Model save
Browse files- README.md +1 -1
- adapter_model.safetensors +1 -1
- all_results.json +4 -4
- runs/May09_01-57-46_sky-3542-mykola-5e9d8a2f-head-4qgcm7zi-compute/events.out.tfevents.1746755944.sky-3542-mykola-5e9d8a2f-head-4qgcm7zi-compute.108697.0 +3 -0
- train_results.json +4 -4
- trainer_state.json +1061 -1061
- 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/polyagent/MamayLM-ORPO/runs/
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with ORPO, a method introduced in [ORPO: Monolithic Preference Optimization without Reference Model](https://huggingface.co/papers/2403.07691).
|
|
|
|
| 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/polyagent/MamayLM-ORPO/runs/phvtipy2)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with ORPO, a method introduced in [ORPO: Monolithic Preference Optimization without Reference Model](https://huggingface.co/papers/2403.07691).
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 286284472
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5e2ffefd7726f56b73abf0a62cfdb7dfc3ddb1348597013f6aef0a158ddcd843
|
| 3 |
size 286284472
|
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 1.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 12400,
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.0152948588709678,
|
| 5 |
+
"train_runtime": 1543.1546,
|
| 6 |
"train_samples": 12400,
|
| 7 |
+
"train_samples_per_second": 8.035,
|
| 8 |
+
"train_steps_per_second": 0.502
|
| 9 |
}
|
runs/May09_01-57-46_sky-3542-mykola-5e9d8a2f-head-4qgcm7zi-compute/events.out.tfevents.1746755944.sky-3542-mykola-5e9d8a2f-head-4qgcm7zi-compute.108697.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e6223107d3acc4f718f114398edf0b0f3f8d3e11db05f6bb063b761147a56d7f
|
| 3 |
+
size 72151
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 1.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 12400,
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.0152948588709678,
|
| 5 |
+
"train_runtime": 1543.1546,
|
| 6 |
"train_samples": 12400,
|
| 7 |
+
"train_samples_per_second": 8.035,
|
| 8 |
+
"train_steps_per_second": 0.502
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -11,1398 +11,1398 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.012903225806451613,
|
| 14 |
-
"grad_norm":
|
| 15 |
-
"learning_rate":
|
| 16 |
-
"log_odds_chosen": 1.
|
| 17 |
-
"log_odds_ratio": -0.
|
| 18 |
-
"logits/chosen": -
|
| 19 |
-
"logits/rejected": 15.
|
| 20 |
-
"logps/chosen": -1.
|
| 21 |
-
"logps/rejected": -2.
|
| 22 |
-
"loss":
|
| 23 |
-
"nll_loss": 1.
|
| 24 |
"rewards/accuracies": 0.949999988079071,
|
| 25 |
-
"rewards/chosen": -0.
|
| 26 |
-
"rewards/margins": 0.
|
| 27 |
-
"rewards/rejected": -0.
|
| 28 |
"step": 10
|
| 29 |
},
|
| 30 |
{
|
| 31 |
"epoch": 0.025806451612903226,
|
| 32 |
-
"grad_norm":
|
| 33 |
-
"learning_rate":
|
| 34 |
-
"log_odds_chosen":
|
| 35 |
-
"log_odds_ratio": -0.
|
| 36 |
-
"logits/chosen":
|
| 37 |
-
"logits/rejected":
|
| 38 |
-
"logps/chosen": -1.
|
| 39 |
-
"logps/rejected": -
|
| 40 |
-
"loss": 1.
|
| 41 |
-
"nll_loss": 1.
|
| 42 |
-
"rewards/accuracies": 0.
|
| 43 |
-
"rewards/chosen": -0.
|
| 44 |
-
"rewards/margins": 0.
|
| 45 |
-
"rewards/rejected": -0.
|
| 46 |
"step": 20
|
| 47 |
},
|
| 48 |
{
|
| 49 |
"epoch": 0.03870967741935484,
|
| 50 |
-
"grad_norm":
|
| 51 |
-
"learning_rate":
|
| 52 |
-
"log_odds_chosen": 0.
|
| 53 |
-
"log_odds_ratio": -0.
|
| 54 |
-
"logits/chosen":
|
| 55 |
-
"logits/rejected":
|
| 56 |
-
"logps/chosen": -1.
|
| 57 |
-
"logps/rejected": -
|
| 58 |
-
"loss": 1.
|
| 59 |
-
"nll_loss": 1.
|
| 60 |
-
"rewards/accuracies": 0.
|
| 61 |
-
"rewards/chosen": -0.
|
| 62 |
-
"rewards/margins": 0.
|
| 63 |
-
"rewards/rejected": -0.
|
| 64 |
"step": 30
|
| 65 |
},
|
| 66 |
{
|
| 67 |
"epoch": 0.05161290322580645,
|
| 68 |
-
"grad_norm":
|
| 69 |
-
"learning_rate":
|
| 70 |
-
"log_odds_chosen": 0.
|
| 71 |
-
"log_odds_ratio": -0.
|
| 72 |
-
"logits/chosen": 0.
|
| 73 |
-
"logits/rejected": 17.
|
| 74 |
-
"logps/chosen": -1.
|
| 75 |
-
"logps/rejected": -1.
|
| 76 |
-
"loss": 1.
|
| 77 |
-
"nll_loss": 1.
|
| 78 |
-
"rewards/accuracies": 0.
