| 750 |
+
"grad_norm": 0.35609665513038635,
|
| 751 |
+
"learning_rate": 4.1300292020280645e-05,
|
| 752 |
+
"loss": 0.056,
|
| 753 |
+
"num_tokens": 5710483.0,
|
| 754 |
"step": 465
|
| 755 |
},
|
| 756 |
{
|
| 757 |
+
"epoch": 0.9873949579831933,
|
| 758 |
+
"grad_norm": 0.2831043601036072,
|
| 759 |
+
"learning_rate": 4.10935127816205e-05,
|
| 760 |
+
"loss": 0.0574,
|
| 761 |
+
"num_tokens": 5771914.0,
|
| 762 |
"step": 470
|
| 763 |
},
|
| 764 |
{
|
| 765 |
+
"epoch": 0.9978991596638656,
|
| 766 |
+
"grad_norm": 0.2836240530014038,
|
| 767 |
+
"learning_rate": 4.088490944894539e-05,
|
| 768 |
+
"loss": 0.0515,
|
| 769 |
+
"num_tokens": 5833354.0,
|
| 770 |
"step": 475
|
| 771 |
},
|
| 772 |
{
|
| 773 |
+
"epoch": 1.0084033613445378,
|
| 774 |
+
"grad_norm": 2.6926634311676025,
|
| 775 |
+
"learning_rate": 4.06745100144378e-05,
|
| 776 |
+
"loss": 0.0558,
|
| 777 |
+
"num_tokens": 5894794.0,
|
| 778 |
"step": 480
|
| 779 |
},
|
| 780 |
{
|
| 781 |
+
"epoch": 1.01890756302521,
|
| 782 |
+
"grad_norm": 0.17681336402893066,
|
| 783 |
+
"learning_rate": 4.0462342711296584e-05,
|
| 784 |
+
"loss": 0.0523,
|
| 785 |
+
"num_tokens": 5956077.0,
|
| 786 |
"step": 485
|
| 787 |
},
|
| 788 |
{
|
| 789 |
+
"epoch": 1.0294117647058822,
|
| 790 |
+
"grad_norm": 0.2838696539402008,
|
| 791 |
+
"learning_rate": 4.024843600994833e-05,
|
| 792 |
+
"loss": 0.0537,
|
| 793 |
+
"num_tokens": 6017517.0,
|
| 794 |
"step": 490
|
| 795 |
},
|
| 796 |
{
|
| 797 |
+
"epoch": 1.0399159663865547,
|
| 798 |
+
"grad_norm": 0.2431504875421524,
|
| 799 |
+
"learning_rate": 4.003281861422699e-05,
|
| 800 |
+
"loss": 0.0537,
|
| 801 |
+
"num_tokens": 6078801.0,
|
| 802 |
"step": 495
|
| 803 |
},
|
| 804 |
{
|
| 805 |
+
"epoch": 1.050420168067227,
|
| 806 |
+
"grad_norm": 0.2204369157552719,
|
| 807 |
+
"learning_rate": 3.981551945752215e-05,
|
| 808 |
+
"loss": 0.0538,
|
| 809 |
+
"num_tokens": 6140232.0,
|
| 810 |
"step": 500
|
| 811 |
},
|
| 812 |
{
|
| 813 |
+
"epoch": 1.0609243697478992,
|
| 814 |
+
"grad_norm": 0.2458706945180893,
|
| 815 |
+
"learning_rate": 3.959656769889646e-05,
|
| 816 |
+
"loss": 0.0545,
|
| 817 |
+
"num_tokens": 6201672.0,
|
| 818 |
"step": 505
|
| 819 |
},
|
| 820 |
{
|
| 821 |
+
"epoch": 1.0714285714285714,
|
| 822 |
+
"grad_norm": 0.21258144080638885,
|
| 823 |
+
"learning_rate": 3.937599271917292e-05,
|
| 824 |
+
"loss": 0.056,
|
| 825 |
+
"num_tokens": 6263112.0,
|
| 826 |
"step": 510
|
| 827 |
},
|
| 828 |
{
|
| 829 |
+
"epoch": 1.0819327731092436,
|
| 830 |
+
"grad_norm": 0.2708013355731964,
|
| 831 |
+
"learning_rate": 3.915382411699218e-05,
|
| 832 |
+
"loss": 0.0547,
|
| 833 |
+
"num_tokens": 6324552.0,
|
| 834 |
"step": 515
|
| 835 |
},
|
| 836 |
{
|
| 837 |
+
"epoch": 1.092436974789916,
|
| 838 |
+
"grad_norm": 2.9274137020111084,
|
| 839 |
+
"learning_rate": 3.893009170484085e-05,
|
| 840 |
+
"loss": 0.0524,
|
| 841 |
+
"num_tokens": 6385992.0,
|
| 842 |
"step": 520
|
| 843 |
},
|
| 844 |
{
|
| 845 |
+
"epoch": 1.1029411764705883,
|
| 846 |
+
"grad_norm": 0.3301822543144226,
|
| 847 |
+
"learning_rate": 3.870482550505094e-05,
|
| 848 |
+
"loss": 0.0554,
|
| 849 |
+
"num_tokens": 6447432.0,
|
| 850 |
"step": 525
|
| 851 |
},
|
| 852 |
{
|
| 853 |
+
"epoch": 1.1134453781512605,
|
| 854 |
+
"grad_norm": 0.4145117700099945,
|
| 855 |
+
"learning_rate": 3.847805574577123e-05,
|
| 856 |
+
"loss": 0.0551,
|
| 857 |
+
"num_tokens": 6508872.0,
|
| 858 |
"step": 530
|
| 859 |
},
|
| 860 |
{
|
| 861 |
+
"epoch": 1.1239495798319328,
|
| 862 |
+
"grad_norm": 0.2403404861688614,
|
| 863 |
+
"learning_rate": 3.8249812856910985e-05,
|
| 864 |
+
"loss": 0.0576,
|
| 865 |
+
"num_tokens": 6570312.0,
|
| 866 |
"step": 535
|
| 867 |
},
|
| 868 |
{
|
| 869 |
+
"epoch": 1.134453781512605,
|
| 870 |
+
"grad_norm": 0.2703566551208496,
|
| 871 |
+
"learning_rate": 3.8020127466056636e-05,
|
| 872 |
+
"loss": 0.0526,
|
| 873 |
+
"num_tokens": 6631553.0,
|
| 874 |
"step": 540
|
| 875 |
},
|
| 876 |
{
|
| 877 |
+
"epoch": 1.1449579831932772,
|
| 878 |
+
"grad_norm": 0.23629266023635864,
|
| 879 |
+
"learning_rate": 3.778903039436189e-05,
|
| 880 |
+
"loss": 0.053,
|
| 881 |
+
"num_tokens": 6692993.0,
|
| 882 |
"step": 545
|
| 883 |
},
|
| 884 |
{
|
| 885 |
+
"epoch": 1.1554621848739495,
|
| 886 |
+
"grad_norm": 31.853532791137695,
|
| 887 |
+
"learning_rate": 3.755655265241187e-05,
|
| 888 |
+
"loss": 0.0551,
|
| 889 |
+
"num_tokens": 6754394.0,
|
| 890 |
"step": 550
|
| 891 |
},
|
| 892 |
{
|
| 893 |
+
"epoch": 1.165966386554622,
|
| 894 |
+
"grad_norm": 0.274700790643692,
|
| 895 |
+
"learning_rate": 3.7322725436061875e-05,
|
| 896 |
+
"loss": 0.0534,
|
| 897 |
+
"num_tokens": 6815834.0,
|
| 898 |
"step": 555
|
| 899 |
},
|
| 900 |
{
|
| 901 |
+
"epoch": 1.1764705882352942,
|
| 902 |
+
"grad_norm": 0.26774168014526367,
|
| 903 |
+
"learning_rate": 3.708758012225125e-05,
|
| 904 |
+
"loss": 0.0528,
|
| 905 |
+
"num_tokens": 6877269.0,
|
| 906 |
"step": 560
|
| 907 |
},
|
| 908 |
{
|
| 909 |
+
"epoch": 1.1869747899159664,
|
| 910 |
+
"grad_norm": 0.2192794382572174,
|
| 911 |
+
"learning_rate": 3.685114826479292e-05,
|
| 912 |
+
"loss": 0.0543,
|
| 913 |
+
"num_tokens": 6938555.0,
|
| 914 |
"step": 565
|
| 915 |
},
|
| 916 |
{
|
| 917 |
+
"epoch": 1.1974789915966386,
|
| 918 |
+
"grad_norm": 0.4298264980316162,
|
| 919 |
+
"learning_rate": 3.661346159013929e-05,
|
| 920 |
+
"loss": 0.0536,
|
| 921 |
+
"num_tokens": 6999704.0,
|
| 922 |
"step": 570
|
| 923 |
},
|
| 924 |
{
|
| 925 |
+
"epoch": 1.2079831932773109,
|
| 926 |
+
"grad_norm": 0.19733546674251556,
|
| 927 |
+
"learning_rate": 3.637455199312488e-05,
|
| 928 |
+
"loss": 0.053,
|
| 929 |
+
"num_tokens": 7061144.0,
|
| 930 |
"step": 575
|
| 931 |
},
|
| 932 |
{
|
| 933 |
+
"epoch": 1.2184873949579833,
|
| 934 |
+
"grad_norm": 0.22298553586006165,
|
| 935 |
+
"learning_rate": 3.61344515326864e-05,
|
| 936 |
+
"loss": 0.0532,
|
| 937 |
+
"num_tokens": 7122407.0,
|
| 938 |
"step": 580
|
| 939 |
},
|
| 940 |
{
|
| 941 |
+
"epoch": 1.2289915966386555,
|
| 942 |
+
"grad_norm": 0.19753730297088623,
|
| 943 |
+
"learning_rate": 3.5893192427560834e-05,
|
| 944 |
+
"loss": 0.0536,
|
| 945 |
+
"num_tokens": 7183847.0,
|
| 946 |
"step": 585
|
| 947 |
},
|
| 948 |
{
|
| 949 |
+
"epoch": 1.2394957983193278,
|
| 950 |
+
"grad_norm": 0.20278260111808777,
|
| 951 |
+
"learning_rate": 3.565080705196202e-05,
|
| 952 |
+
"loss": 0.0525,
|
| 953 |
+
"num_tokens": 7245125.0,
|
| 954 |
"step": 590
|
| 955 |
