Model save
Browse files- README.md +1 -1
- all_results.json +3 -3
- model-00001-of-00004.safetensors +1 -1
- model-00002-of-00004.safetensors +1 -1
- model-00003-of-00004.safetensors +1 -1
- model-00004-of-00004.safetensors +1 -1
- train_results.json +3 -3
- trainer_state.json +876 -510
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -27,7 +27,7 @@ print(output["generated_text"])
|
|
| 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/jun-liang-chen-the-hong-kong-polytechnic-university/huggingface/runs/
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
|
|
|
| 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/jun-liang-chen-the-hong-kong-polytechnic-university/huggingface/runs/bomcw3b5)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 7500,
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.2665646580783036,
|
| 4 |
+
"train_runtime": 484901.6937,
|
| 5 |
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.015,
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
model-00001-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4877660776
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29eaeace24738554f7e6e05e2ed92758752732e7e4cd1e62af9519cbed06d2cc
|
| 3 |
size 4877660776
|
model-00002-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4932751008
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d00d4d8b88b12a1c5903e1bd82c1eb6b5a34d3091ce4a1d1a703b8d9276fcf26
|
| 3 |
size 4932751008
|
model-00003-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4330865200
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:736625042166c57ba72783a3df20e860c03a83719e89257119c8b83ec51ccfad
|
| 3 |
size 4330865200
|
model-00004-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1089994880
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18ffebf5a0285c8b31fd711c517f01fb1dc3ee9de44769b1c9053dacafea52b9
|
| 3 |
size 1089994880
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
-
"train_loss":
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 7500,
|
| 6 |
-
"train_samples_per_second": 0.
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.2665646580783036,
|
| 4 |
+
"train_runtime": 484901.6937,
|
| 5 |
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.015,
|
| 7 |
"train_steps_per_second": 0.001
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,796 +1,1162 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"eval_steps":
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"clip_ratio": 0.0,
|
| 13 |
-
"completion_length":
|
| 14 |
-
"epoch": 0.
|
| 15 |
-
"grad_norm": 0.
|
| 16 |
-
"kl":
|
| 17 |
-
"learning_rate":
|
| 18 |
-
"loss":
|
| 19 |
-
"reward": 0.
|
| 20 |
-
"reward_std": 0.
|
| 21 |
-
"rewards/accuracy_reward": 0.
|
| 22 |
"rewards/format_reward": 0.0,
|
| 23 |
"step": 1
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"clip_ratio": 0.0,
|
| 27 |
-
"completion_length":
|
| 28 |
-
"epoch": 0.
|
| 29 |
-
"grad_norm": 0.
|
| 30 |
-
"kl": 0.
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss": -0.
|
| 33 |
-
"reward": 0.
|
| 34 |
-
"reward_std": 0.
|
| 35 |
-
"rewards/accuracy_reward": 0.
|
| 36 |
"rewards/format_reward": 0.0,
|
| 37 |
"step": 5
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"clip_ratio": 0.0,
|
| 41 |
-
"completion_length":
|
| 42 |
-
"epoch": 0.
|
| 43 |
-
"grad_norm": 0.
|
| 44 |
-
"kl": 0.
|
| 45 |
-
"learning_rate":
|
| 46 |
-
"loss": -0.
|
| 47 |
-
"reward": 0.
|
| 48 |
-
"reward_std": 0.
|
| 49 |
-
"rewards/accuracy_reward": 0.
|
| 50 |
"rewards/format_reward": 0.0,
|
| 51 |
"step": 10
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"clip_ratio": 0.0,
|
| 55 |
-
"completion_length":
|
| 56 |
-
"epoch": 0.
|
| 57 |
-
"grad_norm": 0.
|
| 58 |
-
"kl": 0.
|
| 59 |
-
"learning_rate": 1.
|
| 60 |
-
"loss": -0.
|
| 61 |
-
"reward": 0.
|
| 62 |
-
"reward_std": 0.
|
| 63 |
-
"rewards/accuracy_reward": 0.
|
| 64 |
"rewards/format_reward": 0.0,
|
| 65 |
"step": 15
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"clip_ratio": 0.0,
|
| 69 |
-
"completion_length":
|
| 70 |
-
"epoch": 0.
|
| 71 |
-
"grad_norm":
|
| 72 |
-
"kl": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": -0.
|
| 75 |
-
"reward": 0.
|
| 76 |
-
"reward_std": 0.
|
| 77 |
-
"rewards/accuracy_reward": 0.
|
| 78 |
"rewards/format_reward": 0.0,
|
| 79 |
"step": 20
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"clip_ratio": 0.0,
|
| 83 |
-
"completion_length":
|
| 84 |
-
"epoch": 0.
|
| 85 |
-
"grad_norm": 0.
|
| 86 |
-
"kl": 0.
|
| 87 |
-
"learning_rate":
|
| 88 |
-
"loss": -0.
|
| 89 |
-
"reward": 0.
|
| 90 |
-
"reward_std": 0.
|
| 91 |
-
"rewards/accuracy_reward": 0.
|
| 92 |
"rewards/format_reward": 0.0,
|
| 93 |
"step": 25
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"clip_ratio": 0.0,
|
| 97 |
-
"completion_length":
|
| 98 |
-
"epoch": 0.
|
| 99 |
-
"grad_norm": 0.
|
| 100 |
-
"kl": 0.
|
| 101 |
-
"learning_rate": 2.
|
| 102 |
-
"loss": -0.
|
| 103 |
-
"reward": 0.
|
| 104 |
-
"reward_std": 0.
|
| 105 |
-
"rewards/accuracy_reward": 0.
|
| 106 |
"rewards/format_reward": 0.0,
|
| 107 |
"step": 30
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"clip_ratio": 0.0,
|
| 111 |
-
"completion_length":
|
| 112 |
-
"epoch": 0.
|
| 113 |
-
"grad_norm": 0.
|
| 114 |
-
"kl":
|
| 115 |
-
"learning_rate": 2.
|
| 116 |
-
"loss": 0.
|
| 117 |
-
"reward": 0.
|
| 118 |
-
"reward_std": 0.
|
| 119 |
-
"rewards/accuracy_reward": 0.
|
| 120 |
"rewards/format_reward": 0.0,
|
| 121 |
"step": 35
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"clip_ratio": 0.0,
|
| 125 |
-
"completion_length":
|
| 126 |
-
"epoch": 0.
|
| 127 |
-
"grad_norm":
|
| 128 |
-
"kl": 0.
|
| 129 |
-
"learning_rate":
|
| 130 |
-
"loss": -0.
|
| 131 |
-
"reward": 0.
|
| 132 |
-
"reward_std": 0.
|
| 133 |
-
"rewards/accuracy_reward": 0.
|
| 134 |
-
"rewards/format_reward": 0.
|
| 135 |
"step": 40
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"clip_ratio": 0.0,
|
| 139 |
-
"completion_length":
|
| 140 |
-
"epoch": 0.
|
| 141 |
-
"grad_norm": 0.
|
| 142 |
-
"kl": 0.
|
| 143 |
-
"learning_rate": 2.
|
| 144 |
-
"loss": 0.
|
| 145 |
-
"reward": 0.
|
| 146 |
-
"reward_std": 0.
|
| 147 |
-
"rewards/accuracy_reward": 0.
|
| 148 |
"rewards/format_reward": 0.0,
|
| 149 |
"step": 45
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"clip_ratio": 0.0,
|
| 153 |
-
"completion_length":
|
| 154 |
-
"epoch": 0.
|
| 155 |
-
"grad_norm": 0.
|
| 156 |
-
"kl": 0.
|
| 157 |
-
"learning_rate": 2.
|
| 158 |
-
"loss": 0.
|
| 159 |
-
"reward": 0.
|
| 160 |
-
"reward_std": 0.
|
| 161 |
-
"rewards/accuracy_reward": 0.
|
| 162 |
"rewards/format_reward": 0.0,
|
| 163 |
"step": 50
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"clip_ratio": 0.0,
|
| 167 |
-
"completion_length":
|
| 168 |
-
"epoch": 0.
|
| 169 |
-
"grad_norm":
|
| 170 |
-
"kl": 0.
|
| 171 |
-
"learning_rate": 2.
|
| 172 |
-
"loss": 0.
|
| 173 |
-
"reward": 0.
|
| 174 |
-
"reward_std": 0.
|
| 175 |
-
"rewards/accuracy_reward": 0.
|
| 176 |
"rewards/format_reward": 0.0,
|
| 177 |
"step": 55
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"clip_ratio": 0.0,
|
| 181 |
-
"completion_length":
|
| 182 |
-
"epoch": 0.
|
| 183 |
-
"grad_norm": 0.
|
| 184 |
-
"kl":
|
| 185 |
-
"learning_rate": 2.
|
| 186 |
-
"loss":
|
| 187 |
-
"reward": 0.
|
| 188 |
-
"reward_std": 0.
|
| 189 |
-
"rewards/accuracy_reward": 0.
|
| 190 |
"rewards/format_reward": 0.0,
|
| 191 |
"step": 60
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"clip_ratio": 0.0,
|
| 195 |
-
"completion_length":
|
| 196 |
-
"epoch": 0.
|
| 197 |
-
"grad_norm":
|
| 198 |
-
"kl": 0.
|
| 199 |
-
"learning_rate": 2.
|
| 200 |
-
"loss":
|
| 201 |
-
"reward": 0.
|
| 202 |
-
"reward_std": 0.
|
| 203 |
-
"rewards/accuracy_reward": 0.
|
| 204 |
"rewards/format_reward": 0.0,
|
| 205 |
"step": 65
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"clip_ratio": 0.0,
|
| 209 |
-
"completion_length":
|
| 210 |
-
"epoch": 0.
|
| 211 |
-
"grad_norm":
|
| 212 |
-
"kl":
|
| 213 |
-
"learning_rate": 2.
|
| 214 |
-
"loss":
|
| 215 |
-
"reward": 0.
|
| 216 |
-
"reward_std": 0.
|
| 217 |
-
"rewards/accuracy_reward": 0.
|
| 218 |
"rewards/format_reward": 0.0,
|
| 219 |
"step": 70
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"clip_ratio": 0.0,
|
| 223 |
-
"completion_length":
|
| 224 |
-
"epoch": 0.
|
| 225 |
-
"grad_norm": 0.
|
| 226 |
-
"kl": 0.
|
| 227 |
-
"learning_rate": 2.
|
| 228 |
-
"loss": -0.
|
| 229 |
-
"reward": 0.
|
| 230 |
-
"reward_std": 0.
|
| 231 |
-
"rewards/accuracy_reward": 0.
|
| 232 |
"rewards/format_reward": 0.0,
|
| 233 |
"step": 75
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"clip_ratio": 0.0,
|
| 237 |
-
"completion_length":
|
| 238 |
-
"epoch": 0.
|
| 239 |
-
"grad_norm": 0.
|
| 240 |
-
"kl": 0.
|
| 241 |
-
"learning_rate": 2.
|
| 242 |
-
"loss": -0.
|
| 243 |
-
"reward": 0.
|
| 244 |
-
"reward_std": 0.
|
| 245 |
-
"rewards/accuracy_reward": 0.
|
| 246 |
"rewards/format_reward": 0.0,
|
| 247 |
"step": 80
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"clip_ratio": 0.0,
|
| 251 |
-
"completion_length":
|
| 252 |
-
"epoch": 0.
|
| 253 |
-
"grad_norm": 0.
|
| 254 |
-
"kl": 0.
|
| 255 |
-
"learning_rate": 2.
|
| 256 |
-
"loss": -0.
|
| 257 |
-
"reward": 0.
|
| 258 |
-
"reward_std": 0.
|
| 259 |
-
"rewards/accuracy_reward": 0.
|
| 260 |
"rewards/format_reward": 0.0,
|
| 261 |
"step": 85
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"clip_ratio": 0.0,
|
| 265 |
-
"completion_length":
|
| 266 |
-
"epoch": 0.
|
| 267 |
-
"grad_norm":
|
| 268 |
-
"kl": 0.