|
| 79 |
-
"rewards/chosen": -0.
|
| 80 |
-
"rewards/margins": 0.
|
| 81 |
-
"rewards/rejected": -0.
|
| 82 |
"step": 40
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 0.06451612903225806,
|
| 86 |
-
"grad_norm":
|
| 87 |
-
"learning_rate":
|
| 88 |
-
"log_odds_chosen": 0.
|
| 89 |
-
"log_odds_ratio": -0.
|
| 90 |
-
"logits/chosen":
|
| 91 |
-
"logits/rejected":
|
| 92 |
-
"logps/chosen": -1.
|
| 93 |
-
"logps/rejected": -1.
|
| 94 |
-
"loss": 1.
|
| 95 |
-
"nll_loss": 1.
|
| 96 |
-
"rewards/accuracies": 0.
|
| 97 |
-
"rewards/chosen": -0.
|
| 98 |
-
"rewards/margins": 0.
|
| 99 |
-
"rewards/rejected": -0.
|
| 100 |
"step": 50
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.07741935483870968,
|
| 104 |
-
"grad_norm": 11.
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"log_odds_chosen": 0.
|
| 107 |
-
"log_odds_ratio": -0.
|
| 108 |
-
"logits/chosen":
|
| 109 |
-
"logits/rejected":
|
| 110 |
-
"logps/chosen": -1.
|
| 111 |
-
"logps/rejected": -1.
|
| 112 |
-
"loss": 1.
|
| 113 |
-
"nll_loss": 1.
|
| 114 |
-
"rewards/accuracies": 0.
|
| 115 |
-
"rewards/chosen": -0.
|
| 116 |
-
"rewards/margins": 0.
|
| 117 |
-
"rewards/rejected": -0.
|
| 118 |
"step": 60
|
| 119 |
},
|
| 120 |
{
|
| 121 |
"epoch": 0.09032258064516129,
|
| 122 |
-
"grad_norm":
|
| 123 |
-
"learning_rate":
|
| 124 |
-
"log_odds_chosen":
|
| 125 |
-
"log_odds_ratio": -0.
|
| 126 |
-
"logits/chosen":
|
| 127 |
-
"logits/rejected":
|
| 128 |
-
"logps/chosen": -1.
|
| 129 |
-
"logps/rejected": -
|
| 130 |
-
"loss": 1.
|
| 131 |
-
"nll_loss": 1.
|
| 132 |
-
"rewards/accuracies": 0.
|
| 133 |
-
"rewards/chosen": -0.
|
| 134 |
-
"rewards/margins": 0.
|
| 135 |
-
"rewards/rejected": -0.
|
| 136 |
"step": 70
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.1032258064516129,
|
| 140 |
-
"grad_norm":
|
| 141 |
-
"learning_rate":
|
| 142 |
-
"log_odds_chosen":
|
| 143 |
-
"log_odds_ratio": -0.
|
| 144 |
-
"logits/chosen": -
|
| 145 |
-
"logits/rejected":
|
| 146 |
-
"logps/chosen": -1.
|
| 147 |
-
"logps/rejected": -
|
| 148 |
-
"loss": 1.
|
| 149 |
-
"nll_loss": 1.
|
| 150 |
-
"rewards/accuracies": 0.
|
| 151 |
-
"rewards/chosen": -0.
|
| 152 |
-
"rewards/margins": 0.
|
| 153 |
-
"rewards/rejected": -0.
|
| 154 |
"step": 80
|
| 155 |
},
|
| 156 |
{
|
| 157 |
"epoch": 0.11612903225806452,
|
| 158 |
-
"grad_norm":
|
| 159 |
-
"learning_rate":
|
| 160 |
-
"log_odds_chosen":
|
| 161 |
-
"log_odds_ratio": -0.
|
| 162 |
-
"logits/chosen": -
|
| 163 |
-
"logits/rejected":
|
| 164 |
-
"logps/chosen": -1.
|
| 165 |
-
"logps/rejected": -
|
| 166 |
-
"loss": 1.
|
| 167 |
-
"nll_loss": 1.
|
| 168 |
-
"rewards/accuracies": 0.
|
| 169 |
-
"rewards/chosen": -0.
|
| 170 |
-
"rewards/margins": 0.
|
| 171 |
-
"rewards/rejected": -0.
|
| 172 |
"step": 90
|
| 173 |
},
|
| 174 |
{
|
| 175 |
"epoch": 0.12903225806451613,
|
| 176 |
-
"grad_norm":
|
| 177 |
-
"learning_rate":
|
| 178 |
-
"log_odds_chosen":
|
| 179 |
-
"log_odds_ratio": -0.
|
| 180 |
-
"logits/chosen": -1.
|
| 181 |
-
"logits/rejected": 15.
|
| 182 |
-
"logps/chosen": -1.
|
| 183 |
-
"logps/rejected": -
|
| 184 |
-
"loss": 1.
|
| 185 |
-
"nll_loss": 1.
|
| 186 |
-
"rewards/accuracies": 0.
|
| 187 |
-
"rewards/chosen": -0.
|
| 188 |
-
"rewards/margins": 0.
|
| 189 |
-
"rewards/rejected": -0.
|
| 190 |
"step": 100
|
| 191 |
},
|
| 192 |
{
|
| 193 |
"epoch": 0.14193548387096774,
|
| 194 |
-
"grad_norm":
|
| 195 |
-
"learning_rate":
|
| 196 |
-
"log_odds_chosen":
|
| 197 |
-
"log_odds_ratio": -0.