},
|
| 956 |
{
|
| 957 |
+
"epoch": 1.25,
|
| 958 |
+
"grad_norm": 0.8037598133087158,
|
| 959 |
+
"learning_rate": 3.5407327931236434e-05,
|
| 960 |
+
"loss": 0.0536,
|
| 961 |
+
"num_tokens": 7306565.0,
|
| 962 |
"step": 595
|
| 963 |
},
|
| 964 |
{
|
| 965 |
+
"epoch": 1.2605042016806722,
|
| 966 |
+
"grad_norm": 0.25609511137008667,
|
| 967 |
+
"learning_rate": 3.516278773749863e-05,
|
| 968 |
+
"loss": 0.0534,
|
| 969 |
+
"num_tokens": 7368005.0,
|
| 970 |
"step": 600
|
| 971 |
},
|
| 972 |
{
|
| 973 |
+
"epoch": 1.2710084033613445,
|
| 974 |
+
"grad_norm": 0.18984173238277435,
|
| 975 |
+
"learning_rate": 3.4917219285247036e-05,
|
| 976 |
+
"loss": 0.0517,
|
| 977 |
+
"num_tokens": 7429445.0,
|
| 978 |
"step": 605
|
| 979 |
},
|
| 980 |
{
|
| 981 |
+
"epoch": 1.2815126050420167,
|
| 982 |
+
"grad_norm": 0.17902691662311554,
|
| 983 |
+
"learning_rate": 3.4670655526960627e-05,
|
| 984 |
+
"loss": 0.0538,
|
| 985 |
+
"num_tokens": 7490885.0,
|
| 986 |
"step": 610
|
| 987 |
},
|
| 988 |
{
|
| 989 |
+
"epoch": 1.2920168067226891,
|
| 990 |
+
"grad_norm": 0.19289465248584747,
|
| 991 |
+
"learning_rate": 3.4423129548677055e-05,
|
| 992 |
+
"loss": 0.0526,
|
| 993 |
+
"num_tokens": 7552325.0,
|
| 994 |
"step": 615
|
| 995 |
},
|
| 996 |
{
|
| 997 |
+
"epoch": 1.3025210084033614,
|
| 998 |
+
"grad_norm": 0.21860499680042267,
|
| 999 |
+
"learning_rate": 3.41746745655529e-05,
|
| 1000 |
+
"loss": 0.0546,
|
| 1001 |
+
"num_tokens": 7613765.0,
|
| 1002 |
"step": 620
|
| 1003 |
},
|
| 1004 |
{
|
| 1005 |
+
"epoch": 1.3130252100840336,
|
| 1006 |
+
"grad_norm": 0.19712497293949127,
|
| 1007 |
+
"learning_rate": 3.3925323917406574e-05,
|
| 1008 |
+
"loss": 0.0538,
|
| 1009 |
+
"num_tokens": 7675205.0,
|
| 1010 |
"step": 625
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
+
"epoch": 1.3235294117647058,
|
| 1014 |
+
"grad_norm": 0.22082890570163727,
|
| 1015 |
+
"learning_rate": 3.3675111064244504e-05,
|
| 1016 |
+
"loss": 0.0537,
|
| 1017 |
+
"num_tokens": 7736645.0,
|
| 1018 |
"step": 630
|
| 1019 |
},
|
| 1020 |
{
|
| 1021 |
+
"epoch": 1.334033613445378,
|
| 1022 |
+
"grad_norm": 0.20152664184570312,
|
| 1023 |
+
"learning_rate": 3.3424069581771155e-05,
|
| 1024 |
+
"loss": 0.0529,
|
| 1025 |
+
"num_tokens": 7798085.0,
|
| 1026 |
"step": 635
|
| 1027 |
},
|
| 1028 |
{
|
| 1029 |
+
"epoch": 1.3445378151260505,
|
| 1030 |
+
"grad_norm": 0.22678756713867188,
|
| 1031 |
+
"learning_rate": 3.317223315688358e-05,
|
| 1032 |
+
"loss": 0.0539,
|
| 1033 |
+
"num_tokens": 7859525.0,
|
| 1034 |
"step": 640
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
+
"epoch": 1.3550420168067228,
|
| 1038 |
+
"grad_norm": 0.2191995084285736,
|
| 1039 |
+
"learning_rate": 3.2919635583151025e-05,
|
| 1040 |
+
"loss": 0.0529,
|
| 1041 |
+
"num_tokens": 7920965.0,
|
| 1042 |
"step": 645
|
| 1043 |
},
|
| 1044 |
{
|
| 1045 |
+
"epoch": 1.365546218487395,
|
| 1046 |
+
"grad_norm": 0.18456892669200897,
|
| 1047 |
+
"learning_rate": 3.2666310756280194e-05,
|
| 1048 |
+
"loss": 0.0544,
|
| 1049 |
+
"num_tokens": 7982405.0,
|
| 1050 |
"step": 650
|
| 1051 |
},
|
| 1052 |
{
|
| 1053 |
+
"epoch": 1.3760504201680672,
|
| 1054 |
+
"grad_norm": 0.1721736490726471,
|
| 1055 |
+
"learning_rate": 3.241229266956687e-05,
|
| 1056 |
+
"loss": 0.054,
|
| 1057 |
+
"num_tokens": 8043845.0,
|
| 1058 |
"step": 655
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
+
"epoch": 1.3865546218487395,
|
| 1062 |
+
"grad_norm": 0.22381868958473206,
|
| 1063 |
+
"learning_rate": 3.215761540933436e-05,
|
| 1064 |
+
"loss": 0.0525,
|
| 1065 |
+
"num_tokens": 8105285.0,
|
| 1066 |
"step": 660
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
+
"epoch": 1.3970588235294117,
|
| 1070 |
+
"grad_norm": 0.19951923191547394,
|
| 1071 |
+
"learning_rate": 3.190231315035954e-05,
|
| 1072 |
+
"loss": 0.0514,
|
| 1073 |
+
"num_tokens": 8166725.0,
|
| 1074 |
"step": 665
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
+
"epoch": 1.407563025210084,
|
| 1078 |
+
"grad_norm": 0.21700704097747803,
|
| 1079 |
+
"learning_rate": 3.164642015128694e-05,
|
| 1080 |
+
"loss": 0.0531,
|
| 1081 |
+
"num_tokens": 8228159.0,
|
| 1082 |
"step": 670
|
| 1083 |
},
|
| 1084 |
{
|
| 1085 |
+
"epoch": 1.4180672268907564,
|
| 1086 |
+
"grad_norm": 0.13238979876041412,
|
| 1087 |
+
"learning_rate": 3.13899707500317e-05,
|
| 1088 |
+
"loss": 0.0503,
|
| 1089 |
+
"num_tokens": 8289370.0,
|
| 1090 |
"step": 675
|
| 1091 |
},
|
| 1092 |
{
|
| 1093 |
+
"epoch": 1.4285714285714286,
|
| 1094 |
+
"grad_norm": 0.16383114457130432,
|
| 1095 |
+
"learning_rate": 3.1132999359171737e-05,
|
| 1096 |
+
"loss": 0.0513,
|
| 1097 |
+
"num_tokens": 8350810.0,
|
| 1098 |
"step": 680
|
| 1099 |
},
|
| 1100 |
{
|
| 1101 |
+
"epoch": 1.4390756302521008,
|
| 1102 |
+
"grad_norm": 0.18902327120304108,
|
| 1103 |
+
"learning_rate": 3.087554046133004e-05,
|
| 1104 |
+
"loss": 0.052,
|
| 1105 |
+
"num_tokens": 8412174.0,
|
| 1106 |
"step": 685
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
+
"epoch": 1.449579831932773,
|
| 1110 |
+
"grad_norm": 0.16599516570568085,
|
| 1111 |
+
"learning_rate": 3.0617628604547424e-05,
|
| 1112 |
+
"loss": 0.0533,
|
| 1113 |
+
"num_tokens": 8473614.0,
|
| 1114 |
"step": 690
|
| 1115 |
},
|
| 1116 |
{
|
| 1117 |
+
"epoch": 1.4600840336134453,
|
| 1118 |
+
"grad_norm": 0.20513266324996948,
|
| 1119 |
+
"learning_rate": 3.035929839764665e-05,
|
| 1120 |
+
"loss": 0.0507,
|
| 1121 |
+
"num_tokens": 8535054.0,
|
| 1122 |
"step": 695
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
+
"epoch": 1.4705882352941178,
|
| 1126 |
+
"grad_norm": 0.22719748318195343,
|
| 1127 |
+
"learning_rate": 3.0100584505588275e-05,
|
| 1128 |
+
"loss": 0.052,
|
| 1129 |
+
"num_tokens": 8596494.0,
|
| 1130 |
"step": 700
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
+
"epoch": 1.48109243697479,
|
| 1134 |
+
"grad_norm": 0.18906480073928833,
|
| 1135 |
+
"learning_rate": 2.9841521644818976e-05,
|
| 1136 |
+
"loss": 0.0516,
|
| 1137 |
+
"num_tokens": 8657934.0,
|
| 1138 |
"step": 705
|
| 1139 |
},
|
| 1140 |
{
|
| 1141 |
+
"epoch": 1.4915966386554622,
|
| 1142 |
+
"grad_norm": 0.17335395514965057,
|
| 1143 |
+
"learning_rate": 2.9582144578613102e-05,
|
| 1144 |
+
"loss": 0.0496,
|
| 1145 |
+
"num_tokens": 8719374.0,
|
| 1146 |
"step": 710
|
| 1147 |
},
|
| 1148 |
{
|
| 1149 |
+
"epoch": 1.5021008403361344,
|
| 1150 |
+
"grad_norm": 0.20907028019428253,
|
| 1151 |
+
"learning_rate": 2.9322488112407743e-05,