|
| 269 |
-
"learning_rate": 2.
|
| 270 |
-
"loss": 0.
|
| 271 |
-
"reward": 0.
|
| 272 |
-
"reward_std": 0.
|
| 273 |
-
"rewards/accuracy_reward": 0.
|
| 274 |
"rewards/format_reward": 0.0,
|
| 275 |
"step": 90
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"clip_ratio": 0.0,
|
| 279 |
-
"completion_length":
|
| 280 |
-
"epoch": 0.
|
| 281 |
-
"grad_norm": 0.
|
| 282 |
-
"kl": 0.
|
| 283 |
-
"learning_rate": 2.
|
| 284 |
-
"loss": -0.
|
| 285 |
-
"reward": 0.
|
| 286 |
-
"reward_std": 0.
|
| 287 |
-
"rewards/accuracy_reward": 0.
|
| 288 |
"rewards/format_reward": 0.0,
|
| 289 |
"step": 95
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"
|
| 293 |
-
"
|
| 294 |
-
"
|
| 295 |
-
"
|
| 296 |
-
"
|
| 297 |
-
|
| 298 |
-
|
| 299 |
-
"
|
| 300 |
-
"
|
| 301 |
-
"
|
| 302 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 303 |
"step": 100
|
| 304 |
},
|
| 305 |
{
|
| 306 |
"clip_ratio": 0.0,
|
| 307 |
-
"completion_length":
|
| 308 |
-
"epoch": 0.
|
| 309 |
-
"grad_norm":
|
| 310 |
-
"kl": 0.
|
| 311 |
-
"learning_rate": 2.
|
| 312 |
-
"loss": -0.
|
| 313 |
-
"reward": 0.
|
| 314 |
-
"reward_std": 0.
|
| 315 |
-
"rewards/accuracy_reward": 0.
|
| 316 |
"rewards/format_reward": 0.0,
|
| 317 |
"step": 105
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"clip_ratio": 0.0,
|
| 321 |
-
"completion_length":
|
| 322 |
-
"epoch": 0.
|
| 323 |
-
"grad_norm":
|
| 324 |
-
"kl":
|
| 325 |
-
"learning_rate": 2.
|
| 326 |
-
"loss": -0.
|
| 327 |
-
"reward": 0.
|
| 328 |
-
"reward_std": 0.
|
| 329 |
-
"rewards/accuracy_reward": 0.
|
| 330 |
"rewards/format_reward": 0.0,
|
| 331 |
"step": 110
|
| 332 |
},
|
| 333 |
{
|
| 334 |
"clip_ratio": 0.0,
|
| 335 |
-
"completion_length":
|
| 336 |
-
"epoch": 0.
|
| 337 |
-
"grad_norm": 0.
|
| 338 |
-
"kl": 0.
|
| 339 |
-
"learning_rate": 2.
|
| 340 |
-
"loss":
|
| 341 |
-
"reward": 0.
|
| 342 |
-
"reward_std": 0.
|
| 343 |
-
"rewards/accuracy_reward": 0.
|
| 344 |
"rewards/format_reward": 0.0,
|
| 345 |
"step": 115
|
| 346 |
},
|
| 347 |
{
|
| 348 |
"clip_ratio": 0.0,
|
| 349 |
-
"completion_length":
|
| 350 |
-
"epoch": 0.
|
| 351 |
-
"grad_norm":
|
| 352 |
-
"kl": 0.
|
| 353 |
-
"learning_rate": 2.
|
| 354 |
-
"loss": -0.
|
| 355 |
-
"reward": 0.
|
| 356 |
-
"reward_std": 0.
|
| 357 |
-
"rewards/accuracy_reward": 0.
|
| 358 |
"rewards/format_reward": 0.0,
|
| 359 |
"step": 120
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"clip_ratio": 0.0,
|
| 363 |
-
"completion_length":
|
| 364 |
-
"epoch": 0.
|
| 365 |
-
"grad_norm":
|
| 366 |
-
"kl": 0.
|
| 367 |
-
"learning_rate":
|
| 368 |
-
"loss": -0.
|
| 369 |
-
"reward": 0.
|
| 370 |
-
"reward_std": 0.
|
| 371 |
-
"rewards/accuracy_reward": 0.
|
| 372 |
"rewards/format_reward": 0.0,
|
| 373 |
"step": 125
|
| 374 |
},
|
| 375 |
{
|
| 376 |
"clip_ratio": 0.0,
|
| 377 |
-
"completion_length":
|
| 378 |
-
"epoch": 0.
|
| 379 |
-
"grad_norm":
|
| 380 |
-
"kl": 0.
|
| 381 |
-
"learning_rate":
|
| 382 |
-
"loss":
|
| 383 |
-
"reward": 0.
|
| 384 |
-
"reward_std": 0.
|
| 385 |
-
"rewards/accuracy_reward": 0.
|
| 386 |
"rewards/format_reward": 0.0,
|
| 387 |
"step": 130
|
| 388 |
},
|
| 389 |
{
|
| 390 |
"clip_ratio": 0.0,
|
| 391 |
-
"completion_length":
|
| 392 |
-
"epoch": 0.
|
| 393 |
-
"grad_norm":
|
| 394 |
-
"kl":
|
| 395 |
-
"learning_rate":
|
| 396 |
-
"loss":
|
| 397 |
-
"reward": 0.
|
| 398 |
-
"reward_std": 0.
|
| 399 |
-
"rewards/accuracy_reward": 0.
|
| 400 |
"rewards/format_reward": 0.0,
|
| 401 |
"step": 135
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"clip_ratio": 0.0,
|
| 405 |
-
"completion_length":
|
| 406 |
-
"epoch": 0.
|
| 407 |
-
"grad_norm":
|
| 408 |
-
"kl": 0.
|
| 409 |
-
"learning_rate":
|
| 410 |
-
"loss": -0.
|
| 411 |
-
"reward": 0.
|
| 412 |
-
"reward_std": 0.
|
| 413 |
-
"rewards/accuracy_reward": 0.
|
| 414 |
"rewards/format_reward": 0.0,
|
| 415 |
"step": 140
|
| 416 |
},
|
| 417 |
{
|
| 418 |
"clip_ratio": 0.0,
|
| 419 |
-
"completion_length":
|
| 420 |
-
"epoch": 0.
|
| 421 |
-
"grad_norm":
|
| 422 |
-
"kl":
|
| 423 |
-
"learning_rate":
|
| 424 |
-
"loss":
|
| 425 |
-
"reward": 0.
|
| 426 |
-
"reward_std": 0.
|
| 427 |
-
"rewards/accuracy_reward": 0.
|
| 428 |
"rewards/format_reward": 0.0,
|
| 429 |
"step": 145
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"clip_ratio": 0.0,
|
| 433 |
-
"completion_length": 596.
|
| 434 |
-
"epoch": 0.
|
| 435 |
-
"grad_norm":
|
| 436 |
-
"kl":
|
| 437 |
-
"learning_rate":
|
| 438 |
-
"loss":
|
| 439 |
-
"reward": 0.
|
| 440 |
-
"reward_std": 0.
|
| 441 |
-
"rewards/accuracy_reward": 0.
|
| 442 |
"rewards/format_reward": 0.0,
|
| 443 |
"step": 150
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"clip_ratio": 0.0,
|
| 447 |
-
"completion_length": 579.
|
| 448 |
-
"epoch": 0.
|
| 449 |
-
"grad_norm":
|
| 450 |
-
"kl":
|
| 451 |
-
"learning_rate":
|
| 452 |
-
"loss":
|
| 453 |
-
"reward": 0.
|
| 454 |
-
"reward_std": 0.
|
| 455 |
-
"rewards/accuracy_reward": 0.
|
| 456 |
"rewards/format_reward": 0.0,
|
| 457 |
"step": 155
|
| 458 |
},
|
| 459 |
{
|
| 460 |
"clip_ratio": 0.0,
|
| 461 |
-
"completion_length":
|
| 462 |
-
"epoch": 0.
|
| 463 |
-
"grad_norm":
|
| 464 |
-
"kl":
|
| 465 |
-
"learning_rate":
|
| 466 |
-
"loss":
|
| 467 |
-
"reward": 0.
|
| 468 |
-
"reward_std": 0.
|
| 469 |
-
"rewards/accuracy_reward": 0.
|
| 470 |
"rewards/format_reward": 0.0,
|
| 471 |
"step": 160
|
| 472 |
},
|
| 473 |
{
|
| 474 |
"clip_ratio": 0.0,
|
| 475 |
-
"completion_length":
|
| 476 |
-
"epoch": 0.
|
| 477 |
-
"grad_norm":
|
| 478 |
-
"kl":
|
| 479 |
-
"learning_rate":
|
| 480 |
-
"loss":
|
| 481 |
-
"reward": 0.
|
| 482 |
-
"reward_std": 0.
|
| 483 |
-
"rewards/accuracy_reward": 0.
|
| 484 |
"rewards/format_reward": 0.0,
|
| 485 |
"step": 165
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"clip_ratio": 0.0,
|
| 489 |
-
"completion_length":
|
| 490 |
-
"epoch": 0.
|
| 491 |
-
"grad_norm":
|
| 492 |
-
"kl":
|
| 493 |
-
"learning_rate":
|
| 494 |
-
"loss":
|
| 495 |
-
"reward": 0.
|
| 496 |
-
"reward_std": 0.
|
| 497 |
-
"rewards/accuracy_reward": 0.
|
| 498 |
"rewards/format_reward": 0.0,
|
| 499 |
"step": 170
|
| 500 |
},
|
| 501 |
{
|
| 502 |
"clip_ratio": 0.0,
|
| 503 |
-
"completion_length":
|
| 504 |
-
"epoch": 0.
|
| 505 |
-
"grad_norm":
|
| 506 |
-
"kl":
|
| 507 |
-
"learning_rate":
|
| 508 |
-
"loss": 0.
|
| 509 |
-
"reward": 0.
|
| 510 |
-
"reward_std": 0.
|
| 511 |
-
"rewards/accuracy_reward": 0.
|
| 512 |
"rewards/format_reward": 0.0,
|
| 513 |
"step": 175
|
| 514 |
},
|
| 515 |
{
|
| 516 |
"clip_ratio": 0.0,
|
| 517 |
-
"completion_length":
|
| 518 |
-
"epoch": 0.
|
| 519 |
-
"grad_norm":
|
| 520 |
-
"kl":
|
| 521 |
-
"learning_rate":
|
| 522 |
-
"loss": 0.
|
| 523 |
-
"reward": 0.
|
| 524 |
-
"reward_std": 0.
|
| 525 |
-
"rewards/accuracy_reward": 0.
|
| 526 |
"rewards/format_reward": 0.0,
|
| 527 |
"step": 180
|
| 528 |
},
|
| 529 |
{
|
| 530 |
"clip_ratio": 0.0,
|
| 531 |
-
"completion_length":
|
| 532 |
-
"epoch": 0.
|
| 533 |
-
"grad_norm":
|
| 534 |
-
"kl":
|
| 535 |
-
"learning_rate":
|
| 536 |
-
"loss":
|
| 537 |
-
"reward": 0.
|
| 538 |
-
"reward_std": 0.
|
| 539 |
-
"rewards/accuracy_reward": 0.
|
| 540 |
"rewards/format_reward": 0.0,
|
| 541 |
"step": 185
|
| 542 |
},
|
| 543 |
{
|
| 544 |
"clip_ratio": 0.0,
|
| 545 |
-
"completion_length":
|
| 546 |
-
"epoch": 0.
|
| 547 |
-
"grad_norm":
|
| 548 |
-
"kl":
|
| 549 |
-
"learning_rate":
|
| 550 |
-
"loss":
|
| 551 |
-
"reward": 0.
|
| 552 |
-
"reward_std": 0.
|
| 553 |
-
"rewards/accuracy_reward": 0.
|
| 554 |
"rewards/format_reward": 0.0,
|
| 555 |
"step": 190
|
| 556 |
},
|
| 557 |
{
|
| 558 |
"clip_ratio": 0.0,
|
| 559 |
-
"completion_length":
|
| 560 |
-
"epoch": 0.