|
| 198 |
-
"logits/chosen": -
|
| 199 |
-
"logits/rejected":
|
| 200 |
-
"logps/chosen": -1.
|
| 201 |
-
"logps/rejected": -
|
| 202 |
-
"loss": 1.
|
| 203 |
-
"nll_loss": 1.
|
| 204 |
-
"rewards/accuracies": 0.
|
| 205 |
-
"rewards/chosen": -0.
|
| 206 |
-
"rewards/margins": 0.
|
| 207 |
-
"rewards/rejected": -0.
|
| 208 |
"step": 110
|
| 209 |
},
|
| 210 |
{
|
| 211 |
"epoch": 0.15483870967741936,
|
| 212 |
-
"grad_norm":
|
| 213 |
-
"learning_rate":
|
| 214 |
-
"log_odds_chosen":
|
| 215 |
-
"log_odds_ratio": -0.
|
| 216 |
-
"logits/chosen": -
|
| 217 |
-
"logits/rejected": 16.
|
| 218 |
-
"logps/chosen": -1.
|
| 219 |
-
"logps/rejected": -
|
| 220 |
-
"loss": 1.
|
| 221 |
-
"nll_loss": 1.
|
| 222 |
-
"rewards/accuracies": 0.
|
| 223 |
-
"rewards/chosen": -0.
|
| 224 |
-
"rewards/margins": 0.
|
| 225 |
-
"rewards/rejected": -0.
|
| 226 |
"step": 120
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.16774193548387098,
|
| 230 |
-
"grad_norm":
|
| 231 |
-
"learning_rate":
|
| 232 |
-
"log_odds_chosen":
|
| 233 |
-
"log_odds_ratio": -0.
|
| 234 |
-
"logits/chosen": -
|
| 235 |
-
"logits/rejected":
|
| 236 |
-
"logps/chosen": -1.
|
| 237 |
-
"logps/rejected": -
|
| 238 |
-
"loss": 1.
|
| 239 |
-
"nll_loss": 1.
|
| 240 |
-
"rewards/accuracies": 0.
|
| 241 |
-
"rewards/chosen": -0.
|
| 242 |
-
"rewards/margins": 0.
|
| 243 |
-
"rewards/rejected": -0.
|
| 244 |
"step": 130
|
| 245 |
},
|
| 246 |
{
|
| 247 |
"epoch": 0.18064516129032257,
|
| 248 |
-
"grad_norm":
|
| 249 |
-
"learning_rate":
|
| 250 |
-
"log_odds_chosen":
|
| 251 |
-
"log_odds_ratio": -0.
|
| 252 |
-
"logits/chosen": -1.
|
| 253 |
-
"logits/rejected": 15.
|
| 254 |
-
"logps/chosen": -1.
|
| 255 |
-
"logps/rejected": -
|
| 256 |
-
"loss": 1.
|
| 257 |
-
"nll_loss": 1.
|
| 258 |
-
"rewards/accuracies": 0.
|
| 259 |
-
"rewards/chosen": -0.
|
| 260 |
-
"rewards/margins": 0.
|
| 261 |
-
"rewards/rejected": -0.
|
| 262 |
"step": 140
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.1935483870967742,
|
| 266 |
-
"grad_norm":
|
| 267 |
-
"learning_rate":
|
| 268 |
-
"log_odds_chosen":
|
| 269 |
-
"log_odds_ratio": -0.
|
| 270 |
-
"logits/chosen": -
|
| 271 |
-
"logits/rejected": 15.
|
| 272 |
-
"logps/chosen": -1.
|
| 273 |
-
"logps/rejected": -
|
| 274 |
-
"loss": 1.
|
| 275 |
-
"nll_loss": 1.
|
| 276 |
"rewards/accuracies": 0.981249988079071,
|
| 277 |
-
"rewards/chosen": -0.
|
| 278 |
-
"rewards/margins": 0.
|
| 279 |
-
"rewards/rejected": -0.
|
| 280 |
"step": 150
|
| 281 |
},
|
| 282 |
{
|
| 283 |
"epoch": 0.2064516129032258,
|
| 284 |
-
"grad_norm":
|
| 285 |
-
"learning_rate":
|
| 286 |
-
"log_odds_chosen":
|
| 287 |
-
"log_odds_ratio": -0.
|
| 288 |
-
"logits/chosen": -2.
|
| 289 |
-
"logits/rejected":
|
| 290 |
-
"logps/chosen": -
|
| 291 |
-
"logps/rejected": -
|
| 292 |
-
"loss": 1.
|
| 293 |
-
"nll_loss": 1.
|
| 294 |
"rewards/accuracies": 0.9750000238418579,
|
| 295 |
-
"rewards/chosen": -0.
|
| 296 |
-
"rewards/margins": 0.
|
| 297 |
-
"rewards/rejected": -0.
|
| 298 |
"step": 160
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.21935483870967742,
|
| 302 |
-
"grad_norm":
|
| 303 |
-
"learning_rate":
|
| 304 |
-
"log_odds_chosen":
|
| 305 |
-
"log_odds_ratio": -0.
|
| 306 |
-
"logits/chosen": -
|
| 307 |
-
"logits/rejected":
|
| 308 |
-
"logps/chosen": -0.