|
| 1152 |
+
"loss": 0.0523,
|
| 1153 |
+
"num_tokens": 8780740.0,
|
| 1154 |
"step": 715
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
+
"epoch": 1.5126050420168067,
|
| 1158 |
+
"grad_norm": 0.21596895158290863,
|
| 1159 |
+
"learning_rate": 2.906258708913228e-05,
|
| 1160 |
+
"loss": 0.053,
|
| 1161 |
+
"num_tokens": 8842180.0,
|
| 1162 |
"step": 720
|
| 1163 |
},
|
| 1164 |
{
|
| 1165 |
+
"epoch": 1.523109243697479,
|
| 1166 |
+
"grad_norm": 0.21814534068107605,
|
| 1167 |
+
"learning_rate": 2.880247638453288e-05,
|
| 1168 |
+
"loss": 0.0535,
|
| 1169 |
+
"num_tokens": 8903620.0,
|
| 1170 |
"step": 725
|
| 1171 |
},
|
| 1172 |
{
|
| 1173 |
+
"epoch": 1.5336134453781511,
|
| 1174 |
+
"grad_norm": 0.17172180116176605,
|
| 1175 |
+
"learning_rate": 2.854219090249251e-05,
|
| 1176 |
+
"loss": 0.0511,
|
| 1177 |
+
"num_tokens": 8965060.0,
|
| 1178 |
"step": 730
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
+
"epoch": 1.5441176470588234,
|
| 1182 |
+
"grad_norm": 0.144153892993927,
|
| 1183 |
+
"learning_rate": 2.8281765570347306e-05,
|
| 1184 |
+
"loss": 0.0509,
|
| 1185 |
+
"num_tokens": 9026344.0,
|
| 1186 |
"step": 735
|
| 1187 |
},
|
| 1188 |
{
|
| 1189 |
+
"epoch": 1.5546218487394958,
|
| 1190 |
+
"grad_norm": 0.1880050003528595,
|
| 1191 |
+
"learning_rate": 2.802123533419966e-05,
|
| 1192 |
+
"loss": 0.0546,
|
| 1193 |
+
"num_tokens": 9087784.0,
|
| 1194 |
"step": 740
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
+
"epoch": 1.565126050420168,
|
| 1198 |
+
"grad_norm": 0.15667995810508728,
|
| 1199 |
+
"learning_rate": 2.7760635154228896e-05,
|
| 1200 |
+
"loss": 0.051,
|
| 1201 |
+
"num_tokens": 9149063.0,
|
| 1202 |
"step": 745
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
+
"epoch": 1.5756302521008403,
|
| 1206 |
+
"grad_norm": 0.2708027958869934,
|
| 1207 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 1208 |
+
"loss": 0.0544,
|
| 1209 |
+
"num_tokens": 9210503.0,
|
| 1210 |
"step": 750
|
| 1211 |
},
|
| 1212 |
{
|
| 1213 |
+
"epoch": 1.5861344537815127,
|
| 1214 |
+
"grad_norm": 0.1797020584344864,
|
| 1215 |
+
"learning_rate": 2.723936484577111e-05,
|
| 1216 |
+
"loss": 0.0528,
|
| 1217 |
+
"num_tokens": 9271881.0,
|
| 1218 |
"step": 755
|
| 1219 |
},
|
| 1220 |
{
|
| 1221 |
+
"epoch": 1.596638655462185,
|
| 1222 |
+
"grad_norm": 0.21367360651493073,
|
| 1223 |
+
"learning_rate": 2.6978764665800343e-05,
|
| 1224 |
+
"loss": 0.0535,
|
| 1225 |
+
"num_tokens": 9333321.0,
|
| 1226 |
"step": 760
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
+
"epoch": 1.6071428571428572,
|
| 1230 |
+
"grad_norm": 0.21819233894348145,
|
| 1231 |
+
"learning_rate": 2.67182344296527e-05,
|
| 1232 |
+
"loss": 0.0522,
|
| 1233 |
+
"num_tokens": 9394761.0,
|
| 1234 |
"step": 765
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
+
"epoch": 1.6176470588235294,
|
| 1238 |
+
"grad_norm": 0.1992005854845047,
|
| 1239 |
+
"learning_rate": 2.6457809097507496e-05,
|
| 1240 |
+
"loss": 0.0506,
|
| 1241 |
+
"num_tokens": 9456201.0,
|
| 1242 |
"step": 770
|
| 1243 |
},
|
| 1244 |
{
|
| 1245 |
+
"epoch": 1.6281512605042017,
|
| 1246 |
+
"grad_norm": 0.19866126775741577,
|
| 1247 |
+
"learning_rate": 2.619752361546713e-05,
|
| 1248 |
+
"loss": 0.0518,
|
| 1249 |
+
"num_tokens": 9517492.0,
|
| 1250 |
"step": 775
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
+
"epoch": 1.638655462184874,
|
| 1254 |
+
"grad_norm": 0.174868643283844,
|
| 1255 |
+
"learning_rate": 2.593741291086772e-05,
|
| 1256 |
+
"loss": 0.0532,
|
| 1257 |
+
"num_tokens": 9578932.0,
|
| 1258 |
"step": 780
|
| 1259 |
},
|
| 1260 |
{
|
| 1261 |
+
"epoch": 1.6491596638655461,
|
| 1262 |
+
"grad_norm": 0.22887223958969116,
|
| 1263 |
+
"learning_rate": 2.567751188759227e-05,
|
| 1264 |
+
"loss": 0.0523,
|
| 1265 |
+
"num_tokens": 9640372.0,
|
| 1266 |
"step": 785
|
| 1267 |
},
|
| 1268 |
{
|
| 1269 |
+
"epoch": 1.6596638655462184,
|
| 1270 |
+
"grad_norm": 0.17208142578601837,
|
| 1271 |
+
"learning_rate": 2.541785542138691e-05,
|
| 1272 |
+
"loss": 0.0502,
|
| 1273 |
+
"num_tokens": 9701812.0,
|
| 1274 |
"step": 790
|
| 1275 |
},
|
| 1276 |
{
|
| 1277 |
+
"epoch": 1.6701680672268906,
|
| 1278 |
+
"grad_norm": 0.21313603222370148,
|
| 1279 |
+
"learning_rate": 2.515847835518103e-05,
|
| 1280 |
+
"loss": 0.0526,
|
| 1281 |
+
"num_tokens": 9763075.0,
|
| 1282 |
"step": 795
|
| 1283 |
},
|
| 1284 |
{
|
| 1285 |
+
"epoch": 1.680672268907563,
|
| 1286 |
+
"grad_norm": 0.15035264194011688,
|
| 1287 |
+
"learning_rate": 2.4899415494411737e-05,
|
| 1288 |
+
"loss": 0.0507,
|
| 1289 |
+
"num_tokens": 9824515.0,
|
| 1290 |
"step": 800
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
+
"epoch": 1.6911764705882353,
|
| 1294 |
+
"grad_norm": 0.2601998746395111,
|
| 1295 |
+
"learning_rate": 2.464070160235335e-05,
|
| 1296 |
+
"loss": 0.0526,
|
| 1297 |
+
"num_tokens": 9885955.0,
|
| 1298 |
"step": 805
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
+
"epoch": 1.7016806722689075,
|
| 1302 |
+
"grad_norm": 0.17123112082481384,
|
| 1303 |
+
"learning_rate": 2.438237139545258e-05,
|
| 1304 |
+
"loss": 0.0521,
|
| 1305 |
+
"num_tokens": 9947395.0,
|
| 1306 |
"step": 810
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
+
"epoch": 1.71218487394958,
|
| 1310 |
+
"grad_norm": 0.17243990302085876,
|
| 1311 |
+
"learning_rate": 2.412445953866997e-05,
|
| 1312 |
+
"loss": 0.0502,
|
| 1313 |
+
"num_tokens": 10008835.0,
|
| 1314 |
"step": 815
|
| 1315 |
},
|
| 1316 |
{
|
| 1317 |
+
"epoch": 1.7226890756302522,
|
| 1318 |
+
"grad_norm": 0.21723228693008423,
|
| 1319 |
+
"learning_rate": 2.386700064082827e-05,
|
| 1320 |
+
"loss": 0.0517,
|
| 1321 |
+
"num_tokens": 10070123.0,
|
| 1322 |
"step": 820
|
| 1323 |
},
|
| 1324 |
{
|
| 1325 |
+
"epoch": 1.7331932773109244,
|
| 1326 |
+
"grad_norm": 0.13738787174224854,
|
| 1327 |
+
"learning_rate": 2.361002924996831e-05,
|
| 1328 |
+
"loss": 0.051,
|
| 1329 |
+
"num_tokens": 10131563.0,
|
| 1330 |
"step": 825
|
| 1331 |
},
|
| 1332 |
{
|
| 1333 |
+
"epoch": 1.7436974789915967,
|
| 1334 |
+
"grad_norm": 0.21257147192955017,
|
| 1335 |
+
"learning_rate": 2.3353579848713063e-05,
|
| 1336 |
+
"loss": 0.0522,
|
| 1337 |
+
"num_tokens": 10192967.0,
|
| 1338 |
"step": 830
|
| 1339 |
},
|
| 1340 |
{
|
| 1341 |
+
"epoch": 1.754201680672269,
|
| 1342 |
+
"grad_norm": 0.20029078423976898,
|
| 1343 |
+
"learning_rate": 2.3097686849640476e-05,
|
| 1344 |
+
"loss": 0.0543,
|
| 1345 |
+
"num_tokens": 10254407.0,