|
| 561 |
-
"grad_norm": 0.
|
| 562 |
-
"kl":
|
| 563 |
-
"learning_rate":
|
| 564 |
-
"loss":
|
| 565 |
-
"reward": 0.
|
| 566 |
-
"reward_std": 0.
|
| 567 |
-
"rewards/accuracy_reward": 0.
|
| 568 |
"rewards/format_reward": 0.0,
|
| 569 |
"step": 195
|
| 570 |
},
|
| 571 |
{
|
| 572 |
-
"epoch": 0.
|
| 573 |
-
"grad_norm": 0.
|
| 574 |
-
"learning_rate":
|
| 575 |
-
"loss": 0.
|
| 576 |
"step": 200
|
| 577 |
},
|
| 578 |
{
|
| 579 |
-
"epoch": 0.
|
| 580 |
"eval_clip_ratio": 0.0,
|
| 581 |
-
"eval_completion_length": 581.
|
| 582 |
-
"eval_kl":
|
| 583 |
-
"eval_loss":
|
| 584 |
-
"eval_reward": 0.
|
| 585 |
-
"eval_reward_std": 0.
|
| 586 |
-
"eval_rewards/accuracy_reward": 0.
|
| 587 |
-
"eval_rewards/format_reward":
|
| 588 |
-
"eval_runtime":
|
| 589 |
-
"eval_samples_per_second": 0.
|
| 590 |
-
"eval_steps_per_second": 0.
|
| 591 |
"step": 200
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"clip_ratio": 0.0,
|
| 595 |
-
"completion_length":
|
| 596 |
-
"epoch": 0.
|
| 597 |
-
"grad_norm": 0.
|
| 598 |
-
"kl":
|
| 599 |
-
"learning_rate":
|
| 600 |
-
"loss": 0.
|
| 601 |
-
"reward": 0.
|
| 602 |
-
"reward_std": 0.
|
| 603 |
-
"rewards/accuracy_reward": 0.
|
| 604 |
"rewards/format_reward": 0.0,
|
| 605 |
"step": 205
|
| 606 |
},
|
| 607 |
{
|
| 608 |
"clip_ratio": 0.0,
|
| 609 |
-
"completion_length":
|
| 610 |
-
"epoch": 0.
|
| 611 |
-
"grad_norm": 1.
|
| 612 |
-
"kl": 0.
|
| 613 |
-
"learning_rate":
|
| 614 |
-
"loss": 0.
|
| 615 |
-
"reward": 0.
|
| 616 |
-
"reward_std": 0.
|
| 617 |
-
"rewards/accuracy_reward": 0.
|
| 618 |
-
"rewards/format_reward": 0.
|
| 619 |
"step": 210
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"clip_ratio": 0.0,
|
| 623 |
-
"completion_length":
|
| 624 |
-
"epoch": 0.
|
| 625 |
-
"grad_norm":
|
| 626 |
-
"kl":
|
| 627 |
-
"learning_rate":
|
| 628 |
-
"loss": 0.
|
| 629 |
-
"reward": 0.
|
| 630 |
-
"reward_std": 0.
|
| 631 |
-
"rewards/accuracy_reward": 0.
|
| 632 |
"rewards/format_reward": 0.0,
|
| 633 |
"step": 215
|
| 634 |
},
|
| 635 |
{
|
| 636 |
"clip_ratio": 0.0,
|
| 637 |
-
"completion_length":
|
| 638 |
-
"epoch": 0.
|
| 639 |
-
"grad_norm":
|
| 640 |
-
"kl": 0.
|
| 641 |
-
"learning_rate":
|
| 642 |
-
"loss": -0.
|
| 643 |
-
"reward": 0.
|
| 644 |
-
"reward_std": 0.
|
| 645 |
-
"rewards/accuracy_reward": 0.
|
| 646 |
"rewards/format_reward": 0.0,
|
| 647 |
"step": 220
|
| 648 |
},
|
| 649 |
{
|
| 650 |
"clip_ratio": 0.0,
|
| 651 |
-
"completion_length":
|
| 652 |
-
"epoch": 0.
|
| 653 |
-
"grad_norm":
|
| 654 |
-
"kl":
|
| 655 |
-
"learning_rate":
|
| 656 |
-
"loss": 0.
|
| 657 |
-
"reward": 0.
|
| 658 |
-
"reward_std": 0.
|
| 659 |
-
"rewards/accuracy_reward": 0.
|
| 660 |
"rewards/format_reward": 0.0,
|
| 661 |
"step": 225
|
| 662 |
},
|
| 663 |
{
|
| 664 |
"clip_ratio": 0.0,
|
| 665 |
-
"completion_length": 609.
|
| 666 |
-
"epoch": 0.
|
| 667 |
-
"grad_norm":
|
| 668 |
-
"kl":
|
| 669 |
-
"learning_rate": 1.
|
| 670 |
-
"loss":
|
| 671 |
-
"reward": 0.
|
| 672 |
-
"reward_std": 0.
|
| 673 |
-
"rewards/accuracy_reward": 0.
|
| 674 |
"rewards/format_reward": 0.0,
|
| 675 |
"step": 230
|
| 676 |
},
|
| 677 |
{
|
| 678 |
"clip_ratio": 0.0,
|
| 679 |
-
"completion_length":
|
| 680 |
-
"epoch": 0.
|
| 681 |
-
"grad_norm":
|
| 682 |
-
"kl":
|
| 683 |
-
"learning_rate": 1.
|
| 684 |
-
"loss": 0.
|
| 685 |
-
"reward": 0.
|
| 686 |
-
"reward_std": 0.
|
| 687 |
-
"rewards/accuracy_reward": 0.
|
| 688 |
"rewards/format_reward": 0.0,
|
| 689 |
"step": 235
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"clip_ratio": 0.0,
|
| 693 |
-
"completion_length":
|
| 694 |
-
"epoch": 0.
|
| 695 |
-
"grad_norm":
|
| 696 |
-
"kl": 1.
|
| 697 |
-
"learning_rate":
|
| 698 |
-
"loss":
|
| 699 |
-
"reward": 0.
|
| 700 |
-
"reward_std": 0.
|
| 701 |
-
"rewards/accuracy_reward": 0.
|
| 702 |
"rewards/format_reward": 0.0,
|
| 703 |
"step": 240
|
| 704 |
},
|
| 705 |
{
|
| 706 |
"clip_ratio": 0.0,
|
| 707 |
-
"completion_length": 598.
|
| 708 |
-
"epoch": 0.
|
| 709 |
-
"grad_norm":
|
| 710 |
-
"kl":
|
| 711 |
-
"learning_rate":
|
| 712 |
-
"loss":
|
| 713 |
-
"reward": 0.
|
| 714 |
-
"reward_std": 0.
|
| 715 |
-
"rewards/accuracy_reward": 0.
|
| 716 |
"rewards/format_reward": 0.0,
|
| 717 |
"step": 245
|
| 718 |
},
|
| 719 |
{
|
| 720 |
"clip_ratio": 0.0,
|
| 721 |
-
"completion_length":
|
| 722 |
-
"epoch": 0.
|
| 723 |
-
"grad_norm": 1.
|
| 724 |
-
"kl":
|
| 725 |
-
"learning_rate":
|
| 726 |
-
"loss":
|
| 727 |
-
"reward": 0.
|
| 728 |
-
"reward_std": 0.
|
| 729 |
-
"rewards/accuracy_reward": 0.
|
| 730 |
"rewards/format_reward": 0.0,
|
| 731 |
"step": 250
|
| 732 |
},
|
| 733 |
{
|
| 734 |
"clip_ratio": 0.0,
|
| 735 |
-
"completion_length":
|
| 736 |
-
"epoch": 0.
|
| 737 |
-
"grad_norm":
|
| 738 |
-
"kl":
|
| 739 |
-
"learning_rate": 1.
|
| 740 |
-
"loss":
|
| 741 |
-
"reward": 0.
|
| 742 |
-
"reward_std": 0.
|
| 743 |
-
"rewards/accuracy_reward": 0.
|
| 744 |
"rewards/format_reward": 0.0,
|
| 745 |
"step": 255
|
| 746 |
},
|
| 747 |
{
|
| 748 |
"clip_ratio": 0.0,
|
| 749 |
-
"completion_length":
|
| 750 |
-
"epoch": 0.
|
| 751 |
-
"grad_norm":
|
| 752 |
-
"kl":
|
| 753 |
-
"learning_rate":
|
| 754 |
-
"loss": 0.
|
| 755 |
-
"reward": 0.
|
| 756 |
-
"reward_std": 0.
|
| 757 |
-
"rewards/accuracy_reward": 0.
|
| 758 |
"rewards/format_reward": 0.0,
|
| 759 |
"step": 260
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"clip_ratio": 0.0,
|
| 763 |
-
"completion_length":
|
| 764 |
-
"epoch": 0.
|
| 765 |
-
"grad_norm":
|
| 766 |
-
"kl":
|
| 767 |
-
"learning_rate":
|
| 768 |
-
"loss": 0.
|
| 769 |
-
"reward": 0.
|
| 770 |
-
"reward_std": 0.
|
| 771 |
-
"rewards/accuracy_reward": 0.
|
| 772 |
"rewards/format_reward": 0.0,
|
| 773 |
"step": 265
|
| 774 |
},
|
| 775 |
{
|
| 776 |
"clip_ratio": 0.0,
|
| 777 |
-
"completion_length":
|
| 778 |
-
"epoch": 0.
|
| 779 |
-
"
|
| 780 |
-
"
|
| 781 |
-
"
|
| 782 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 783 |
"rewards/format_reward": 0.0,
|
| 784 |
-
"step":
|
| 785 |
"total_flos": 0.0,
|
| 786 |
-
"train_loss":
|
| 787 |
-
"train_runtime":
|
| 788 |
-
"train_samples_per_second": 0.