|
| 309 |
-
"logps/rejected": -
|
| 310 |
-
"loss":
|
| 311 |
-
"nll_loss":
|
| 312 |
-
"rewards/accuracies": 0.
|
| 313 |
-
"rewards/chosen": -0.
|
| 314 |
-
"rewards/margins": 0.
|
| 315 |
-
"rewards/rejected": -0.
|
| 316 |
"step": 170
|
| 317 |
},
|
| 318 |
{
|
| 319 |
"epoch": 0.23225806451612904,
|
| 320 |
-
"grad_norm":
|
| 321 |
-
"learning_rate":
|
| 322 |
-
"log_odds_chosen":
|
| 323 |
-
"log_odds_ratio": -0.
|
| 324 |
-
"logits/chosen": -
|
| 325 |
-
"logits/rejected": 15.
|
| 326 |
-
"logps/chosen": -
|
| 327 |
-
"logps/rejected": -
|
| 328 |
-
"loss":
|
| 329 |
-
"nll_loss":
|
| 330 |
-
"rewards/accuracies": 0.
|
| 331 |
-
"rewards/chosen": -0.
|
| 332 |
-
"rewards/margins": 0.
|
| 333 |
-
"rewards/rejected": -0.
|
| 334 |
"step": 180
|
| 335 |
},
|
| 336 |
{
|
| 337 |
"epoch": 0.24516129032258063,
|
| 338 |
-
"grad_norm":
|
| 339 |
-
"learning_rate":
|
| 340 |
-
"log_odds_chosen":
|
| 341 |
-
"log_odds_ratio": -0.
|
| 342 |
-
"logits/chosen": -1.
|
| 343 |
-
"logits/rejected":
|
| 344 |
-
"logps/chosen": -
|
| 345 |
-
"logps/rejected": -
|
| 346 |
-
"loss": 1.
|
| 347 |
-
"nll_loss":
|
| 348 |
-
"rewards/accuracies": 0.
|
| 349 |
-
"rewards/chosen": -0.
|
| 350 |
-
"rewards/margins": 0.
|
| 351 |
-
"rewards/rejected": -0.
|
| 352 |
"step": 190
|
| 353 |
},
|
| 354 |
{
|
| 355 |
"epoch": 0.25806451612903225,
|
| 356 |
-
"grad_norm":
|
| 357 |
-
"learning_rate":
|
| 358 |
-
"log_odds_chosen":
|
| 359 |
-
"log_odds_ratio": -0.
|
| 360 |
-
"logits/chosen": -
|
| 361 |
-
"logits/rejected":
|
| 362 |
-
"logps/chosen": -1.
|
| 363 |
-
"logps/rejected": -
|
| 364 |
-
"loss": 1.
|
| 365 |
-
"nll_loss": 1.
|
| 366 |
-
"rewards/accuracies": 0.
|
| 367 |
-
"rewards/chosen": -0.
|
| 368 |
-
"rewards/margins": 0.
|
| 369 |
-
"rewards/rejected": -0.
|
| 370 |
"step": 200
|
| 371 |
},
|
| 372 |
{
|
| 373 |
"epoch": 0.2709677419354839,
|
| 374 |
-
"grad_norm":
|
| 375 |
-
"learning_rate":
|
| 376 |
-
"log_odds_chosen":
|
| 377 |
-
"log_odds_ratio": -0.
|
| 378 |
-
"logits/chosen": -2.
|
| 379 |
-
"logits/rejected": 15.
|
| 380 |
-
"logps/chosen": -
|
| 381 |
-
"logps/rejected": -
|
| 382 |
-
"loss":
|
| 383 |
-
"nll_loss":
|
| 384 |
-
"rewards/accuracies": 0.
|
| 385 |
-
"rewards/chosen": -0.
|
| 386 |
-
"rewards/margins": 0.
|
| 387 |
-
"rewards/rejected": -0.
|
| 388 |
"step": 210
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.2838709677419355,
|
| 392 |
-
"grad_norm":
|
| 393 |
-
"learning_rate":
|
| 394 |
-
"log_odds_chosen": 2.
|
| 395 |
-
"log_odds_ratio": -0.
|
| 396 |
-
"logits/chosen": -1.
|
| 397 |
-
"logits/rejected": 15.
|
| 398 |
-
"logps/chosen": -1.
|
| 399 |
-
"logps/rejected": -
|
| 400 |
-
"loss": 1.
|
| 401 |
-
"nll_loss": 1.
|
| 402 |
-
"rewards/accuracies": 0.
|
| 403 |
-
"rewards/chosen": -0.
|
| 404 |
-
"rewards/margins": 0.
|
| 405 |
-
"rewards/rejected": -0.
|
| 406 |
"step": 220
|
| 407 |
},
|
| 408 |
{
|
| 409 |
"epoch": 0.2967741935483871,
|
| 410 |
-
"grad_norm":
|
| 411 |
-
"learning_rate":
|
| 412 |
-
"log_odds_chosen":
|
| 413 |
-
"log_odds_ratio": -0.
|
| 414 |
-
"logits/chosen": -
|
| 415 |
-
"logits/rejected": 15.
|
| 416 |
-
"logps/chosen": -1.
|
| 417 |
-
"logps/rejected": -
|
| 418 |
-
"loss": 1.
|
| 419 |
-
"nll_loss": 1.