|
| 1346 |
"step": 835
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
+
"epoch": 1.7647058823529411,
|
| 1350 |
+
"grad_norm": 0.20497262477874756,
|
| 1351 |
+
"learning_rate": 2.2842384590665645e-05,
|
| 1352 |
+
"loss": 0.0526,
|
| 1353 |
+
"num_tokens": 10315847.0,
|
| 1354 |
"step": 840
|
| 1355 |
},
|
| 1356 |
{
|
| 1357 |
+
"epoch": 1.7752100840336134,
|
| 1358 |
+
"grad_norm": 0.19529034197330475,
|
| 1359 |
+
"learning_rate": 2.2587707330433133e-05,
|
| 1360 |
+
"loss": 0.052,
|
| 1361 |
+
"num_tokens": 10377287.0,
|
| 1362 |
"step": 845
|
| 1363 |
},
|
| 1364 |
{
|
| 1365 |
+
"epoch": 1.7857142857142856,
|
| 1366 |
+
"grad_norm": 0.19968290627002716,
|
| 1367 |
+
"learning_rate": 2.23336892437198e-05,
|
| 1368 |
+
"loss": 0.0511,
|
| 1369 |
+
"num_tokens": 10438565.0,
|
| 1370 |
"step": 850
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
+
"epoch": 1.7962184873949578,
|
| 1374 |
+
"grad_norm": 0.20312048494815826,
|
| 1375 |
+
"learning_rate": 2.2080364416848987e-05,
|
| 1376 |
+
"loss": 0.0508,
|
| 1377 |
+
"num_tokens": 10500005.0,
|
| 1378 |
"step": 855
|
| 1379 |
},
|
| 1380 |
{
|
| 1381 |
+
"epoch": 1.8067226890756303,
|
| 1382 |
+
"grad_norm": 0.2170594483613968,
|
| 1383 |
+
"learning_rate": 2.1827766843116428e-05,
|
| 1384 |
+
"loss": 0.052,
|
| 1385 |
+
"num_tokens": 10561445.0,
|
| 1386 |
"step": 860
|
| 1387 |
},
|
| 1388 |
{
|
| 1389 |
+
"epoch": 1.8172268907563025,
|
| 1390 |
+
"grad_norm": 0.20793762803077698,
|
| 1391 |
+
"learning_rate": 2.157593041822885e-05,
|
| 1392 |
+
"loss": 0.0507,
|
| 1393 |
+
"num_tokens": 10622885.0,
|
| 1394 |
"step": 865
|
| 1395 |
},
|
| 1396 |
{
|
| 1397 |
+
"epoch": 1.8277310924369747,
|
| 1398 |
+
"grad_norm": 0.18194827437400818,
|
| 1399 |
+
"learning_rate": 2.1324888935755498e-05,
|
| 1400 |
+
"loss": 0.0512,
|
| 1401 |
+
"num_tokens": 10684325.0,
|
| 1402 |
"step": 870
|
| 1403 |
},
|
| 1404 |
{
|
| 1405 |
+
"epoch": 1.8382352941176472,
|
| 1406 |
+
"grad_norm": 0.14043785631656647,
|
| 1407 |
+
"learning_rate": 2.1074676082593425e-05,
|
| 1408 |
+
"loss": 0.0507,
|
| 1409 |
+
"num_tokens": 10745533.0,
|
| 1410 |
"step": 875
|
| 1411 |
},
|
| 1412 |
{
|
| 1413 |
+
"epoch": 1.8487394957983194,
|
| 1414 |
+
"grad_norm": 0.17620113492012024,
|
| 1415 |
+
"learning_rate": 2.0825325434447106e-05,
|
| 1416 |
+
"loss": 0.0526,
|
| 1417 |
+
"num_tokens": 10806971.0,
|
| 1418 |
"step": 880
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
+
"epoch": 1.8592436974789917,
|
| 1422 |
+
"grad_norm": 0.17084655165672302,
|
| 1423 |
+
"learning_rate": 2.0576870451322953e-05,
|
| 1424 |
+
"loss": 0.05,
|
| 1425 |
+
"num_tokens": 10868411.0,
|
| 1426 |
"step": 885
|
| 1427 |
},
|
| 1428 |
{
|
| 1429 |
+
"epoch": 1.8697478991596639,
|
| 1430 |
+
"grad_norm": 0.17954443395137787,
|
| 1431 |
+
"learning_rate": 2.032934447303938e-05,
|
| 1432 |
+
"loss": 0.0479,
|
| 1433 |
+
"num_tokens": 10929851.0,
|
| 1434 |
"step": 890
|
| 1435 |
},
|
| 1436 |
{
|
| 1437 |
+
"epoch": 1.8802521008403361,
|
| 1438 |
+
"grad_norm": 0.19004443287849426,
|
| 1439 |
+
"learning_rate": 2.0082780714752963e-05,
|
| 1440 |
+
"loss": 0.0516,
|
| 1441 |
+
"num_tokens": 10991291.0,
|
| 1442 |
"step": 895
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
+
"epoch": 1.8907563025210083,
|
| 1446 |
+
"grad_norm": 0.1933521330356598,
|
| 1447 |
+
"learning_rate": 1.9837212262501382e-05,
|
| 1448 |
+
"loss": 0.0526,
|
| 1449 |
+
"num_tokens": 11052731.0,
|
| 1450 |
"step": 900
|
| 1451 |
},
|
| 1452 |
{
|
| 1453 |
+
"epoch": 1.9012605042016806,
|
| 1454 |
+
"grad_norm": 0.1794319450855255,
|
| 1455 |
+
"learning_rate": 1.9592672068763574e-05,
|
| 1456 |
+
"loss": 0.052,
|
| 1457 |
+
"num_tokens": 11114068.0,
|
| 1458 |
"step": 905
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
+
"epoch": 1.9117647058823528,
|
| 1462 |
+
"grad_norm": 0.16216090321540833,
|
| 1463 |
+
"learning_rate": 1.934919294803798e-05,
|
| 1464 |
+
"loss": 0.0519,
|
| 1465 |
+
"num_tokens": 11175508.0,
|
| 1466 |
"step": 910
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
+
"epoch": 1.9222689075630253,
|
| 1470 |
+
"grad_norm": 0.19467321038246155,
|
| 1471 |
+
"learning_rate": 1.9106807572439168e-05,
|
| 1472 |
"loss": 0.0506,
|
| 1473 |
+
"num_tokens": 11236948.0,
|
| 1474 |
"step": 915
|
| 1475 |
},
|
| 1476 |
{
|
| 1477 |
+
"epoch": 1.9327731092436975,
|
| 1478 |
+
"grad_norm": 0.13739857077598572,
|
| 1479 |
+
"learning_rate": 1.88655484673136e-05,
|
| 1480 |
+
"loss": 0.0516,
|
| 1481 |
+
"num_tokens": 11298388.0,
|
| 1482 |
"step": 920
|
| 1483 |
},
|
| 1484 |
{
|
| 1485 |
+
"epoch": 1.9432773109243697,
|
| 1486 |
+
"grad_norm": 0.15686306357383728,
|
| 1487 |
+
"learning_rate": 1.8625448006875123e-05,
|
| 1488 |
+
"loss": 0.0505,
|
| 1489 |
+
"num_tokens": 11359828.0,
|
| 1490 |
"step": 925
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
+
"epoch": 1.9537815126050422,
|
| 1494 |
+
"grad_norm": 0.12999138236045837,
|
| 1495 |
+
"learning_rate": 1.8386538409860708e-05,
|
| 1496 |
+
"loss": 0.051,
|
| 1497 |
+
"num_tokens": 11421268.0,
|
| 1498 |
"step": 930
|
| 1499 |
},
|
| 1500 |
{
|
| 1501 |
+
"epoch": 1.9642857142857144,
|
| 1502 |
+
"grad_norm": 0.18375808000564575,
|
| 1503 |
+
"learning_rate": 1.8148851735207083e-05,
|
| 1504 |
"loss": 0.0523,
|
| 1505 |
+
"num_tokens": 11482548.0,
|
| 1506 |
"step": 935
|
| 1507 |
},
|
| 1508 |
{
|
| 1509 |
+
"epoch": 1.9747899159663866,
|
| 1510 |
+
"grad_norm": 0.19671285152435303,
|
| 1511 |
+
"learning_rate": 1.791241987774876e-05,
|
| 1512 |
+
"loss": 0.0509,
|
| 1513 |
+
"num_tokens": 11543988.0,
|
| 1514 |
"step": 940
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
+
"epoch": 1.9852941176470589,
|
| 1518 |
+
"grad_norm": 0.1805330216884613,
|
| 1519 |
+
"learning_rate": 1.7677274563938134e-05,
|
| 1520 |
+
"loss": 0.0503,
|
| 1521 |
+
"num_tokens": 11605268.0,
|
| 1522 |
"step": 945
|
| 1523 |
},
|
| 1524 |
{
|
| 1525 |
+
"epoch": 1.995798319327731,
|
| 1526 |
+
"grad_norm": 0.19455303251743317,
|
| 1527 |
+
"learning_rate": 1.744344734758814e-05,
|
| 1528 |
+
"loss": 0.0517,
|
| 1529 |
+
"num_tokens": 11666708.0,
|
| 1530 |
"step": 950
|
| 1531 |
},
|
| 1532 |
{
|
| 1533 |
+
"epoch": 2.0063025210084033,
|
| 1534 |
+
"grad_norm": 0.17816315591335297,
|
| 1535 |
+
"learning_rate": 1.721096960563812e-05,
|
| 1536 |
+
"loss": 0.0507,
|
| 1537 |
+
"num_tokens": 11728148.0,
|
| 1538 |
"step": 955
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