|
| 789 |
"train_steps_per_second": 0.001
|
| 790 |
}
|
| 791 |
],
|
| 792 |
"logging_steps": 5,
|
| 793 |
-
"max_steps":
|
| 794 |
"num_input_tokens_seen": 0,
|
| 795 |
"num_train_epochs": 1,
|
| 796 |
"save_steps": 500,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9981333333333333,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 394,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"clip_ratio": 0.0,
|
| 13 |
+
"completion_length": 579.8596701371042,
|
| 14 |
+
"epoch": 0.002533333333333333,
|
| 15 |
+
"grad_norm": 0.2689627707004547,
|
| 16 |
+
"kl": -9.097551044664885e-08,
|
| 17 |
+
"learning_rate": 7.500000000000001e-08,
|
| 18 |
+
"loss": 0.0054,
|
| 19 |
+
"reward": 0.6754386079938788,
|
| 20 |
+
"reward_std": 0.29693952202796936,
|
| 21 |
+
"rewards/accuracy_reward": 0.6754386079938788,
|
| 22 |
"rewards/format_reward": 0.0,
|
| 23 |
"step": 1
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"clip_ratio": 0.0,
|
| 27 |
+
"completion_length": 623.1096651177658,
|
| 28 |
+
"epoch": 0.012666666666666666,
|
| 29 |
+
"grad_norm": 0.34043097496032715,
|
| 30 |
+
"kl": 0.00013973210987291838,
|
| 31 |
+
"learning_rate": 3.75e-07,
|
| 32 |
+
"loss": -0.0101,
|
| 33 |
+
"reward": 0.6052631694627436,
|
| 34 |
+
"reward_std": 0.3655273945708024,
|
| 35 |
+
"rewards/accuracy_reward": 0.6052631694627436,
|
| 36 |
"rewards/format_reward": 0.0,
|
| 37 |
"step": 5
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"clip_ratio": 0.0,
|
| 41 |
+
"completion_length": 574.2947516190378,
|
| 42 |
+
"epoch": 0.025333333333333333,
|
| 43 |
+
"grad_norm": 0.16866758465766907,
|
| 44 |
+
"kl": 0.0002021940130936472,
|
| 45 |
+
"learning_rate": 7.5e-07,
|
| 46 |
+
"loss": -0.0133,
|
| 47 |
+
"reward": 0.6228070329678687,
|
| 48 |
+
"reward_std": 0.3261075675487518,
|
| 49 |
+
"rewards/accuracy_reward": 0.6228070329678687,
|
| 50 |
"rewards/format_reward": 0.0,
|
| 51 |
"step": 10
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"clip_ratio": 0.0,
|
| 55 |
+
"completion_length": 593.8473849647924,
|
| 56 |
+
"epoch": 0.038,
|
| 57 |
+
"grad_norm": 0.13506826758384705,
|
| 58 |
+
"kl": 0.00030391090794613484,
|
| 59 |
+
"learning_rate": 1.125e-06,
|
| 60 |
+
"loss": -0.0171,
|
| 61 |
+
"reward": 0.6192982615608918,
|
| 62 |
+
"reward_std": 0.3493922035945089,
|
| 63 |
+
"rewards/accuracy_reward": 0.6192982615608918,
|
| 64 |
"rewards/format_reward": 0.0,
|
| 65 |
"step": 15
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"clip_ratio": 0.0,
|
| 69 |
+
"completion_length": 614.5087890625,
|
| 70 |
+
"epoch": 0.050666666666666665,
|
| 71 |
+
"grad_norm": 0.37117841839790344,
|
| 72 |
+
"kl": 0.0014222195273951481,
|
| 73 |
+
"learning_rate": 1.5e-06,
|
| 74 |
+
"loss": -0.0263,
|
| 75 |
+
"reward": 0.6087719443597291,
|
| 76 |
+
"reward_std": 0.35981847079176654,
|
| 77 |
+
"rewards/accuracy_reward": 0.6087719443597291,
|
| 78 |
"rewards/format_reward": 0.0,
|
| 79 |
"step": 20
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"clip_ratio": 0.0,
|
| 83 |
+
"completion_length": 610.6105422170539,
|
| 84 |
+
"epoch": 0.06333333333333334,
|
| 85 |
+
"grad_norm": 0.6702502369880676,
|
| 86 |
+
"kl": 0.008063848395096627,
|
| 87 |
+
"learning_rate": 1.875e-06,
|
| 88 |
+
"loss": -0.0323,
|
| 89 |
+
"reward": 0.6000000163128502,
|
| 90 |
+
"reward_std": 0.3776336585220538,
|
| 91 |
+
"rewards/accuracy_reward": 0.6000000163128502,
|
| 92 |
"rewards/format_reward": 0.0,
|
| 93 |
"step": 25
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"clip_ratio": 0.0,
|
| 97 |
+
"completion_length": 581.708786171361,
|
| 98 |
+
"epoch": 0.076,
|
| 99 |
+
"grad_norm": 0.13385862112045288,
|
| 100 |
+
"kl": 0.08222049913908305,
|
| 101 |
+
"learning_rate": 2.25e-06,
|
| 102 |
+
"loss": -0.003,
|
| 103 |
+
"reward": 0.5719298386260083,
|
| 104 |
+
"reward_std": 0.3216610023849889,
|
| 105 |
+
"rewards/accuracy_reward": 0.5719298386260083,
|
| 106 |
"rewards/format_reward": 0.0,
|
| 107 |
"step": 30
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"clip_ratio": 0.0,
|
| 111 |
+
"completion_length": 580.1789640727796,
|
| 112 |
+
"epoch": 0.08866666666666667,
|
| 113 |
+
"grad_norm": 0.1613887995481491,
|
| 114 |
+
"kl": 0.4072471217105263,
|
| 115 |
+
"learning_rate": 2.6250000000000003e-06,
|
| 116 |
+
"loss": -0.0034,
|
| 117 |
+
"reward": 0.5684210676895944,
|
| 118 |
+
"reward_std": 0.3606692671775818,
|
| 119 |
+
"rewards/accuracy_reward": 0.5684210676895944,
|
| 120 |
"rewards/format_reward": 0.0,
|
| 121 |
"step": 35
|
| 122 |
},
|
| 123 |
{
|
| 124 |
"clip_ratio": 0.0,
|
| 125 |
+
"completion_length": 573.603520443565,
|
| 126 |
+
"epoch": 0.10133333333333333,
|
| 127 |
+
"grad_norm": 4.28112268447876,
|
| 128 |
+
"kl": 0.28351083052785775,
|
| 129 |
+
"learning_rate": 3e-06,
|
| 130 |
+
"loss": -0.0148,
|
| 131 |
+
"reward": 0.6192982620314548,
|
| 132 |
+
"reward_std": 0.34681932079164607,
|
| 133 |
+
"rewards/accuracy_reward": 0.6192982620314548,
|
| 134 |
+
"rewards/format_reward": 0.0,
|
| 135 |
"step": 40
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"clip_ratio": 0.0,
|
| 139 |
+
"completion_length": 597.3666823537726,
|
| 140 |
+
"epoch": 0.114,
|
| 141 |
+
"grad_norm": 0.09016856551170349,
|
| 142 |
+
"kl": 0.1709989447342722,
|
| 143 |
+
"learning_rate": 2.998523534736735e-06,
|
| 144 |
+
"loss": -0.0274,
|
| 145 |
+
"reward": 0.6052631729527523,
|
| 146 |
+
"reward_std": 0.37054079206366286,
|
| 147 |
+
"rewards/accuracy_reward": 0.6052631729527523,
|
| 148 |
"rewards/format_reward": 0.0,
|
| 149 |
"step": 45
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"clip_ratio": 0.0,
|
| 153 |
+
"completion_length": 588.608786171361,
|
| 154 |
+
"epoch": 0.12666666666666668,
|
| 155 |
+
"grad_norm": 0.2084578275680542,
|
| 156 |
+
"kl": 0.2026009107890882,
|
| 157 |
+
"learning_rate": 2.994097045546504e-06,
|
| 158 |
+
"loss": -0.0133,
|
| 159 |
+
"reward": 0.5824561528469386,
|
| 160 |
+
"reward_std": 0.3644034429600364,
|
| 161 |
+
"rewards/accuracy_reward": 0.5824561528469386,
|
| 162 |
"rewards/format_reward": 0.0,
|
| 163 |
"step": 50
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"clip_ratio": 0.0,
|
| 167 |
+
"completion_length": 592.9579096743935,
|
| 168 |
+
"epoch": 0.13933333333333334,
|
| 169 |
+
"grad_norm": 1.332377552986145,
|
| 170 |
+
"kl": 0.253752979479338,
|
| 171 |
+
"learning_rate": 2.986729246506011e-06,
|
| 172 |
+
"loss": -0.0199,
|
| 173 |
+
"reward": 0.6315789619558736,
|
| 174 |
+
"reward_std": 0.34024513646175986,
|
| 175 |
+
"rewards/accuracy_reward": 0.6315789619558736,
|
| 176 |
"rewards/format_reward": 0.0,
|
| 177 |
"step": 55
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"clip_ratio": 0.0,
|
| 181 |
+
"completion_length": 580.366682032535,
|
| 182 |
+
"epoch": 0.152,
|
| 183 |
+
"grad_norm": 0.056244488805532455,
|
| 184 |
+
"kl": 17.54536361694336,
|
| 185 |
+
"learning_rate": 2.976434642014389e-06,
|
| 186 |
+
"loss": 0.8101,
|
| 187 |
+
"reward": 0.6280701911763141,
|
| 188 |
+
"reward_std": 0.3611795199544806,
|
| 189 |
+
"rewards/accuracy_reward": 0.6280701911763141,
|
| 190 |
"rewards/format_reward": 0.0,
|
| 191 |
"step": 60
|
| 192 |
},
|
| 193 |
{
|
| 194 |
"clip_ratio": 0.0,
|
| 195 |
+
"completion_length": 602.4929979826275,
|
| 196 |
+
"epoch": 0.16466666666666666,
|
| 197 |
+
"grad_norm": 2.9207770824432373,
|
| 198 |
+
"kl": 0.25210153680098685,
|
| 199 |
+
"learning_rate": 2.9632334982395456e-06,
|
| 200 |
+
"loss": 0.0016,
|
| 201 |
+
"reward": 0.6473684381497534,
|
| 202 |
+
"reward_std": 0.3406482906718003,
|
| 203 |
+
"rewards/accuracy_reward": 0.6473684381497534,
|
| 204 |
"rewards/format_reward": 0.0,
|
| 205 |
"step": 65
|
| 206 |
},
|
| 207 |
{
|
| 208 |
"clip_ratio": 0.0,
|
| 209 |
+
"completion_length": 550.657911119963,
|
| 210 |
+
"epoch": 0.17733333333333334,
|
| 211 |
+
"grad_norm": 2.1469576358795166,
|
| 212 |
+
"kl": 1.1785144203587583,
|
| 213 |
+
"learning_rate": 2.947151803221774e-06,
|
| 214 |
+
"loss": 0.0246,
|
| 215 |
+
"reward": 0.5982456306093618,
|
| 216 |
+
"reward_std": 0.3625405713131553,
|
| 217 |
+
"rewards/accuracy_reward": 0.5982456306093618,
|
| 218 |
"rewards/format_reward": 0.0,
|
| 219 |
"step": 70
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"clip_ratio": 0.0,
|
| 223 |
+
"completion_length": 630.1982622648541,
|
| 224 |
+
"epoch": 0.19,
|
| 225 |
+
"grad_norm": 0.437906414270401,
|
| 226 |
+
"kl": 0.2049952456825658,
|
| 227 |
+
"learning_rate": 2.928221215713164e-06,
|
| 228 |
+
"loss": -0.0136,
|
| 229 |
+
"reward": 0.6368421231445514,
|
| 230 |
+
"reward_std": 0.3507869362831116,
|
| 231 |
+
"rewards/accuracy_reward": 0.6368421231445514,
|
| 232 |
"rewards/format_reward": 0.0,
|
| 233 |
"step": 75
|
| 234 |
},
|
| 235 |
{
|
| 236 |
"clip_ratio": 0.0,
|
| 237 |
+