|
| 420 |
-
"rewards/accuracies": 0.
|
| 421 |
-
"rewards/chosen": -0.
|
| 422 |
-
"rewards/margins": 0.
|
| 423 |
-
"rewards/rejected": -0.
|
| 424 |
"step": 230
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 0.3096774193548387,
|
| 428 |
-
"grad_norm":
|
| 429 |
-
"learning_rate":
|
| 430 |
-
"log_odds_chosen": 2.
|
| 431 |
-
"log_odds_ratio": -0.
|
| 432 |
-
"logits/chosen": -2.
|
| 433 |
-
"logits/rejected": 14.
|
| 434 |
-
"logps/chosen": -
|
| 435 |
-
"logps/rejected": -
|
| 436 |
-
"loss": 1.
|
| 437 |
-
"nll_loss": 1.
|
| 438 |
-
"rewards/accuracies": 0.
|
| 439 |
-
"rewards/chosen": -0.
|
| 440 |
-
"rewards/margins": 0.
|
| 441 |
-
"rewards/rejected": -0.
|
| 442 |
"step": 240
|
| 443 |
},
|
| 444 |
{
|
| 445 |
"epoch": 0.3225806451612903,
|
| 446 |
-
"grad_norm":
|
| 447 |
-
"learning_rate":
|
| 448 |
-
"log_odds_chosen":
|
| 449 |
-
"log_odds_ratio": -0.
|
| 450 |
-
"logits/chosen": -
|
| 451 |
-
"logits/rejected": 15.
|
| 452 |
-
"logps/chosen": -0.
|
| 453 |
-
"logps/rejected": -3.
|
| 454 |
-
"loss":
|
| 455 |
-
"nll_loss":
|
| 456 |
"rewards/accuracies": 0.987500011920929,
|
| 457 |
-
"rewards/chosen": -0.
|
| 458 |
-
"rewards/margins": 0.
|
| 459 |
-
"rewards/rejected": -0.
|
| 460 |
"step": 250
|
| 461 |
},
|
| 462 |
{
|
| 463 |
"epoch": 0.33548387096774196,
|
| 464 |
-
"grad_norm":
|
| 465 |
-
"learning_rate":
|
| 466 |
-
"log_odds_chosen": 2.
|
| 467 |
-
"log_odds_ratio": -0.
|
| 468 |
-
"logits/chosen": -2.
|
| 469 |
-
"logits/rejected": 15.
|
| 470 |
-
"logps/chosen": -
|
| 471 |
-
"logps/rejected": -
|
| 472 |
-
"loss":
|
| 473 |
-
"nll_loss":
|
| 474 |
-
"rewards/accuracies": 0.
|
| 475 |
-
"rewards/chosen": -0.
|
| 476 |
-
"rewards/margins": 0.
|
| 477 |
-
"rewards/rejected": -0.
|
| 478 |
"step": 260
|
| 479 |
},
|
| 480 |
{
|
| 481 |
"epoch": 0.34838709677419355,
|
| 482 |
-
"grad_norm":
|
| 483 |
-
"learning_rate":
|
| 484 |
-
"log_odds_chosen":
|
| 485 |
-
"log_odds_ratio": -0.
|
| 486 |
-
"logits/chosen": -
|
| 487 |
-
"logits/rejected":
|
| 488 |
-
"logps/chosen": -0.
|
| 489 |
-
"logps/rejected": -
|
| 490 |
-
"loss":
|
| 491 |
-
"nll_loss":
|
| 492 |
-
"rewards/accuracies": 0
|
| 493 |
-
"rewards/chosen": -0.
|
| 494 |
-
"rewards/margins": 0.
|
| 495 |
-
"rewards/rejected": -0.
|
| 496 |
"step": 270
|
| 497 |
},
|
| 498 |
{
|
| 499 |
"epoch": 0.36129032258064514,
|
| 500 |
-
"grad_norm":
|
| 501 |
-
"learning_rate":
|
| 502 |
-
"log_odds_chosen":
|
| 503 |
-
"log_odds_ratio": -0.
|
| 504 |
-
"logits/chosen": -
|
| 505 |
-
"logits/rejected": 15.
|
| 506 |
-
"logps/chosen": -
|
| 507 |
-
"logps/rejected": -
|
| 508 |
-
"loss":
|
| 509 |
-
"nll_loss":
|
| 510 |
"rewards/accuracies": 0.981249988079071,
|
| 511 |
-
"rewards/chosen": -0.
|
| 512 |
-
"rewards/margins": 0.
|
| 513 |
-
"rewards/rejected": -0.
|
| 514 |
"step": 280
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 0.3741935483870968,
|
| 518 |
-
"grad_norm":
|
| 519 |
-
"learning_rate":
|
| 520 |
-
"log_odds_chosen":
|
| 521 |
-
"log_odds_ratio": -0.
|
| 522 |
-
"logits/chosen": -
|
| 523 |
-
"logits/rejected": 15.
|
| 524 |
-
"logps/chosen": -
|
| 525 |
-
"logps/rejected": -
|
| 526 |
-
"loss":
|
| 527 |
-
"nll_loss":
|
| 528 |
"rewards/accuracies": 0.981249988079071,
|
| 529 |
-
"rewards/chosen": -0.
|
| 530 |
-
"rewards/margins": 0.
|
| 531 |
-
"rewards/rejected": -0.