+
"epoch": 2.0168067226890756,
|
| 1542 |
+
"grad_norm": 0.12756673991680145,
|
| 1543 |
+
"learning_rate": 1.697987253394337e-05,
|
| 1544 |
+
"loss": 0.0491,
|
| 1545 |
+
"num_tokens": 11789273.0,
|
| 1546 |
"step": 960
|
| 1547 |
},
|
| 1548 |
{
|
| 1549 |
+
"epoch": 2.027310924369748,
|
| 1550 |
+
"grad_norm": 0.19657427072525024,
|
| 1551 |
+
"learning_rate": 1.675018714308902e-05,
|
| 1552 |
+
"loss": 0.0504,
|
| 1553 |
+
"num_tokens": 11850713.0,
|
| 1554 |
"step": 965
|
| 1555 |
},
|
| 1556 |
{
|
| 1557 |
+
"epoch": 2.03781512605042,
|
| 1558 |
+
"grad_norm": 0.1950300633907318,
|
| 1559 |
+
"learning_rate": 1.652194425422878e-05,
|
| 1560 |
+
"loss": 0.0505,
|
| 1561 |
+
"num_tokens": 11912153.0,
|
| 1562 |
"step": 970
|
| 1563 |
},
|
| 1564 |
{
|
| 1565 |
+
"epoch": 2.0483193277310923,
|
| 1566 |
+
"grad_norm": 0.16631367802619934,
|
| 1567 |
+
"learning_rate": 1.629517449494906e-05,
|
| 1568 |
+
"loss": 0.0502,
|
| 1569 |
+
"num_tokens": 11973593.0,
|
| 1570 |
"step": 975
|
| 1571 |
},
|
| 1572 |
{
|
| 1573 |
+
"epoch": 2.0588235294117645,
|
| 1574 |
+
"grad_norm": 0.17350395023822784,
|
| 1575 |
+
"learning_rate": 1.6069908295159146e-05,
|
| 1576 |
+
"loss": 0.0526,
|
| 1577 |
+
"num_tokens": 12035033.0,
|
| 1578 |
"step": 980
|
| 1579 |
},
|
| 1580 |
{
|
| 1581 |
+
"epoch": 2.069327731092437,
|
| 1582 |
+
"grad_norm": 0.18997882306575775,
|
| 1583 |
+
"learning_rate": 1.5846175883007815e-05,
|
| 1584 |
+
"loss": 0.0493,
|
| 1585 |
+
"num_tokens": 12096473.0,
|
| 1586 |
"step": 985
|
| 1587 |
},
|
| 1588 |
{
|
| 1589 |
+
"epoch": 2.0798319327731094,
|
| 1590 |
+
"grad_norm": 0.1386975198984146,
|
| 1591 |
+
"learning_rate": 1.562400728082709e-05,
|
| 1592 |
+
"loss": 0.0497,
|
| 1593 |
+
"num_tokens": 12157913.0,
|
| 1594 |
"step": 990
|
| 1595 |
},
|
| 1596 |
{
|
| 1597 |
+
"epoch": 2.0903361344537816,
|
| 1598 |
+
"grad_norm": 0.1656985878944397,
|
| 1599 |
+
"learning_rate": 1.540343230110354e-05,
|
| 1600 |
+
"loss": 0.0509,
|
| 1601 |
+
"num_tokens": 12219353.0,
|
| 1602 |
"step": 995
|
| 1603 |
},
|
| 1604 |
{
|
| 1605 |
+
"epoch": 2.100840336134454,
|
| 1606 |
+
"grad_norm": 0.19251607358455658,
|
| 1607 |
+
"learning_rate": 1.5184480542477869e-05,
|
| 1608 |
+
"loss": 0.0503,
|
| 1609 |
+
"num_tokens": 12280793.0,
|
| 1610 |
"step": 1000
|
| 1611 |
},
|
| 1612 |
{
|
| 1613 |
+
"epoch": 2.111344537815126,
|
| 1614 |
+
"grad_norm": 0.17274506390094757,
|
| 1615 |
+
"learning_rate": 1.4967181385773022e-05,
|
| 1616 |
+
"loss": 0.0491,
|
| 1617 |
+
"num_tokens": 12342004.0,
|
| 1618 |
"step": 1005
|
| 1619 |
},
|
| 1620 |
{
|
| 1621 |
+
"epoch": 2.1218487394957983,
|
| 1622 |
+
"grad_norm": 0.20883677899837494,
|
| 1623 |
+
"learning_rate": 1.4751563990051675e-05,
|
| 1624 |
+
"loss": 0.0495,
|
| 1625 |
+
"num_tokens": 12403444.0,
|
| 1626 |
"step": 1010
|
| 1627 |
},
|
| 1628 |
{
|
| 1629 |
+
"epoch": 2.1323529411764706,
|
| 1630 |
+
"grad_norm": 0.20437228679656982,
|
| 1631 |
+
"learning_rate": 1.453765728870343e-05,
|
| 1632 |
+
"loss": 0.0514,
|
| 1633 |
+
"num_tokens": 12464884.0,
|
| 1634 |
"step": 1015
|
| 1635 |
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 2.142857142857143,
|
| 1638 |
+
"grad_norm": 0.20462237298488617,
|
| 1639 |
+
"learning_rate": 1.432548998556221e-05,
|
| 1640 |
+
"loss": 0.051,
|
| 1641 |
+
"num_tokens": 12526175.0,
|
| 1642 |
+
"step": 1020
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 2.153361344537815,
|
| 1646 |
+
"grad_norm": 0.2599621117115021,
|
| 1647 |
+
"learning_rate": 1.4115090551054622e-05,
|
| 1648 |
+
"loss": 0.0517,
|
| 1649 |
+
"num_tokens": 12587615.0,
|
| 1650 |
+
"step": 1025
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 2.1638655462184873,
|
| 1654 |
+
"grad_norm": 0.1801358163356781,
|
| 1655 |
+
"learning_rate": 1.3906487218379504e-05,
|
| 1656 |
+
"loss": 0.0499,
|
| 1657 |
+
"num_tokens": 12649055.0,
|
| 1658 |
+
"step": 1030
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 2.1743697478991595,
|
| 1662 |
+
"grad_norm": 0.1843215674161911,
|
| 1663 |
+
"learning_rate": 1.3699707979719357e-05,
|
| 1664 |
+
"loss": 0.0513,
|
| 1665 |
+
"num_tokens": 12710459.0,
|
| 1666 |
+
"step": 1035
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 2.184873949579832,
|
| 1670 |
+
"grad_norm": 0.19053132832050323,
|
| 1671 |
+
"learning_rate": 1.3494780582484126e-05,
|
| 1672 |
+
"loss": 0.0496,
|
| 1673 |
+
"num_tokens": 12771899.0,
|
| 1674 |
+
"step": 1040
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 2.1953781512605044,
|
| 1678 |
+
"grad_norm": 0.15285778045654297,
|
| 1679 |
+
"learning_rate": 1.329173252558779e-05,
|
| 1680 |
+
"loss": 0.0497,
|
| 1681 |
+
"num_tokens": 12833339.0,
|
| 1682 |
+
"step": 1045
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.2058823529411766,
|
| 1686 |
+
"grad_norm": 0.14396464824676514,
|
| 1687 |
+
"learning_rate": 1.3090591055758356e-05,
|
| 1688 |
+
"loss": 0.0507,
|
| 1689 |
+
"num_tokens": 12894779.0,
|
| 1690 |
+
"step": 1050
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 2.216386554621849,
|
| 1694 |
+
"grad_norm": 0.14991876482963562,
|
| 1695 |
+
"learning_rate": 1.2891383163881633e-05,
|
| 1696 |
+
"loss": 0.05,
|
| 1697 |
+
"num_tokens": 12956219.0,
|
| 1698 |
+
"step": 1055
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 2.226890756302521,
|
| 1702 |
+
"grad_norm": 0.14839011430740356,
|
| 1703 |
+
"learning_rate": 1.2694135581379383e-05,
|
| 1704 |
+
"loss": 0.0499,
|
| 1705 |
+
"num_tokens": 13017659.0,
|
| 1706 |
+
"step": 1060
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 2.2373949579831933,
|
| 1710 |
+
"grad_norm": 0.12264993786811829,
|
| 1711 |
+
"learning_rate": 1.2498874776622246e-05,
|
| 1712 |
+
"loss": 0.0462,
|
| 1713 |
+
"num_tokens": 13079099.0,
|
| 1714 |
+
"step": 1065
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 2.2478991596638656,
|
| 1718 |
+
"grad_norm": 0.1659439504146576,
|
| 1719 |
+
"learning_rate": 1.2305626951378019e-05,
|
| 1720 |
+
"loss": 0.0492,
|
| 1721 |
+
"num_tokens": 13140539.0,
|
| 1722 |
+
"step": 1070
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 2.258403361344538,
|
| 1726 |
+
"grad_norm": 0.16605842113494873,
|
| 1727 |
+
"learning_rate": 1.2114418037295636e-05,
|
| 1728 |
+
"loss": 0.0502,
|
| 1729 |
+
"num_tokens": 13201979.0,
|
| 1730 |
+
"step": 1075
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 2.26890756302521,
|
| 1734 |
+
"grad_norm": 0.16009701788425446,
|