"completion_length": 554.66141614412,
|
| 238 |
+
"epoch": 0.20266666666666666,
|
| 239 |
+
"grad_norm": 0.20336699485778809,
|
| 240 |
+
"kl": 0.840125154194079,
|
| 241 |
+
"learning_rate": 2.906479002853542e-06,
|
| 242 |
+
"loss": -0.0331,
|
| 243 |
+
"reward": 0.6403508934535478,
|
| 244 |
+
"reward_std": 0.33991540262573644,
|
| 245 |
+
"rewards/accuracy_reward": 0.6403508934535478,
|
| 246 |
"rewards/format_reward": 0.0,
|
| 247 |
"step": 80
|
| 248 |
},
|
| 249 |
{
|
| 250 |
"clip_ratio": 0.0,
|
| 251 |
+
"completion_length": 571.3140498111122,
|
| 252 |
+
"epoch": 0.21533333333333332,
|
| 253 |
+
"grad_norm": 0.566848874092102,
|
| 254 |
+
"kl": 0.42301218133223684,
|
| 255 |
+
"learning_rate": 2.8819679668056195e-06,
|
| 256 |
+
"loss": -0.0154,
|
| 257 |
+
"reward": 0.6754386137974889,
|
| 258 |
+
"reward_std": 0.3516690385969062,
|
| 259 |
+
"rewards/accuracy_reward": 0.6754386137974889,
|
| 260 |
"rewards/format_reward": 0.0,
|
| 261 |
"step": 85
|
| 262 |
},
|
| 263 |
{
|
| 264 |
"clip_ratio": 0.0,
|
| 265 |
+
"completion_length": 576.1280825966282,
|
| 266 |
+
"epoch": 0.228,
|
| 267 |
+
"grad_norm": 5.783878326416016,
|
| 268 |
+
"kl": 0.3097421746504934,
|
| 269 |
+
"learning_rate": 2.8547363604937856e-06,
|
| 270 |
+
"loss": 0.0077,
|
| 271 |
+
"reward": 0.612280716237269,
|
| 272 |
+
"reward_std": 0.355850856241427,
|
| 273 |
+
"rewards/accuracy_reward": 0.612280716237269,
|
| 274 |
"rewards/format_reward": 0.0,
|
| 275 |
"step": 90
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"clip_ratio": 0.0,
|
| 279 |
+
"completion_length": 615.8421213250411,
|
| 280 |
+
"epoch": 0.24066666666666667,
|
| 281 |
+
"grad_norm": 0.29518190026283264,
|
| 282 |
+
"kl": 0.10967632092927632,
|
| 283 |
+
"learning_rate": 2.824837792612416e-06,
|
| 284 |
+
"loss": -0.0353,
|
| 285 |
+
"reward": 0.6122807160804146,
|
| 286 |
+
"reward_std": 0.3359477858794363,
|
| 287 |
+
"rewards/accuracy_reward": 0.6122807160804146,
|
| 288 |
"rewards/format_reward": 0.0,
|
| 289 |
"step": 95
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 0.25333333333333335,
|
| 293 |
+
"grad_norm": 0.09735328704118729,
|
| 294 |
+
"learning_rate": 2.792331122090709e-06,
|
| 295 |
+
"loss": -0.0192,
|
| 296 |
+
"step": 100
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.25333333333333335,
|
| 300 |
+
"eval_clip_ratio": 0.0,
|
| 301 |
+
"eval_completion_length": 581.6460822784423,
|
| 302 |
+
"eval_kl": 0.4462999359130859,
|
| 303 |
+
"eval_loss": -0.020424701273441315,
|
| 304 |
+
"eval_reward": 0.4882333454877138,
|
| 305 |
+
"eval_reward_std": 0.3652301513493061,
|
| 306 |
+
"eval_rewards/accuracy_reward": 0.4882333454877138,
|
| 307 |
+
"eval_rewards/format_reward": 0.0,
|
| 308 |
+
"eval_runtime": 95085.0702,
|
| 309 |
+
"eval_samples_per_second": 0.053,
|
| 310 |
+
"eval_steps_per_second": 0.009,
|
| 311 |
"step": 100
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"clip_ratio": 0.0,
|
| 315 |
+
"completion_length": 560.9368566412675,
|
| 316 |
+
"epoch": 0.266,
|
| 317 |
+
"grad_norm": 13.26487922668457,
|
| 318 |
+
"kl": 0.22525361713610198,
|
| 319 |
+
"learning_rate": 2.7572803422217976e-06,
|
| 320 |
+
"loss": -0.0193,
|
| 321 |
+
"reward": 0.6263158046885541,
|
| 322 |
+
"reward_std": 0.3673114654264952,
|
| 323 |
+
"rewards/accuracy_reward": 0.6263158046885541,
|
| 324 |
"rewards/format_reward": 0.0,
|
| 325 |
"step": 105
|
| 326 |
},
|
| 327 |
{
|
| 328 |
"clip_ratio": 0.0,
|
| 329 |
+
"completion_length": 572.0772078664679,
|
| 330 |
+
"epoch": 0.2786666666666667,
|
| 331 |
+
"grad_norm": 5.774922847747803,
|
| 332 |
+
"kl": 1.1548378392269736,
|
| 333 |
+
"learning_rate": 2.71975445468425e-06,
|
| 334 |
+
"loss": -0.0186,
|
| 335 |
+
"reward": 0.6631579132456529,
|
| 336 |
+
"reward_std": 0.3699968554471668,
|
| 337 |
+
"rewards/accuracy_reward": 0.6631579132456529,
|
| 338 |
"rewards/format_reward": 0.0,
|
| 339 |
"step": 110
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"clip_ratio": 0.0,
|
| 343 |
+
"completion_length": 617.7017713044819,
|
| 344 |
+
"epoch": 0.29133333333333333,
|
| 345 |
+
"grad_norm": 0.3488950729370117,
|
| 346 |
+
"kl": 0.4739940442537007,
|
| 347 |
+
"learning_rate": 2.679827333703964e-06,
|
| 348 |
+
"loss": 0.0054,
|
| 349 |
+
"reward": 0.5789473808125446,
|
| 350 |
+
"reward_std": 0.359339523942847,
|
| 351 |
+
"rewards/accuracy_reward": 0.5789473808125446,
|
| 352 |
"rewards/format_reward": 0.0,
|
| 353 |
"step": 115
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"clip_ratio": 0.0,
|
| 357 |
+
"completion_length": 584.6894883005242,
|
| 358 |
+
"epoch": 0.304,
|
| 359 |
+
"grad_norm": 2.7635438442230225,
|
| 360 |
+
"kl": 0.8890348735608553,
|
| 361 |
+
"learning_rate": 2.637577580623858e-06,
|
| 362 |
+
"loss": -0.0151,
|
| 363 |
+
"reward": 0.5947368579475503,
|
| 364 |
+
"reward_std": 0.34380959868431094,
|
| 365 |
+
"rewards/accuracy_reward": 0.5947368579475503,
|
| 366 |
"rewards/format_reward": 0.0,
|
| 367 |
"step": 120
|
| 368 |
},
|
| 369 |
{
|
| 370 |
"clip_ratio": 0.0,
|
| 371 |
+
"completion_length": 615.0772094726562,
|
| 372 |
+
"epoch": 0.31666666666666665,
|
| 373 |
+
"grad_norm": 1.1497012376785278,
|
| 374 |
+
"kl": 0.36164293791118424,
|
| 375 |
+
"learning_rate": 2.593088369167671e-06,
|
| 376 |
+
"loss": -0.0181,
|
| 377 |
+
"reward": 0.6192982593649312,
|
| 378 |
+
"reward_std": 0.3435111723448101,
|
| 379 |
+
"rewards/accuracy_reward": 0.6192982593649312,
|
| 380 |
"rewards/format_reward": 0.0,
|
| 381 |
"step": 125
|
| 382 |
},
|
| 383 |
{
|
| 384 |
"clip_ratio": 0.0,
|
| 385 |
+
"completion_length": 576.5666825143915,
|
| 386 |
+
"epoch": 0.3293333333333333,
|
| 387 |
+
"grad_norm": 6.164768695831299,
|
| 388 |
+
"kl": 0.5363926937705592,
|
| 389 |
+
"learning_rate": 2.5464472817024772e-06,
|
| 390 |
+
"loss": 0.018,
|
| 391 |
+
"reward": 0.6684210694149921,
|
| 392 |
+
"reward_std": 0.368413172583831,
|
| 393 |
+
"rewards/accuracy_reward": 0.6684210694149921,
|
| 394 |
"rewards/format_reward": 0.0,
|
| 395 |
"step": 130
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"clip_ratio": 0.0,
|
| 399 |
+
"completion_length": 566.4719445479543,
|
| 400 |
+
"epoch": 0.342,
|
| 401 |
+
"grad_norm": 2.0130836963653564,
|
| 402 |
+
"kl": 1.256256103515625,
|
| 403 |
+
"learning_rate": 2.497746136822254e-06,
|
| 404 |
+
"loss": 0.0444,
|
| 405 |
+
"reward": 0.642105276176804,
|
| 406 |
+
"reward_std": 0.33451331728383116,
|
| 407 |
+
"rewards/accuracy_reward": 0.642105276176804,
|
| 408 |
"rewards/format_reward": 0.0,
|
| 409 |
"step": 135
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"clip_ratio": 0.0,
|
| 413 |
+
"completion_length": 615.5193124871505,
|
| 414 |
+
"epoch": 0.3546666666666667,
|
| 415 |
+
"grad_norm": 2.369405746459961,
|
| 416 |
+
"kl": 0.7050068102384869,
|
| 417 |
+
"learning_rate": 2.4470808085919304e-06,
|
| 418 |
+
"loss": -0.0144,
|
| 419 |
+
"reward": 0.6421052804118709,
|
| 420 |
+
"reward_std": 0.3574345387910542,
|
| 421 |
+
"rewards/accuracy_reward": 0.6421052804118709,
|
| 422 |
"rewards/format_reward": 0.0,
|
| 423 |
"step": 140
|
| 424 |
},
|
| 425 |
{
|
| 426 |
"clip_ratio": 0.0,
|
| 427 |
+
"completion_length": 576.3912459524055,
|
| 428 |
+
"epoch": 0.36733333333333335,
|
| 429 |
+
"grad_norm": 4.594287872314453,
|
| 430 |
+
"kl": 2.437645841899671,
|
| 431 |
+
"learning_rate": 2.3945510378077523e-06,
|
| 432 |
+
"loss": 0.0796,
|
| 433 |
+
"reward": 0.6842105448246002,
|
| 434 |
+
"reward_std": 0.350753252757223,
|
| 435 |
+
"rewards/accuracy_reward": 0.6842105448246002,
|
| 436 |
"rewards/format_reward": 0.0,
|
| 437 |
"step": 145
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"clip_ratio": 0.0,
|
| 441 |
+
"completion_length": 596.2193150570519,
|
| 442 |
+
"epoch": 0.38,
|
| 443 |
+
"grad_norm": 2.5179603099823,
|
| 444 |
+
"kl": 1.6508487099095395,
|
| 445 |
+
"learning_rate": 2.340260235645519e-06,
|
| 446 |
+
"loss": 0.0431,
|
| 447 |
+
"reward": 0.663157911363401,
|
| 448 |
+
"reward_std": 0.3722002707029644,
|
| 449 |
+
"rewards/accuracy_reward": 0.663157911363401,
|
| 450 |
"rewards/format_reward": 0.0,
|
| 451 |
"step": 150
|
| 452 |
},
|
| 453 |
{
|
| 454 |
"clip_ratio": 0.0,
|
| 455 |
+
"completion_length": 579.5965046129728,
|
| 456 |
+
"epoch": 0.39266666666666666,
|
| 457 |
+
"grad_norm": 76.81166076660156,
|
| 458 |
+
"kl": 4.629216886821546,
|
| 459 |
+
"learning_rate": 2.2843152800832416e-06,
|
| 460 |
+
"loss": 0.094,
|
| 461 |
+
"reward": 0.5807017698099739,
|
| 462 |
+
"reward_std": 0.37164790002923265,
|
| 463 |
+
"rewards/accuracy_reward": 0.5807017698099739,
|
| 464 |
"rewards/format_reward": 0.0,
|
| 465 |
"step": 155
|
| 466 |
},
|
| 467 |
{
|
| 468 |
"clip_ratio": 0.0,
|
| 469 |
+
"completion_length": 604.2456311677631,
|
| 470 |
+
"epoch": 0.4053333333333333,
|
| 471 |
+