|
| 532 |
"step": 290
|
| 533 |
},
|
| 534 |
{
|
| 535 |
"epoch": 0.3870967741935484,
|
| 536 |
-
"grad_norm":
|
| 537 |
-
"learning_rate":
|
| 538 |
-
"log_odds_chosen":
|
| 539 |
-
"log_odds_ratio": -0.
|
| 540 |
-
"logits/chosen": -2.
|
| 541 |
-
"logits/rejected": 15.
|
| 542 |
-
"logps/chosen": -
|
| 543 |
-
"logps/rejected": -
|
| 544 |
-
"loss":
|
| 545 |
-
"nll_loss":
|
| 546 |
-
"rewards/accuracies": 0.
|
| 547 |
-
"rewards/chosen": -0.
|
| 548 |
-
"rewards/margins": 0.
|
| 549 |
-
"rewards/rejected": -0.
|
| 550 |
"step": 300
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 0.4,
|
| 554 |
-
"grad_norm":
|
| 555 |
-
"learning_rate":
|
| 556 |
-
"log_odds_chosen":
|
| 557 |
-
"log_odds_ratio": -0.
|
| 558 |
-
"logits/chosen": -
|
| 559 |
-
"logits/rejected":
|
| 560 |
-
"logps/chosen": -
|
| 561 |
-
"logps/rejected": -
|
| 562 |
-
"loss":
|
| 563 |
-
"nll_loss":
|
| 564 |
-
"rewards/accuracies": 0.
|
| 565 |
-
"rewards/chosen": -0.
|
| 566 |
-
"rewards/margins": 0.
|
| 567 |
-
"rewards/rejected": -0.
|
| 568 |
"step": 310
|
| 569 |
},
|
| 570 |
{
|
| 571 |
"epoch": 0.4129032258064516,
|
| 572 |
-
"grad_norm":
|
| 573 |
-
"learning_rate":
|
| 574 |
-
"log_odds_chosen":
|
| 575 |
-
"log_odds_ratio": -0.
|
| 576 |
-
"logits/chosen": -
|
| 577 |
-
"logits/rejected": 15.
|
| 578 |
-
"logps/chosen": -
|
| 579 |
-
"logps/rejected": -
|
| 580 |
-
"loss":
|
| 581 |
-
"nll_loss":
|
| 582 |
-
"rewards/accuracies": 0.
|
| 583 |
-
"rewards/chosen": -0.
|
| 584 |
-
"rewards/margins": 0.
|
| 585 |
-
"rewards/rejected": -0.
|
| 586 |
"step": 320
|
| 587 |
},
|
| 588 |
{
|
| 589 |
"epoch": 0.4258064516129032,
|
| 590 |
-
"grad_norm":
|
| 591 |
-
"learning_rate":
|
| 592 |
-
"log_odds_chosen":
|
| 593 |
-
"log_odds_ratio": -0.
|
| 594 |
-
"logits/chosen": -2.
|
| 595 |
-
"logits/rejected": 15.
|
| 596 |
-
"logps/chosen": -
|
| 597 |
-
"logps/rejected": -
|
| 598 |
-
"loss": 1.
|
| 599 |
-
"nll_loss": 1.
|
| 600 |
-
"rewards/accuracies": 0.
|
| 601 |
-
"rewards/chosen": -0.
|
| 602 |
-
"rewards/margins": 0.
|
| 603 |
-
"rewards/rejected": -0.
|
| 604 |
"step": 330
|
| 605 |
},
|
| 606 |
{
|
| 607 |
"epoch": 0.43870967741935485,
|
| 608 |
-
"grad_norm":
|
| 609 |
-
"learning_rate":
|
| 610 |
-
"log_odds_chosen":
|
| 611 |
-
"log_odds_ratio": -0.
|
| 612 |
-
"logits/chosen": -
|
| 613 |
-
"logits/rejected": 15.
|
| 614 |
-
"logps/chosen": -
|
| 615 |
-
"logps/rejected": -
|
| 616 |
-
"loss":
|
| 617 |
-
"nll_loss":
|
| 618 |
-
"rewards/accuracies": 0
|
| 619 |
-
"rewards/chosen": -0.
|
| 620 |
-
"rewards/margins": 0.
|
| 621 |
-
"rewards/rejected": -0.
|
| 622 |
"step": 340
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 0.45161290322580644,
|
| 626 |
-
"grad_norm":
|
| 627 |
-
"learning_rate":
|
| 628 |
-
"log_odds_chosen":
|
| 629 |
-
"log_odds_ratio": -0.
|
| 630 |
-
"logits/chosen": -2.
|
| 631 |
-
"logits/rejected": 15.
|
| 632 |
-
"logps/chosen": -0.
|
| 633 |
-
"logps/rejected": -
|
| 634 |
-
"loss":
|
| 635 |
-
"nll_loss":
|
| 636 |
"rewards/accuracies": 0.9937499761581421,
|
| 637 |
-
"rewards/chosen": -0.
|
| 638 |
-
"rewards/margins": 0.
|
| 639 |
-
"rewards/rejected": -0.
|
| 640 |
"step": 350
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 0.4645161290322581,
|
| 644 |
-
"grad_norm":
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"log_odds_chosen":
|
| 647 |
-
"log_odds_ratio": -0.
|
| 648 |
-
"logits/chosen": -2.
|
| 649 |
-
"logits/rejected": 15.