| 1735 |
+
"learning_rate": 1.1925273692425487e-05,
|
| 1736 |
+
"loss": 0.0496,
|
| 1737 |
+
"num_tokens": 13263419.0,
|
| 1738 |
+
"step": 1080
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 2.2794117647058822,
|
| 1742 |
+
"grad_norm": 0.1512678861618042,
|
| 1743 |
+
"learning_rate": 1.1738219297776371e-05,
|
| 1744 |
+
"loss": 0.0497,
|
| 1745 |
+
"num_tokens": 13324859.0,
|
| 1746 |
+
"step": 1085
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 2.2899159663865545,
|
| 1750 |
+
"grad_norm": 0.18362964689731598,
|
| 1751 |
+
"learning_rate": 1.1553279953909641e-05,
|
| 1752 |
+
"loss": 0.0485,
|
| 1753 |
+
"num_tokens": 13386299.0,
|
| 1754 |
+
"step": 1090
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 2.3004201680672267,
|
| 1758 |
+
"grad_norm": 0.15746116638183594,
|
| 1759 |
+
"learning_rate": 1.1370480477571029e-05,
|
| 1760 |
+
"loss": 0.0503,
|
| 1761 |
+
"num_tokens": 13447730.0,
|
| 1762 |
+
"step": 1095
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 2.310924369747899,
|
| 1766 |
+
"grad_norm": 0.2701464891433716,
|
| 1767 |
+
"learning_rate": 1.118984539836051e-05,
|
| 1768 |
+
"loss": 0.0521,
|
| 1769 |
+
"num_tokens": 13509170.0,
|
| 1770 |
+
"step": 1100
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.3214285714285716,
|
| 1774 |
+
"grad_norm": 0.18647603690624237,
|
| 1775 |
+
"learning_rate": 1.1011398955440702e-05,
|
| 1776 |
+
"loss": 0.0498,
|
| 1777 |
+
"num_tokens": 13570409.0,
|
| 1778 |
+
"step": 1105
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 2.331932773109244,
|
| 1782 |
+
"grad_norm": 0.12975195050239563,
|
| 1783 |
+
"learning_rate": 1.0835165094284264e-05,
|
| 1784 |
+
"loss": 0.0507,
|
| 1785 |
+
"num_tokens": 13631849.0,
|
| 1786 |
+
"step": 1110
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 2.342436974789916,
|
| 1790 |
+
"grad_norm": 0.15623484551906586,
|
| 1791 |
+
"learning_rate": 1.066116746346065e-05,
|
| 1792 |
+
"loss": 0.0499,
|
| 1793 |
+
"num_tokens": 13693289.0,
|
| 1794 |
+
"step": 1115
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.3529411764705883,
|
| 1798 |
+
"grad_norm": 0.1415032297372818,
|
| 1799 |
+
"learning_rate": 1.0489429411462794e-05,
|
| 1800 |
+
"loss": 0.05,
|
| 1801 |
+
"num_tokens": 13754729.0,
|
| 1802 |
+
"step": 1120
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 2.3634453781512605,
|
| 1806 |
+
"grad_norm": 0.188720241189003,
|
| 1807 |
+
"learning_rate": 1.0319973983573971e-05,
|
| 1808 |
+
"loss": 0.053,
|
| 1809 |
+
"num_tokens": 13816169.0,
|
| 1810 |
+
"step": 1125
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 2.3739495798319328,
|
| 1814 |
+
"grad_norm": 0.19719360768795013,
|
| 1815 |
+
"learning_rate": 1.0152823918775408e-05,
|
| 1816 |
+
"loss": 0.0503,
|
| 1817 |
+
"num_tokens": 13877609.0,
|
| 1818 |
+
"step": 1130
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 2.384453781512605,
|
| 1822 |
+
"grad_norm": 0.1669566035270691,
|
| 1823 |
+
"learning_rate": 9.988001646694935e-06,
|
| 1824 |
+
"loss": 0.0499,
|
| 1825 |
+
"num_tokens": 13939049.0,
|
| 1826 |
+
"step": 1135
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 2.3949579831932772,
|
| 1830 |
+
"grad_norm": 0.22400623559951782,
|
| 1831 |
+
"learning_rate": 9.825529284597238e-06,
|
| 1832 |
+
"loss": 0.0534,
|
| 1833 |
+
"num_tokens": 14000489.0,
|
| 1834 |
+
"step": 1140
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.4054621848739495,
|
| 1838 |
+
"grad_norm": 0.15708568692207336,
|
| 1839 |
+
"learning_rate": 9.665428634415923e-06,
|
| 1840 |
+
"loss": 0.0499,
|
| 1841 |
+
"num_tokens": 14061697.0,
|
| 1842 |
+
"step": 1145
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 2.4159663865546217,
|
| 1846 |
+
"grad_norm": 0.16055414080619812,
|
| 1847 |
+
"learning_rate": 9.50772117982799e-06,
|
| 1848 |
+
"loss": 0.0506,
|
| 1849 |
+
"num_tokens": 14123137.0,
|
| 1850 |
+
"step": 1150
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 2.426470588235294,
|
| 1854 |
+
"grad_norm": 0.14245997369289398,
|
| 1855 |
+
"learning_rate": 9.352428083370946e-06,
|
| 1856 |
+
"loss": 0.0497,
|
| 1857 |
+
"num_tokens": 14184577.0,
|
| 1858 |
+
"step": 1155
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 2.4369747899159666,
|
| 1862 |
+
"grad_norm": 0.14547857642173767,
|
| 1863 |
+
"learning_rate": 9.199570183603021e-06,
|
| 1864 |
+
"loss": 0.0501,
|
| 1865 |
+
"num_tokens": 14246017.0,
|
| 1866 |
+
"step": 1160
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 2.447478991596639,
|
| 1870 |
+
"grad_norm": 0.17205478250980377,
|
| 1871 |
+
"learning_rate": 9.049167992306908e-06,
|
| 1872 |
+
"loss": 0.0501,
|
| 1873 |
+
"num_tokens": 14307457.0,
|
| 1874 |
+
"step": 1165
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 2.457983193277311,
|
| 1878 |
+
"grad_norm": 0.16485774517059326,
|
| 1879 |
+
"learning_rate": 8.901241691737286e-06,
|
| 1880 |
+
"loss": 0.0499,
|
| 1881 |
+
"num_tokens": 14368897.0,
|
| 1882 |
+
"step": 1170
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 2.4684873949579833,
|
| 1886 |
+
"grad_norm": 0.1967056393623352,
|
| 1887 |
+
"learning_rate": 8.755811131912612e-06,
|
| 1888 |
+
"loss": 0.051,
|
| 1889 |
+
"num_tokens": 14430337.0,
|
| 1890 |
+
"step": 1175
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 2.4789915966386555,
|
| 1894 |
+
"grad_norm": 0.14919425547122955,
|
| 1895 |
+
"learning_rate": 8.612895827951451e-06,
|
| 1896 |
+
"loss": 0.0495,
|
| 1897 |
+
"num_tokens": 14491744.0,
|
| 1898 |
+
"step": 1180
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 2.4894957983193278,
|
| 1902 |
+
"grad_norm": 0.1448267251253128,
|
| 1903 |
+
"learning_rate": 8.472514957453801e-06,
|
| 1904 |
+
"loss": 0.0512,
|
| 1905 |
+
"num_tokens": 14553007.0,
|
| 1906 |
+
"step": 1185
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 2.5,
|
| 1910 |
+
"grad_norm": 0.17768217623233795,
|
| 1911 |
+
"learning_rate": 8.33468735792765e-06,
|
| 1912 |
+
"loss": 0.0501,
|
| 1913 |
+
"num_tokens": 14614447.0,
|
| 1914 |
+
"step": 1190
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 2.5105042016806722,
|
| 1918 |
+
"grad_norm": 0.1507992148399353,
|
| 1919 |
+
"learning_rate": 8.199431524261223e-06,
|
| 1920 |
+
"loss": 0.0503,
|
| 1921 |
+
"num_tokens": 14675727.0,
|
| 1922 |
+
"step": 1195
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 2.5210084033613445,
|
| 1926 |
+
"grad_norm": 0.16350729763507843,
|
| 1927 |
+
"learning_rate": 8.066765606241163e-06,