"grad_norm": 2.316026210784912,
|
| 472 |
+
"kl": 5.967475971422697,
|
| 473 |
+
"learning_rate": 2.2268263054989753e-06,
|
| 474 |
+
"loss": 0.242,
|
| 475 |
+
"reward": 0.5666666804175627,
|
| 476 |
+
"reward_std": 0.3491695720898478,
|
| 477 |
+
"rewards/accuracy_reward": 0.5666666804175627,
|
| 478 |
"rewards/format_reward": 0.0,
|
| 479 |
"step": 160
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"clip_ratio": 0.0,
|
| 483 |
+
"completion_length": 585.1210690146999,
|
| 484 |
+
"epoch": 0.418,
|
| 485 |
+
"grad_norm": 1.3882242441177368,
|
| 486 |
+
"kl": 1.542066072162829,
|
| 487 |
+
"learning_rate": 2.167906485858047e-06,
|
| 488 |
+
"loss": 0.031,
|
| 489 |
+
"reward": 0.6631579112065465,
|
| 490 |
+
"reward_std": 0.32709676968423945,
|
| 491 |
+
"rewards/accuracy_reward": 0.6631579112065465,
|
| 492 |
"rewards/format_reward": 0.0,
|
| 493 |
"step": 165
|
| 494 |
},
|
| 495 |
{
|
| 496 |
"clip_ratio": 0.0,
|
| 497 |
+
"completion_length": 609.7789647152549,
|
| 498 |
+
"epoch": 0.43066666666666664,
|
| 499 |
+
"grad_norm": 23.364316940307617,
|
| 500 |
+
"kl": 3.0389545641447366,
|
| 501 |
+
"learning_rate": 2.1076718119164804e-06,
|
| 502 |
+
"loss": 0.0985,
|
| 503 |
+
"reward": 0.5368421203211734,
|
| 504 |
+
"reward_std": 0.36661528725373116,
|
| 505 |
+
"rewards/accuracy_reward": 0.5368421203211734,
|
| 506 |
"rewards/format_reward": 0.0,
|
| 507 |
"step": 170
|
| 508 |
},
|
| 509 |
{
|
| 510 |
"clip_ratio": 0.0,
|
| 511 |
+
"completion_length": 589.2105436626233,
|
| 512 |
+
"epoch": 0.44333333333333336,
|
| 513 |
+
"grad_norm": 2.261157274246216,
|
| 514 |
+
"kl": 3.034598581414474,
|
| 515 |
+
"learning_rate": 2.0462408628792335e-06,
|
| 516 |
+
"loss": 0.0464,
|
| 517 |
+
"reward": 0.5877193132513447,
|
| 518 |
+
"reward_std": 0.32118205584977805,
|
| 519 |
+
"rewards/accuracy_reward": 0.5877193132513447,
|
| 520 |
"rewards/format_reward": 0.0,
|
| 521 |
"step": 175
|
| 522 |
},
|
| 523 |
{
|
| 524 |
"clip_ratio": 0.0,
|
| 525 |
+
"completion_length": 583.0210664447985,
|
| 526 |
+
"epoch": 0.456,
|
| 527 |
+
"grad_norm": 1.490301251411438,
|
| 528 |
+
"kl": 4.092078279194079,
|
| 529 |
+
"learning_rate": 1.9837345729627633e-06,
|
| 530 |
+
"loss": 0.0794,
|
| 531 |
+
"reward": 0.6087719440460205,
|
| 532 |
+
"reward_std": 0.34601063853815983,
|
| 533 |
+
"rewards/accuracy_reward": 0.6087719440460205,
|
| 534 |
"rewards/format_reward": 0.0,
|
| 535 |
"step": 180
|
| 536 |
},
|
| 537 |
{
|
| 538 |
"clip_ratio": 0.0,
|
| 539 |
+
"completion_length": 591.0193131296259,
|
| 540 |
+
"epoch": 0.4686666666666667,
|
| 541 |
+
"grad_norm": 3.377087354660034,
|
| 542 |
+
"kl": 39.17373753597862,
|
| 543 |
+
"learning_rate": 1.9202759933214665e-06,
|
| 544 |
+
"loss": 1.5296,
|
| 545 |
+
"reward": 0.5684210672190315,
|
| 546 |
+
"reward_std": 0.3722002681932951,
|
| 547 |
+
"rewards/accuracy_reward": 0.5684210672190315,
|
| 548 |
"rewards/format_reward": 0.0,
|
| 549 |
"step": 185
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"clip_ratio": 0.0,
|
| 553 |
+
"completion_length": 628.3315945273952,
|
| 554 |
+
"epoch": 0.48133333333333334,
|
| 555 |
+
"grad_norm": 15.796894073486328,
|
| 556 |
+
"kl": 8.018802682976974,
|
| 557 |
+
"learning_rate": 1.8559900498066726e-06,
|
| 558 |
+
"loss": 0.2629,
|
| 559 |
+
"reward": 0.5877193149767423,
|
| 560 |
+
"reward_std": 0.3454245896715867,
|
| 561 |
+
"rewards/accuracy_reward": 0.5877193149767423,
|
| 562 |
"rewards/format_reward": 0.0,
|
| 563 |
"step": 190
|
| 564 |
},
|
| 565 |
{
|
| 566 |
"clip_ratio": 0.0,
|
| 567 |
+
"completion_length": 616.9403683311061,
|
| 568 |
+
"epoch": 0.494,
|
| 569 |
+
"grad_norm": 0.7116127014160156,
|
| 570 |
+
"kl": 11.00859888980263,
|
| 571 |
+
"learning_rate": 1.7910032970350677e-06,
|
| 572 |
+
"loss": 0.5039,
|
| 573 |
+
"reward": 0.614035103195592,
|
| 574 |
+
"reward_std": 0.3618389898224881,
|
| 575 |
+
"rewards/accuracy_reward": 0.614035103195592,
|
| 576 |
"rewards/format_reward": 0.0,
|
| 577 |
"step": 195
|
| 578 |
},
|
| 579 |
{
|
| 580 |
+
"epoch": 0.5066666666666667,
|
| 581 |
+
"grad_norm": 0.7055822014808655,
|
| 582 |
+
"learning_rate": 1.7254436692507058e-06,
|
| 583 |
+
"loss": 0.2295,
|
| 584 |
"step": 200
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 0.5066666666666667,
|
| 588 |
"eval_clip_ratio": 0.0,
|
| 589 |
+
"eval_completion_length": 581.6460156143188,
|
| 590 |
+
"eval_kl": 44.771087231445314,
|
| 591 |
+
"eval_loss": 1.8106327056884766,
|
| 592 |
+
"eval_reward": 0.4892333455443382,
|
| 593 |
+
"eval_reward_std": 0.362811917424202,
|
| 594 |
+
"eval_rewards/accuracy_reward": 0.4891666788816452,
|
| 595 |
+
"eval_rewards/format_reward": 6.666666865348816e-05,
|
| 596 |
+
"eval_runtime": 95474.2074,
|
| 597 |
+
"eval_samples_per_second": 0.052,
|
| 598 |
+
"eval_steps_per_second": 0.009,
|
| 599 |
"step": 200
|
| 600 |
},
|
| 601 |
{
|
| 602 |
"clip_ratio": 0.0,
|
| 603 |
+
"completion_length": 624.9587893837377,
|
| 604 |
+
"epoch": 0.5193333333333333,
|
| 605 |
+
"grad_norm": 0.2618827223777771,
|
| 606 |
+
"kl": 4.135612246864721,
|
| 607 |
+
"learning_rate": 1.6594402284710481e-06,
|
| 608 |
+
"loss": 0.0336,
|
| 609 |
+
"reward": 0.564035101783903,
|
| 610 |
+
"reward_std": 0.3579249236144518,
|
| 611 |
+
"rewards/accuracy_reward": 0.564035101783903,
|
| 612 |
"rewards/format_reward": 0.0,
|
| 613 |
"step": 205
|
| 614 |
},
|
| 615 |
{
|
| 616 |
"clip_ratio": 0.0,
|
| 617 |
+
"completion_length": 621.8368573640522,
|
| 618 |
+
"epoch": 0.532,
|
| 619 |
+
"grad_norm": 1.038638710975647,
|
| 620 |
+
"kl": 0.783648681640625,
|
| 621 |
+
"learning_rate": 1.593122910412851e-06,
|
| 622 |
+
"loss": 0.0144,
|
| 623 |
+
"reward": 0.6175438767985294,
|
| 624 |
+
"reward_std": 0.35596638729697777,
|
| 625 |
+
"rewards/accuracy_reward": 0.6175438767985294,
|
| 626 |
+
"rewards/format_reward": 0.0,
|
| 627 |
"step": 210
|
| 628 |
},
|
| 629 |
{
|
| 630 |
"clip_ratio": 0.0,
|
| 631 |
+
"completion_length": 612.468435990183,
|
| 632 |
+
"epoch": 0.5446666666666666,
|
| 633 |
+
"grad_norm": 15.33467960357666,
|
| 634 |
+
"kl": 5.278036338404605,
|
| 635 |
+
"learning_rate": 1.5266222686980693e-06,
|
| 636 |
+
"loss": 0.2523,
|
| 637 |
+
"reward": 0.5964912433373301,
|
| 638 |
+
"reward_std": 0.344139332834043,
|
| 639 |
+
"rewards/accuracy_reward": 0.5964912433373301,
|
| 640 |
"rewards/format_reward": 0.0,
|
| 641 |
"step": 215
|
| 642 |
},
|
| 643 |
{
|
| 644 |
"clip_ratio": 0.0,
|
| 645 |
+
"completion_length": 579.1210691753187,
|
| 646 |
+
"epoch": 0.5573333333333333,
|
| 647 |
+
"grad_norm": 14.52474308013916,
|
| 648 |
+
"kl": 0.7972836143092106,
|
| 649 |
+
"learning_rate": 1.460069217843338e-06,
|
| 650 |
+
"loss": -0.0145,
|
| 651 |
+
"reward": 0.5754386097192764,
|
| 652 |
+
"reward_std": 0.3417132879558362,
|
| 653 |
+
"rewards/accuracy_reward": 0.5754386097192764,
|
| 654 |
"rewards/format_reward": 0.0,
|
| 655 |
"step": 220
|
| 656 |
},
|
| 657 |
{
|
| 658 |
"clip_ratio": 0.0,
|
| 659 |
+
"completion_length": 567.8386103579872,
|
| 660 |
+
"epoch": 0.57,
|
| 661 |
+
"grad_norm": 4.826749324798584,
|
| 662 |
+
"kl": 1.95933837890625,
|
| 663 |
+
"learning_rate": 1.3935947755389924e-06,
|
| 664 |
+
"loss": -0.0023,
|
| 665 |
+
"reward": 0.6614035228365346,
|
| 666 |
+
"reward_std": 0.33425700288069876,
|
| 667 |
+
"rewards/accuracy_reward": 0.6614035228365346,
|
| 668 |
"rewards/format_reward": 0.0,
|
| 669 |
"step": 225
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"clip_ratio": 0.0,
|
| 673 |
+
"completion_length": 609.7438760857833,
|
| 674 |
+
"epoch": 0.5826666666666667,
|
| 675 |
+
"grad_norm": 2.4607062339782715,
|
| 676 |
+
"kl": 2.425996800472862,
|
| 677 |
+
"learning_rate": 1.3273298047249756e-06,
|
| 678 |
+
"loss": 0.0442,
|
| 679 |
+
"reward": 0.542105278059056,
|
| 680 |
+
"reward_std": 0.3630532004331288,
|
| 681 |
+
"rewards/accuracy_reward": 0.542105278059056,
|
| 682 |
"rewards/format_reward": 0.0,
|
| 683 |
"step": 230
|
| 684 |
},
|
| 685 |
{
|
| 686 |
"clip_ratio": 0.0,
|
| 687 |
+
"completion_length": 633.8508912739001,
|
| 688 |
+
"epoch": 0.5953333333333334,
|
| 689 |
+
"grad_norm": 1.241820216178894,
|
| 690 |
+
"kl": 4.220809454666941,
|
| 691 |
+
"learning_rate": 1.2614047559713923e-06,
|
| 692 |
+
"loss": 0.1149,
|
| 693 |
+
"reward": 0.5491228218141355,
|
| 694 |
+
"reward_std": 0.3291172884012524,
|
| 695 |
+
"rewards/accuracy_reward": 0.5491228218141355,
|
| 696 |
"rewards/format_reward": 0.0,
|
| 697 |
"step": 235
|
| 698 |
},
|
| 699 |
{
|
| 700 |
"clip_ratio": 0.0,
|
| 701 |
+
"completion_length": 607.8807173879524,
|
| 702 |
+
"epoch": 0.608,
|
| 703 |
+
"grad_norm": 4.545963287353516,
|
| 704 |
+
"kl": 1.1027640894839639,
|
| 705 |
+
"learning_rate": 1.1959494106708598e-06,
|
| 706 |
+
"loss": 0.0236,
|
| 707 |
+
"reward": 0.6000000144305982,