|
| 650 |
-
"logps/chosen": -
|
| 651 |
-
"logps/rejected": -
|
| 652 |
-
"loss":
|
| 653 |
-
"nll_loss":
|
| 654 |
-
"rewards/accuracies": 0.
|
| 655 |
-
"rewards/chosen": -0.
|
| 656 |
-
"rewards/margins": 0.
|
| 657 |
-
"rewards/rejected": -0.
|
| 658 |
"step": 360
|
| 659 |
},
|
| 660 |
{
|
| 661 |
"epoch": 0.4774193548387097,
|
| 662 |
-
"grad_norm":
|
| 663 |
-
"learning_rate":
|
| 664 |
-
"log_odds_chosen":
|
| 665 |
-
"log_odds_ratio": -0.
|
| 666 |
-
"logits/chosen": -
|
| 667 |
-
"logits/rejected": 15.
|
| 668 |
-
"logps/chosen": -
|
| 669 |
-
"logps/rejected": -
|
| 670 |
-
"loss":
|
| 671 |
-
"nll_loss":
|
| 672 |
-
"rewards/accuracies": 0.
|
| 673 |
-
"rewards/chosen": -0.
|
| 674 |
-
"rewards/margins": 0.
|
| 675 |
-
"rewards/rejected": -0.
|
| 676 |
"step": 370
|
| 677 |
},
|
| 678 |
{
|
| 679 |
"epoch": 0.49032258064516127,
|
| 680 |
-
"grad_norm":
|
| 681 |
-
"learning_rate":
|
| 682 |
-
"log_odds_chosen":
|
| 683 |
-
"log_odds_ratio": -0.
|
| 684 |
-
"logits/chosen": -2.
|
| 685 |
-
"logits/rejected": 15.
|
| 686 |
-
"logps/chosen": -0.
|
| 687 |
-
"logps/rejected": -
|
| 688 |
-
"loss":
|
| 689 |
-
"nll_loss": 0.
|
| 690 |
"rewards/accuracies": 0.9750000238418579,
|
| 691 |
-
"rewards/chosen": -0.
|
| 692 |
-
"rewards/margins": 0.
|
| 693 |
-
"rewards/rejected": -0.
|
| 694 |
"step": 380
|
| 695 |
},
|
| 696 |
{
|
| 697 |
"epoch": 0.5032258064516129,
|
| 698 |
-
"grad_norm":
|
| 699 |
-
"learning_rate":
|
| 700 |
-
"log_odds_chosen":
|
| 701 |
-
"log_odds_ratio": -0.
|
| 702 |
-
"logits/chosen": -2.
|
| 703 |
-
"logits/rejected": 15.
|
| 704 |
-
"logps/chosen": -
|
| 705 |
-
"logps/rejected": -
|
| 706 |
-
"loss":
|
| 707 |
-
"nll_loss":
|
| 708 |
"rewards/accuracies": 0.96875,
|
| 709 |
-
"rewards/chosen": -0.
|
| 710 |
-
"rewards/margins": 0.
|
| 711 |
-
"rewards/rejected": -0.
|
| 712 |
"step": 390
|
| 713 |
},
|
| 714 |
{
|
| 715 |
"epoch": 0.5161290322580645,
|
| 716 |
-
"grad_norm":
|
| 717 |
-
"learning_rate":
|
| 718 |
-
"log_odds_chosen":
|
| 719 |
-
"log_odds_ratio": -0.
|
| 720 |
-
"logits/chosen": -
|
| 721 |
-
"logits/rejected": 15.
|
| 722 |
-
"logps/chosen": -0.
|
| 723 |
-
"logps/rejected": -
|
| 724 |
-
"loss":
|
| 725 |
-
"nll_loss":
|
| 726 |
-
"rewards/accuracies": 0.
|
| 727 |
-
"rewards/chosen": -0.
|
| 728 |
-
"rewards/margins": 0.
|
| 729 |
-
"rewards/rejected": -0.
|
| 730 |
"step": 400
|
| 731 |
},
|
| 732 |
{
|
| 733 |
"epoch": 0.5290322580645161,
|
| 734 |
-
"grad_norm":
|
| 735 |
-
"learning_rate":
|
| 736 |
-
"log_odds_chosen":
|
| 737 |
-
"log_odds_ratio": -0.
|
| 738 |
-
"logits/chosen": -2.
|
| 739 |
-
"logits/rejected": 15.
|
| 740 |
-
"logps/chosen": -0.
|
| 741 |
-
"logps/rejected": -
|
| 742 |
-
"loss":
|
| 743 |
-
"nll_loss":
|
| 744 |
-
"rewards/accuracies": 0
|
| 745 |
-
"rewards/chosen": -0.
|
| 746 |
-
"rewards/margins": 0.
|
| 747 |
-
"rewards/rejected": -0.
|
| 748 |
"step": 410
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 0.5419354838709678,
|
| 752 |
-
"grad_norm":
|
| 753 |
-
"learning_rate":
|
| 754 |
-
"log_odds_chosen":
|
| 755 |
-
"log_odds_ratio": -0.
|
| 756 |
-
"logits/chosen": -
|
| 757 |
-
"logits/rejected":
|
| 758 |
-
"logps/chosen": -
|
| 759 |
-
"logps/rejected": -
|
| 760 |
-
"loss":
|
| 761 |
-
"nll_loss":
|
| 762 |
-
"rewards/accuracies": 0.