|
| 1928 |
+
"loss": 0.0496,
|
| 1929 |
+
"num_tokens": 14737165.0,
|
| 1930 |
+
"step": 1200
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 2.5315126050420167,
|
| 1934 |
+
"grad_norm": 0.16035616397857666,
|
| 1935 |
+
"learning_rate": 7.936707406117028e-06,
|
| 1936 |
+
"loss": 0.0488,
|
| 1937 |
+
"num_tokens": 14798605.0,
|
| 1938 |
+
"step": 1205
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 2.542016806722689,
|
| 1942 |
+
"grad_norm": 0.1894913911819458,
|
| 1943 |
+
"learning_rate": 7.809274376212464e-06,
|
| 1944 |
+
"loss": 0.0508,
|
| 1945 |
+
"num_tokens": 14859883.0,
|
| 1946 |
+
"step": 1210
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 2.552521008403361,
|
| 1950 |
+
"grad_norm": 0.1903340071439743,
|
| 1951 |
+
"learning_rate": 7.68448361658327e-06,
|
| 1952 |
+
"loss": 0.0488,
|
| 1953 |
+
"num_tokens": 14921105.0,
|
| 1954 |
+
"step": 1215
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 2.5630252100840334,
|
| 1958 |
+
"grad_norm": 0.14671629667282104,
|
| 1959 |
+
"learning_rate": 7.5623518727227975e-06,
|
| 1960 |
+
"loss": 0.0495,
|
| 1961 |
+
"num_tokens": 14982545.0,
|
| 1962 |
+
"step": 1220
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.5735294117647056,
|
| 1966 |
+
"grad_norm": 0.16081440448760986,
|
| 1967 |
+
"learning_rate": 7.442895533314856e-06,
|
| 1968 |
+
"loss": 0.0473,
|
| 1969 |
+
"num_tokens": 15043985.0,
|
| 1970 |
+
"step": 1225
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 2.5840336134453783,
|
| 1974 |
+
"grad_norm": 0.1555902659893036,
|
| 1975 |
+
"learning_rate": 7.326130628034581e-06,
|
| 1976 |
+
"loss": 0.0492,
|
| 1977 |
+
"num_tokens": 15105425.0,
|
| 1978 |
+
"step": 1230
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 2.5945378151260505,
|
| 1982 |
+
"grad_norm": 0.1796170324087143,
|
| 1983 |
+
"learning_rate": 7.212072825397413e-06,
|
| 1984 |
+
"loss": 0.0497,
|
| 1985 |
+
"num_tokens": 15166865.0,
|
| 1986 |
+
"step": 1235
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 2.6050420168067228,
|
| 1990 |
+
"grad_norm": 0.13445882499217987,
|
| 1991 |
+
"learning_rate": 7.100737430656561e-06,
|
| 1992 |
+
"loss": 0.0494,
|
| 1993 |
+
"num_tokens": 15228139.0,
|
| 1994 |
+
"step": 1240
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 2.615546218487395,
|
| 1998 |
+
"grad_norm": 0.18797667324543,
|
| 1999 |
+
"learning_rate": 6.992139383749224e-06,
|
| 2000 |
+
"loss": 0.0499,
|
| 2001 |
+
"num_tokens": 15289579.0,
|
| 2002 |
+
"step": 1245
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 2.6260504201680672,
|
| 2006 |
+
"grad_norm": 0.1478380262851715,
|
| 2007 |
+
"learning_rate": 6.886293257291801e-06,
|
| 2008 |
+
"loss": 0.0503,
|
| 2009 |
+
"num_tokens": 15351019.0,
|
| 2010 |
+
"step": 1250
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 2.6365546218487395,
|
| 2014 |
+
"grad_norm": 0.19320227205753326,
|
| 2015 |
+
"learning_rate": 6.78321325462444e-06,
|
| 2016 |
+
"loss": 0.0486,
|
| 2017 |
+
"num_tokens": 15412459.0,
|
| 2018 |
+
"step": 1255
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 2.6470588235294117,
|
| 2022 |
+
"grad_norm": 0.18944524228572845,
|
| 2023 |
+
"learning_rate": 6.682913207905095e-06,
|
| 2024 |
+
"loss": 0.0496,
|
| 2025 |
+
"num_tokens": 15473796.0,
|
| 2026 |
+
"step": 1260
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 2.657563025210084,
|
| 2030 |
+
"grad_norm": 0.17592737078666687,
|
| 2031 |
+
"learning_rate": 6.585406576253404e-06,
|
| 2032 |
+
"loss": 0.0501,
|
| 2033 |
+
"num_tokens": 15535236.0,
|
| 2034 |
+
"step": 1265
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 2.668067226890756,
|
| 2038 |
+
"grad_norm": 0.18211396038532257,
|
| 2039 |
+
"learning_rate": 6.490706443944656e-06,
|
| 2040 |
+
"loss": 0.0491,
|
| 2041 |
+
"num_tokens": 15596676.0,
|
| 2042 |
+
"step": 1270
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 2.678571428571429,
|
| 2046 |
+
"grad_norm": 0.1536846160888672,
|
| 2047 |
+
"learning_rate": 6.398825518653992e-06,
|
| 2048 |
+
"loss": 0.05,
|
| 2049 |
+
"num_tokens": 15658116.0,
|
| 2050 |
+
"step": 1275
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"epoch": 2.689075630252101,
|
| 2054 |
+
"grad_norm": 0.18677380681037903,
|
| 2055 |
+
"learning_rate": 6.30977612975121e-06,
|
| 2056 |
+
"loss": 0.0493,
|
| 2057 |
+
"num_tokens": 15719399.0,
|
| 2058 |
+
"step": 1280
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 2.6995798319327733,
|
| 2062 |
+
"grad_norm": 0.1490916907787323,
|
| 2063 |
+
"learning_rate": 6.223570226646291e-06,
|
| 2064 |
+
"loss": 0.0514,
|
| 2065 |
+
"num_tokens": 15780839.0,
|
| 2066 |
+
"step": 1285
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 2.7100840336134455,
|
| 2070 |
+
"grad_norm": 0.15238384902477264,
|
| 2071 |
+
"learning_rate": 6.140219377185933e-06,
|
| 2072 |
+
"loss": 0.05,
|
| 2073 |
+
"num_tokens": 15842274.0,
|
| 2074 |
+
"step": 1290
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 2.7205882352941178,
|
| 2078 |
+
"grad_norm": 0.15011648833751678,
|
| 2079 |
+
"learning_rate": 6.0597347661012635e-06,
|
| 2080 |
+
"loss": 0.0493,
|
| 2081 |
+
"num_tokens": 15903714.0,
|
| 2082 |
+
"step": 1295
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 2.73109243697479,
|
| 2086 |
+
"grad_norm": 0.1596149504184723,
|
| 2087 |
+
"learning_rate": 5.982127193507003e-06,
|
| 2088 |
+
"loss": 0.0494,
|
| 2089 |
+
"num_tokens": 15965148.0,
|
| 2090 |
+
"step": 1300
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 2.741596638655462,
|
| 2094 |
+
"grad_norm": 0.16487446427345276,
|
| 2095 |
+
"learning_rate": 5.907407073452186e-06,
|
| 2096 |
+
"loss": 0.0506,
|
| 2097 |
+
"num_tokens": 16026588.0,
|
| 2098 |
+
"step": 1305
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 2.7521008403361344,
|
| 2102 |
+
"grad_norm": 0.1454056352376938,
|
| 2103 |
+
"learning_rate": 5.835584432522727e-06,
|
| 2104 |
+
"loss": 0.0492,
|
| 2105 |
+
"num_tokens": 16088028.0,
|
| 2106 |
+
"step": 1310
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 2.7626050420168067,
|
| 2110 |
+
"grad_norm": 0.16204313933849335,
|
| 2111 |
+
"learning_rate": 5.766668908495966e-06,
|
| 2112 |
+
"loss": 0.0509,
|
| 2113 |
+
"num_tokens": 16149468.0,
|
| 2114 |
+
"step": 1315
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 2.773109243697479,
|
| 2118 |
+
"grad_norm": 0.18910805881023407,
|
| 2119 |
+
"learning_rate": 5.700669749047387e-06,
|
| 2120 |
+