|
| 708 |
+
"reward_std": 0.36793422448007684,
|
| 709 |
+
"rewards/accuracy_reward": 0.6000000144305982,
|
| 710 |
"rewards/format_reward": 0.0,
|
| 711 |
"step": 240
|
| 712 |
},
|
| 713 |
{
|
| 714 |
"clip_ratio": 0.0,
|
| 715 |
+
"completion_length": 598.5158052143298,
|
| 716 |
+
"epoch": 0.6206666666666667,
|
| 717 |
+
"grad_norm": 3.975374221801758,
|
| 718 |
+
"kl": 17.801213314658717,
|
| 719 |
+
"learning_rate": 1.1310926255482204e-06,
|
| 720 |
+
"loss": 0.6585,
|
| 721 |
+
"reward": 0.6385965044561185,
|
| 722 |
+
"reward_std": 0.33627751940175105,
|
| 723 |
+
"rewards/accuracy_reward": 0.6385965044561185,
|
| 724 |
"rewards/format_reward": 0.0,
|
| 725 |
"step": 245
|
| 726 |
},
|
| 727 |
{
|
| 728 |
"clip_ratio": 0.0,
|
| 729 |
+
"completion_length": 573.9087850470291,
|
| 730 |
+
"epoch": 0.6333333333333333,
|
| 731 |
+
"grad_norm": 1.1734281778335571,
|
| 732 |
+
"kl": 2.3962697882401316,
|
| 733 |
+
"learning_rate": 1.0669620789905688e-06,
|
| 734 |
+
"loss": 0.027,
|
| 735 |
+
"reward": 0.5877193123102188,
|
| 736 |
+
"reward_std": 0.38500809261673374,
|
| 737 |
+
"rewards/accuracy_reward": 0.5877193123102188,
|
| 738 |
"rewards/format_reward": 0.0,
|
| 739 |
"step": 250
|
| 740 |
},
|
| 741 |
{
|
| 742 |
"clip_ratio": 0.0,
|
| 743 |
+
"completion_length": 606.8912443462171,
|
| 744 |
+
"epoch": 0.646,
|
| 745 |
+
"grad_norm": 2.513296365737915,
|
| 746 |
+
"kl": 2.2693404348273027,
|
| 747 |
+
"learning_rate": 1.0036840196969795e-06,
|
| 748 |
+
"loss": 0.0564,
|
| 749 |
+
"reward": 0.6245614178870854,
|
| 750 |
+
"reward_std": 0.32143837056661906,
|
| 751 |
+
"rewards/accuracy_reward": 0.6245614178870854,
|
| 752 |
"rewards/format_reward": 0.0,
|
| 753 |
"step": 255
|
| 754 |
},
|
| 755 |
{
|
| 756 |
"clip_ratio": 0.0,
|
| 757 |
+
"completion_length": 567.249137316252,
|
| 758 |
+
"epoch": 0.6586666666666666,
|
| 759 |
+
"grad_norm": 1.0174099206924438,
|
| 760 |
+
"kl": 3.5286929481907894,
|
| 761 |
+
"learning_rate": 9.413830181427508e-07,
|
| 762 |
+
"loss": 0.0882,
|
| 763 |
+
"reward": 0.6456140494660327,
|
| 764 |
+
"reward_std": 0.3340006878501491,
|
| 765 |
+
"rewards/accuracy_reward": 0.6456140494660327,
|
| 766 |
"rewards/format_reward": 0.0,
|
| 767 |
"step": 260
|
| 768 |
},
|
| 769 |
{
|
| 770 |
"clip_ratio": 0.0,
|
| 771 |
+
"completion_length": 610.2052803441098,
|
| 772 |
+
"epoch": 0.6713333333333333,
|
| 773 |
+
"grad_norm": 1.029402256011963,
|
| 774 |
+
"kl": 2.918760922080592,
|
| 775 |
+
"learning_rate": 8.801817213474331e-07,
|
| 776 |
+
"loss": 0.0485,
|
| 777 |
+
"reward": 0.5929824714597903,
|
| 778 |
+
"reward_std": 0.34714905211800023,
|
| 779 |
+
"rewards/accuracy_reward": 0.5929824714597903,
|
| 780 |
"rewards/format_reward": 0.0,
|
| 781 |
"step": 265
|
| 782 |
},
|
| 783 |
{
|
| 784 |
"clip_ratio": 0.0,
|
| 785 |
+
"completion_length": 638.8596645957546,
|
| 786 |
+
"epoch": 0.684,
|
| 787 |
+
"grad_norm": 1.5382988452911377,
|
| 788 |
+
"kl": 2.563796193976151,
|
| 789 |
+
"learning_rate": 8.202006114294044e-07,
|
| 790 |
+
"loss": 0.0392,
|
| 791 |
+
"reward": 0.5807017699668282,
|
| 792 |
+
"reward_std": 0.36489082386619165,
|
| 793 |
+
"rewards/accuracy_reward": 0.5807017699668282,
|
| 794 |
+
"rewards/format_reward": 0.0,
|
| 795 |
+
"step": 270
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"clip_ratio": 0.0,
|
| 799 |
+
"completion_length": 608.0421232524671,
|
| 800 |
+
"epoch": 0.6966666666666667,
|
| 801 |
+
"grad_norm": 16.033626556396484,
|
| 802 |
+
"kl": 1.5960372121710527,
|
| 803 |
+
"learning_rate": 7.615577684223272e-07,
|
| 804 |
+
"loss": 0.0303,
|
| 805 |
+
"reward": 0.5894736991116875,
|
| 806 |
+
"reward_std": 0.3749873553451739,
|
| 807 |
+
"rewards/accuracy_reward": 0.5894736991116875,
|
| 808 |
+
"rewards/format_reward": 0.0,
|
| 809 |
+
"step": 275
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"clip_ratio": 0.0,
|
| 813 |
+
"completion_length": 600.415805053711,
|
| 814 |
+
"epoch": 0.7093333333333334,
|
| 815 |
+
"grad_norm": 2.8626461029052734,
|
| 816 |
+
"kl": 1.9747047825863486,
|
| 817 |
+
"learning_rate": 7.043686378203864e-07,
|
| 818 |
+
"loss": 0.0147,
|
| 819 |
+
"reward": 0.6368421217328624,
|
| 820 |
+
"reward_std": 0.37260342334446156,
|
| 821 |
+
"rewards/accuracy_reward": 0.6368421217328624,
|
| 822 |
+
"rewards/format_reward": 0.0,
|
| 823 |
+
"step": 280
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"clip_ratio": 0.0,
|
| 827 |
+
"completion_length": 616.5772094726562,
|
| 828 |
+
"epoch": 0.722,
|
| 829 |
+
"grad_norm": 1.4815430641174316,
|
| 830 |
+
"kl": 2.7695633737664473,
|
| 831 |
+
"learning_rate": 6.487458033099425e-07,
|
| 832 |
+
"loss": 0.0412,
|
| 833 |
+
"reward": 0.5385965032012839,
|
| 834 |
+
"reward_std": 0.35034166982299403,
|
| 835 |
+
"rewards/accuracy_reward": 0.5385965032012839,
|
| 836 |
+
"rewards/format_reward": 0.0,
|
| 837 |
+
"step": 285
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"clip_ratio": 0.0,
|
| 841 |
+
"completion_length": 645.3333529823705,
|
| 842 |
+
"epoch": 0.7346666666666667,
|
| 843 |
+
"grad_norm": 0.6981754302978516,
|
| 844 |
+
"kl": 2.3487503854851974,
|
| 845 |
+
"learning_rate": 5.947987651349942e-07,
|
| 846 |
+
"loss": 0.0472,
|
| 847 |
+
"reward": 0.566666682927232,
|
| 848 |
+
"reward_std": 0.3772305058805566,
|
| 849 |
+
"rewards/accuracy_reward": 0.566666682927232,
|
| 850 |
+
"rewards/format_reward": 0.0,
|
| 851 |
+
"step": 290
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"clip_ratio": 0.0,
|
| 855 |
+
"completion_length": 615.3684377569901,
|
| 856 |
+
"epoch": 0.7473333333333333,
|
| 857 |
+
"grad_norm": 2.056654214859009,
|
| 858 |
+
"kl": 2.483375308388158,
|
| 859 |
+
"learning_rate": 5.426337245327703e-07,
|
| 860 |
+
"loss": 0.0787,
|
| 861 |
+
"reward": 0.5491228204024465,
|
| 862 |
+
"reward_std": 0.3359477874479796,
|
| 863 |
+
"rewards/accuracy_reward": 0.5491228204024465,
|
| 864 |
+
"rewards/format_reward": 0.0,
|
| 865 |
+
"step": 295
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 0.76,
|
| 869 |
+
"grad_norm": 0.973532497882843,
|
| 870 |
+
"learning_rate": 4.923533746638108e-07,
|
| 871 |
+
"loss": 0.0014,
|
| 872 |
+
"step": 300
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.76,
|
| 876 |
+
"eval_clip_ratio": 0.0,
|
| 877 |
+
"eval_completion_length": 579.6517157089233,
|
| 878 |
+
"eval_kl": 2.4090683868408203,
|
| 879 |
+
"eval_loss": 0.046611957252025604,
|
| 880 |
+
"eval_reward": 0.49193334555327894,
|
| 881 |
+
"eval_reward_std": 0.35773685903549196,
|
| 882 |
+
"eval_rewards/accuracy_reward": 0.49193334555327894,
|
| 883 |
+
"eval_rewards/format_reward": 0.0,
|
| 884 |
+
"eval_runtime": 94754.8266,
|
| 885 |
+
"eval_samples_per_second": 0.053,
|
| 886 |
+
"eval_steps_per_second": 0.009,
|
| 887 |
+
"step": 300
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"clip_ratio": 0.0,
|
| 891 |
+
"completion_length": 594.8903664036801,
|
| 892 |
+
"epoch": 0.7726666666666666,
|
| 893 |
+
"grad_norm": 0.28998905420303345,
|
| 894 |
+
"kl": 1.3069589715254935,
|
| 895 |
+
"learning_rate": 4.440566984481256e-07,
|
| 896 |
+
"loss": 0.0171,
|
| 897 |
+
"reward": 0.6210526459311184,
|
| 898 |
+
"reward_std": 0.34825076147129663,
|
| 899 |
+
"rewards/accuracy_reward": 0.6210526459311184,
|
| 900 |
+
"rewards/format_reward": 0.0,
|
| 901 |
+
"step": 305
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"clip_ratio": 0.0,
|
| 905 |
+
"completion_length": 618.7473841616982,
|
| 906 |
+
"epoch": 0.7853333333333333,
|
| 907 |
+
"grad_norm": 0.46500492095947266,
|
| 908 |
+
"kl": 1.2823197214226973,
|
| 909 |
+
"learning_rate": 3.978387737053994e-07,
|
| 910 |
+
"loss": 0.0136,
|
| 911 |
+
"reward": 0.5666666805744172,
|
| 912 |
+
"reward_std": 0.32507625222206116,
|
| 913 |
+
"rewards/accuracy_reward": 0.5666666805744172,
|
| 914 |
+
"rewards/format_reward": 0.0,
|
| 915 |
+
"step": 310
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"clip_ratio": 0.0,
|
| 919 |
+
"completion_length": 618.5772081877055,
|
| 920 |
+
"epoch": 0.798,
|
| 921 |
+
"grad_norm": 0.40139040350914,
|
| 922 |
+
"kl": 1.015325927734375,
|
| 923 |
+
"learning_rate": 3.5379058598286167e-07,
|
| 924 |
+
"loss": -0.0158,
|
| 925 |
+
"reward": 0.5438596634488356,
|
| 926 |
+
"reward_std": 0.3694023759741532,
|
| 927 |
+
"rewards/accuracy_reward": 0.5438596634488356,
|
| 928 |
+
"rewards/format_reward": 0.0,
|
| 929 |
+
"step": 315
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"clip_ratio": 0.0,
|
| 933 |
+
"completion_length": 594.957911119963,
|
| 934 |
+
"epoch": 0.8106666666666666,
|
| 935 |
+
"grad_norm": 1.3908910751342773,
|
| 936 |
+
"kl": 1.2546104029605263,
|
| 937 |
+
"learning_rate": 3.119988494392894e-07,
|
| 938 |
+
"loss": 0.0289,
|
| 939 |
+
"reward": 0.6508772078313325,
|
| 940 |
+
"reward_std": 0.3327888513866224,