"loss": 0.0489,
|
| 2121 |
+
"num_tokens": 16210908.0,
|
| 2122 |
+
"step": 1320
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 2.783613445378151,
|
| 2126 |
+
"grad_norm": 0.17493724822998047,
|
| 2127 |
+
"learning_rate": 5.637595810509689e-06,
|
| 2128 |
+
"loss": 0.05,
|
| 2129 |
+
"num_tokens": 16272348.0,
|
| 2130 |
+
"step": 1325
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 2.7941176470588234,
|
| 2134 |
+
"grad_norm": 0.14875848591327667,
|
| 2135 |
+
"learning_rate": 5.577455556684369e-06,
|
| 2136 |
+
"loss": 0.049,
|
| 2137 |
+
"num_tokens": 16333788.0,
|
| 2138 |
+
"step": 1330
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 2.8046218487394956,
|
| 2142 |
+
"grad_norm": 0.1500275731086731,
|
| 2143 |
+
"learning_rate": 5.520257057705971e-06,
|
| 2144 |
+
"loss": 0.0498,
|
| 2145 |
+
"num_tokens": 16395228.0,
|
| 2146 |
+
"step": 1335
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 2.815126050420168,
|
| 2150 |
+
"grad_norm": 0.1598060131072998,
|
| 2151 |
+
"learning_rate": 5.466007988959163e-06,
|
| 2152 |
+
"loss": 0.0507,
|
| 2153 |
+
"num_tokens": 16456417.0,
|
| 2154 |
+
"step": 1340
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 2.82563025210084,
|
| 2158 |
+
"grad_norm": 0.1572778970003128,
|
| 2159 |
+
"learning_rate": 5.414715630048797e-06,
|
| 2160 |
+
"loss": 0.051,
|
| 2161 |
+
"num_tokens": 16517857.0,
|
| 2162 |
+
"step": 1345
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 2.8361344537815127,
|
| 2166 |
+
"grad_norm": 0.15581144392490387,
|
| 2167 |
+
"learning_rate": 5.366386863823077e-06,
|
| 2168 |
+
"loss": 0.0499,
|
| 2169 |
+
"num_tokens": 16579297.0,
|
| 2170 |
+
"step": 1350
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 2.846638655462185,
|
| 2174 |
+
"grad_norm": 0.18151573836803436,
|
| 2175 |
+
"learning_rate": 5.3210281754499284e-06,
|
| 2176 |
+
"loss": 0.0496,
|
| 2177 |
+
"num_tokens": 16640737.0,
|
| 2178 |
+
"step": 1355
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 2.857142857142857,
|
| 2182 |
+
"grad_norm": 0.1601044237613678,
|
| 2183 |
+
"learning_rate": 5.278645651546797e-06,
|
| 2184 |
+
"loss": 0.0487,
|
| 2185 |
+
"num_tokens": 16702177.0,
|
| 2186 |
+
"step": 1360
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 2.8676470588235294,
|
| 2190 |
+
"grad_norm": 0.1497681736946106,
|
| 2191 |
+
"learning_rate": 5.239244979363877e-06,
|
| 2192 |
+
"loss": 0.0492,
|
| 2193 |
+
"num_tokens": 16763617.0,
|
| 2194 |
+
"step": 1365
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 2.8781512605042017,
|
| 2198 |
+
"grad_norm": 0.15907803177833557,
|
| 2199 |
+
"learning_rate": 5.202831446020945e-06,
|
| 2200 |
+
"loss": 0.0502,
|
| 2201 |
+
"num_tokens": 16824905.0,
|
| 2202 |
+
"step": 1370
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 2.888655462184874,
|
| 2206 |
+
"grad_norm": 0.17641125619411469,
|
| 2207 |
+
"learning_rate": 5.169409937797901e-06,
|
| 2208 |
+
"loss": 0.0502,
|
| 2209 |
+
"num_tokens": 16886146.0,
|
| 2210 |
+
"step": 1375
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 2.899159663865546,
|
| 2214 |
+
"grad_norm": 0.12964367866516113,
|
| 2215 |
+
"learning_rate": 5.138984939479077e-06,
|
| 2216 |
+
"loss": 0.0487,
|
| 2217 |
+
"num_tokens": 16947586.0,
|
| 2218 |
+
"step": 1380
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 2.9096638655462184,
|
| 2222 |
+
"grad_norm": 0.14473247528076172,
|
| 2223 |
+
"learning_rate": 5.111560533751426e-06,
|
| 2224 |
+
"loss": 0.0491,
|
| 2225 |
+
"num_tokens": 17009026.0,
|
| 2226 |
+
"step": 1385
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 2.9201680672268906,
|
| 2230 |
+
"grad_norm": 0.18652838468551636,
|
| 2231 |
+
"learning_rate": 5.087140400656684e-06,
|
| 2232 |
+
"loss": 0.0506,
|
| 2233 |
+
"num_tokens": 17070466.0,
|
| 2234 |
+
"step": 1390
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 2.9306722689075633,
|
| 2238 |
+
"grad_norm": 0.18603888154029846,
|
| 2239 |
+
"learning_rate": 5.065727817097544e-06,
|
| 2240 |
+
"loss": 0.0492,
|
| 2241 |
+
"num_tokens": 17131779.0,
|
| 2242 |
+
"step": 1395
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 2.9411764705882355,
|
| 2246 |
+
"grad_norm": 0.15723615884780884,
|
| 2247 |
+
"learning_rate": 5.047325656397932e-06,
|
| 2248 |
+
"loss": 0.0494,
|
| 2249 |
+
"num_tokens": 17193063.0,
|
| 2250 |
+
"step": 1400
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 2.9516806722689077,
|
| 2254 |
+
"grad_norm": 0.14798587560653687,
|
| 2255 |
+
"learning_rate": 5.031936387917442e-06,
|
| 2256 |
+
"loss": 0.049,
|
| 2257 |
+
"num_tokens": 17254503.0,
|
| 2258 |
+
"step": 1405
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 2.96218487394958,
|
| 2262 |
+
"grad_norm": 0.19435246288776398,
|
| 2263 |
+
"learning_rate": 5.019562076719972e-06,
|
| 2264 |
+
"loss": 0.0494,
|
| 2265 |
+
"num_tokens": 17315742.0,
|
| 2266 |
+
"step": 1410
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 2.972689075630252,
|
| 2270 |
+
"grad_norm": 0.17056235671043396,
|
| 2271 |
+
"learning_rate": 5.0102043832966236e-06,
|
| 2272 |
+
"loss": 0.0493,
|
| 2273 |
+
"num_tokens": 17377182.0,
|
| 2274 |
+
"step": 1415
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 2.9831932773109244,
|
| 2278 |
+
"grad_norm": 0.12487131357192993,
|
| 2279 |
+
"learning_rate": 5.003864563342878e-06,
|
| 2280 |
+
"loss": 0.0477,
|
| 2281 |
+
"num_tokens": 17438622.0,
|
| 2282 |
+
"step": 1420
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 2.9936974789915967,
|
| 2286 |
+
"grad_norm": 0.13770359754562378,
|
| 2287 |
+
"learning_rate": 5.0005434675900966e-06,
|
| 2288 |
+
"loss": 0.0477,
|
| 2289 |
+
"num_tokens": 17500062.0,
|
| 2290 |
+
"step": 1425
|
| 2291 |
+
},
|
| 2292 |
{
|
| 2293 |
"epoch": 3.0,
|
| 2294 |
+
"num_tokens": 17536926.0,
|
| 2295 |
+
"step": 1428,
|
| 2296 |
+
"total_flos": 7.444201440207176e+17,
|
| 2297 |
+
"train_loss": 0.06586769079210378,
|
| 2298 |
+
"train_runtime": 7485.0094,
|
| 2299 |
+
"train_samples_per_second": 9.147,
|
| 2300 |
+
"train_steps_per_second": 0.191
|
| 2301 |
}
|
| 2302 |
],
|
| 2303 |
"logging_steps": 5,
|
| 2304 |
+
"max_steps": 1428,
|
| 2305 |
"num_input_tokens_seen": 0,
|
| 2306 |
"num_train_epochs": 3,
|
| 2307 |
"save_steps": 100,
|
|
|
|
| 2317 |
"attributes": {}
|
| 2318 |
}
|
| 2319 |
},
|
| 2320 |
+
"total_flos": 7.444201440207176e+17,
|
| 2321 |
"train_batch_size": 16,
|
| 2322 |
"trial_name": null,
|
| 2323 |
"trial_params": null
|