|
| 941 |
+
"rewards/accuracy_reward": 0.6508772078313325,
|
| 942 |
+
"rewards/format_reward": 0.0,
|
| 943 |
+
"step": 320
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"clip_ratio": 0.0,
|
| 947 |
+
"completion_length": 582.7333484850432,
|
| 948 |
+
"epoch": 0.8233333333333334,
|
| 949 |
+
"grad_norm": 4.380345344543457,
|
| 950 |
+
"kl": 418.79378180252877,
|
| 951 |
+
"learning_rate": 2.725458361377465e-07,
|
| 952 |
+
"loss": 15.2043,
|
| 953 |
+
"reward": 0.608771941379497,
|
| 954 |
+
"reward_std": 0.34012960415137444,
|
| 955 |
+
"rewards/accuracy_reward": 0.608771941379497,
|
| 956 |
+
"rewards/format_reward": 0.0,
|
| 957 |
+
"step": 325
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"clip_ratio": 0.0,
|
| 961 |
+
"completion_length": 634.2175588507401,
|
| 962 |
+
"epoch": 0.836,
|
| 963 |
+
"grad_norm": 6.396597862243652,
|
| 964 |
+
"kl": 1.1365401418585526,
|
| 965 |
+
"learning_rate": 2.3550921408312737e-07,
|
| 966 |
+
"loss": 0.0132,
|
| 967 |
+
"reward": 0.5859649261361675,
|
| 968 |
+
"reward_std": 0.34545827006038865,
|
| 969 |
+
"rewards/accuracy_reward": 0.5859649261361675,
|
| 970 |
+
"rewards/format_reward": 0.0,
|
| 971 |
+
"step": 330
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"clip_ratio": 0.0,
|
| 975 |
+
"completion_length": 614.2140493292558,
|
| 976 |
+
"epoch": 0.8486666666666667,
|
| 977 |
+
"grad_norm": 0.9369886517524719,
|
| 978 |
+
"kl": 1.5584103232935855,
|
| 979 |
+
"learning_rate": 2.0096189432334195e-07,
|
| 980 |
+
"loss": 0.0201,
|
| 981 |
+
"reward": 0.6035087874061182,
|
| 982 |
+
"reward_std": 0.3459035368342149,
|
| 983 |
+
"rewards/accuracy_reward": 0.6035087874061182,
|
| 984 |
+
"rewards/format_reward": 0.0,
|
| 985 |
+
"step": 335
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"clip_ratio": 0.0,
|
| 989 |
+
"completion_length": 619.8245766087582,
|
| 990 |
+
"epoch": 0.8613333333333333,
|
| 991 |
+
"grad_norm": 0.42530328035354614,
|
| 992 |
+
"kl": 1.4711069207442433,
|
| 993 |
+
"learning_rate": 1.6897188741514286e-07,
|
| 994 |
+
"loss": 0.0519,
|
| 995 |
+
"reward": 0.575438610503548,
|
| 996 |
+
"reward_std": 0.3146836676095661,
|
| 997 |
+
"rewards/accuracy_reward": 0.575438610503548,
|
| 998 |
+
"rewards/format_reward": 0.0,
|
| 999 |
+
"step": 340
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"clip_ratio": 0.0,
|
| 1003 |
+
"completion_length": 577.3737000314812,
|
| 1004 |
+
"epoch": 0.874,
|
| 1005 |
+
"grad_norm": 0.3861980140209198,
|
| 1006 |
+
"kl": 1.7285689504523025,
|
| 1007 |
+
"learning_rate": 1.396021695371582e-07,
|
| 1008 |
+
"loss": -0.0088,
|
| 1009 |
+
"reward": 0.6385965082206224,
|
| 1010 |
+
"reward_std": 0.37675155840421976,
|
| 1011 |
+
"rewards/accuracy_reward": 0.6385965082206224,
|
| 1012 |
+
"rewards/format_reward": 0.0,
|
| 1013 |
+
"step": 345
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"clip_ratio": 0.0,
|
| 1017 |
+
"completion_length": 629.012297941509,
|
| 1018 |
+
"epoch": 0.8866666666666667,
|
| 1019 |
+
"grad_norm": 4.221518039703369,
|
| 1020 |
+
"kl": 1.9362998560855262,
|
| 1021 |
+
"learning_rate": 1.1291055851370623e-07,
|
| 1022 |
+
"loss": 0.056,
|
| 1023 |
+
"reward": 0.582456154886045,
|
| 1024 |
+
"reward_std": 0.38160365223884585,
|
| 1025 |
+
"rewards/accuracy_reward": 0.582456154886045,
|
| 1026 |
+
"rewards/format_reward": 0.0,
|
| 1027 |
+
"step": 350
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"clip_ratio": 0.0,
|
| 1031 |
+
"completion_length": 589.6280856483861,
|
| 1032 |
+
"epoch": 0.8993333333333333,
|
| 1033 |
+
"grad_norm": 2.3517720699310303,
|
| 1034 |
+
"kl": 1.8255171926398026,
|
| 1035 |
+
"learning_rate": 8.894959999345015e-08,
|
| 1036 |
+
"loss": 0.0361,
|
| 1037 |
+
"reward": 0.5754386076801702,
|
| 1038 |
+
"reward_std": 0.34303222267251265,
|
| 1039 |
+
"rewards/accuracy_reward": 0.5754386076801702,
|
| 1040 |
+
"rewards/format_reward": 0.0,
|
| 1041 |
+
"step": 355
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"clip_ratio": 0.0,
|
| 1045 |
+
"completion_length": 629.0754549528423,
|
| 1046 |
+
"epoch": 0.912,
|
| 1047 |
+
"grad_norm": 1.3438383340835571,
|
| 1048 |
+
"kl": 1.5210680509868422,
|
| 1049 |
+
"learning_rate": 6.776646400696212e-08,
|
| 1050 |
+
"loss": 0.0234,
|
| 1051 |
+
"reward": 0.5789473809693989,
|
| 1052 |
+
"reward_std": 0.3563298034040551,
|
| 1053 |
+
"rewards/accuracy_reward": 0.5789473809693989,
|
| 1054 |
+
"rewards/format_reward": 0.0,
|
| 1055 |
+
"step": 360
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"clip_ratio": 0.0,
|
| 1059 |
+
"completion_length": 585.6263320119757,
|
| 1060 |
+
"epoch": 0.9246666666666666,
|
| 1061 |
+
"grad_norm": 1.6730248928070068,
|
| 1062 |
+
"kl": 3.8371864720394737,
|
| 1063 |
+
"learning_rate": 4.940285210684375e-08,
|
| 1064 |
+
"loss": 0.075,
|
| 1065 |
+
"reward": 0.6596491382310264,
|
| 1066 |
+
"reward_std": 0.3641158220015074,
|
| 1067 |
+
"rewards/accuracy_reward": 0.6596491382310264,
|
| 1068 |
+
"rewards/format_reward": 0.0,
|
| 1069 |
+
"step": 365
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"clip_ratio": 0.0,
|
| 1073 |
+
"completion_length": 634.8017714651007,
|
| 1074 |
+
"epoch": 0.9373333333333334,
|
| 1075 |
+
"grad_norm": 0.6605441570281982,
|
| 1076 |
+
"kl": 1.9925395764802631,
|
| 1077 |
+
"learning_rate": 3.389491527319999e-08,
|
| 1078 |
+
"loss": 0.0375,
|
| 1079 |
+
"reward": 0.6578947547234987,
|
| 1080 |
+
"reward_std": 0.41116404282419305,
|
| 1081 |
+
"rewards/accuracy_reward": 0.6578947547234987,
|
| 1082 |
+
"rewards/format_reward": 0.0,
|
| 1083 |
+
"step": 370
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"clip_ratio": 0.0,
|
| 1087 |
+
"completion_length": 600.2000179893092,
|
| 1088 |
+
"epoch": 0.95,
|
| 1089 |
+
"grad_norm": 1.5404088497161865,
|
| 1090 |
+
"kl": 1.4656384919819079,
|
| 1091 |
+
"learning_rate": 2.127318274608381e-08,
|
| 1092 |
+
"loss": 0.0059,
|
| 1093 |
+
"reward": 0.5789473817536706,
|
| 1094 |
+
"reward_std": 0.350234569060175,
|
| 1095 |
+
"rewards/accuracy_reward": 0.5789473817536706,
|
| 1096 |
+
"rewards/format_reward": 0.0,
|
| 1097 |
+
"step": 375
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"clip_ratio": 0.0,
|
| 1101 |
+
"completion_length": 573.8017702604595,
|
| 1102 |
+
"epoch": 0.9626666666666667,
|
| 1103 |
+
"grad_norm": 2.143493175506592,
|
| 1104 |
+
"kl": 1.3233176783511513,
|
| 1105 |
+
"learning_rate": 1.1562501925013125e-08,
|
| 1106 |
+
"loss": 0.0301,
|
| 1107 |
+
"reward": 0.6631579116771096,
|
| 1108 |
+
"reward_std": 0.3379683045964492,
|
| 1109 |
+
"rewards/accuracy_reward": 0.6631579116771096,
|
| 1110 |
+
"rewards/format_reward": 0.0,
|
| 1111 |
+
"step": 380
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"clip_ratio": 0.0,
|
| 1115 |
+
"completion_length": 616.8175617418791,
|
| 1116 |
+
"epoch": 0.9753333333333334,
|
| 1117 |
+
"grad_norm": 0.5864923596382141,
|
| 1118 |
+
"kl": 1.3502071982935855,
|
| 1119 |
+
"learning_rate": 4.781989453874814e-09,
|
| 1120 |
+
"loss": 0.0052,
|
| 1121 |
+
"reward": 0.5807017691825566,
|
| 1122 |
+
"reward_std": 0.3575416382990385,
|
| 1123 |
+
"rewards/accuracy_reward": 0.5807017691825566,
|
| 1124 |
+
"rewards/format_reward": 0.0,
|
| 1125 |
+
"step": 385
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"clip_ratio": 0.0,
|
| 1129 |
+
"completion_length": 612.1122984233656,
|
| 1130 |
+
"epoch": 0.988,
|
| 1131 |
+
"grad_norm": 0.3266775906085968,
|
| 1132 |
+
"kl": 1.6176237407483554,
|
| 1133 |
+
"learning_rate": 9.44993587509657e-10,
|
| 1134 |
+
"loss": 0.0364,
|
| 1135 |
+
"reward": 0.5947368560652984,
|
| 1136 |
+
"reward_std": 0.300396885683662,
|
| 1137 |
+
"rewards/accuracy_reward": 0.5947368560652984,
|
| 1138 |
+
"rewards/format_reward": 0.0,
|
| 1139 |
+
"step": 390
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"clip_ratio": 0.0,
|
| 1143 |
+
"completion_length": 612.8508915148283,
|
| 1144 |
+
"epoch": 0.9981333333333333,
|
| 1145 |
+
"kl": 1.1464434171977795,
|
| 1146 |
+
"reward": 0.603070187529451,
|
| 1147 |
+
"reward_std": 0.27440278663447026,
|
| 1148 |
+
"rewards/accuracy_reward": 0.603070187529451,
|
| 1149 |
"rewards/format_reward": 0.0,
|
| 1150 |
+
"step": 394,
|
| 1151 |
"total_flos": 0.0,
|
| 1152 |
+
"train_loss": 0.2665646580783036,
|
| 1153 |
+
"train_runtime": 484901.6937,
|
| 1154 |
+
"train_samples_per_second": 0.015,
|
| 1155 |
"train_steps_per_second": 0.001
|
| 1156 |
}
|
| 1157 |
],
|
| 1158 |
"logging_steps": 5,
|
| 1159 |
+
"max_steps": 394,
|
| 1160 |
"num_input_tokens_seen": 0,
|
| 1161 |
"num_train_epochs": 1,
|
| 1162 |
"save_steps": 500,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8056
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ae203872f9a89bc16deefba33d39589ccc25f8f69f4ffa2bb08e6a0560638856
|
| 3 |
size 8056
|