|
| 763 |
-
"rewards/chosen": -0.
|
| 764 |
-
"rewards/margins": 0.
|
| 765 |
-
"rewards/rejected": -0.
|
| 766 |
"step": 420
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 0.5548387096774193,
|
| 770 |
-
"grad_norm":
|
| 771 |
-
"learning_rate":
|
| 772 |
-
"log_odds_chosen":
|
| 773 |
-
"log_odds_ratio": -0.
|
| 774 |
-
"logits/chosen": -1.
|
| 775 |
-
"logits/rejected":
|
| 776 |
-
"logps/chosen": -
|
| 777 |
-
"logps/rejected": -
|
| 778 |
-
"loss":
|
| 779 |
-
"nll_loss":
|
| 780 |
-
"rewards/accuracies": 0.
|
| 781 |
-
"rewards/chosen": -0.
|
| 782 |
-
"rewards/margins": 0.
|
| 783 |
-
"rewards/rejected": -0.
|
| 784 |
"step": 430
|
| 785 |
},
|
| 786 |
{
|
| 787 |
"epoch": 0.567741935483871,
|
| 788 |
-
"grad_norm":
|
| 789 |
-
"learning_rate":
|
| 790 |
-
"log_odds_chosen":
|
| 791 |
-
"log_odds_ratio": -0.
|
| 792 |
-
"logits/chosen": -
|
| 793 |
-
"logits/rejected": 15.
|
| 794 |
-
"logps/chosen": -
|
| 795 |
-
"logps/rejected": -
|
| 796 |
-
"loss":
|
| 797 |
-
"nll_loss":
|
| 798 |
"rewards/accuracies": 0.9937499761581421,
|
| 799 |
-
"rewards/chosen": -0.
|
| 800 |
-
"rewards/margins": 0.
|
| 801 |
-
"rewards/rejected": -0.
|
| 802 |
"step": 440
|
| 803 |
},
|
| 804 |
{
|
| 805 |
"epoch": 0.5806451612903226,
|
| 806 |
-
"grad_norm":
|
| 807 |
-
"learning_rate":
|
| 808 |
-
"log_odds_chosen":
|
| 809 |
-
"log_odds_ratio": -0.
|
| 810 |
-
"logits/chosen": -2.
|
| 811 |
-
"logits/rejected":
|
| 812 |
-
"logps/chosen": -
|
| 813 |
-
"logps/rejected": -
|
| 814 |
-
"loss": 1.
|
| 815 |
-
"nll_loss":
|
| 816 |
-
"rewards/accuracies": 0.
|
| 817 |
-
"rewards/chosen": -0.
|
| 818 |
-
"rewards/margins": 0.
|
| 819 |
-
"rewards/rejected": -0.
|
| 820 |
"step": 450
|
| 821 |
},
|
| 822 |
{
|
| 823 |
"epoch": 0.5935483870967742,
|
| 824 |
-
"grad_norm":
|
| 825 |
-
"learning_rate":
|
| 826 |
-
"log_odds_chosen":
|
| 827 |
-
"log_odds_ratio": -0.
|
| 828 |
-
"logits/chosen": -2.
|
| 829 |
-
"logits/rejected": 15.
|
| 830 |
-
"logps/chosen": -0.
|
| 831 |
-
"logps/rejected": -
|
| 832 |
-
"loss":
|
| 833 |
-
"nll_loss":
|
| 834 |
-
"rewards/accuracies": 0.
|
| 835 |
-
"rewards/chosen": -0.
|
| 836 |
-
"rewards/margins": 0.
|
| 837 |
-
"rewards/rejected": -0.
|
| 838 |
"step": 460
|
| 839 |
},
|
| 840 |
{
|
| 841 |
"epoch": 0.6064516129032258,
|
| 842 |
-
"grad_norm":
|
| 843 |
-
"learning_rate":
|
| 844 |
-
"log_odds_chosen":
|
| 845 |
-
"log_odds_ratio": -0.
|
| 846 |
-
"logits/chosen": -
|
| 847 |
-
"logits/rejected": 14.
|
| 848 |
-
"logps/chosen": -
|
| 849 |
-
"logps/rejected": -
|
| 850 |
-
"loss":
|
| 851 |
-
"nll_loss":
|
| 852 |
-
"rewards/accuracies": 0.
|
| 853 |
-
"rewards/chosen": -0.
|
| 854 |
-
"rewards/margins": 0.
|
| 855 |
-
"rewards/rejected": -0.
|
| 856 |
"step": 470
|
| 857 |
},
|
| 858 |
{
|
| 859 |
"epoch": 0.6193548387096774,
|
| 860 |
-
"grad_norm":
|
| 861 |
-
"learning_rate":
|
| 862 |
-
"log_odds_chosen":
|
| 863 |
-
"log_odds_ratio": -0.
|
| 864 |
-
"logits/chosen": -2.
|
| 865 |
-
"logits/rejected": 15.
|
| 866 |
-
"logps/chosen": -0.
|
| 867 |
-
"logps/rejected": -
|
| 868 |
-
"loss": 0.
|
| 869 |
-
"nll_loss": 0.
|
| 870 |
-
"rewards/accuracies": 0.
|
| 871 |
-
"rewards/chosen": -0.
|
| 872 |
-
"rewards/margins": 0.
|
| 873 |
-
"rewards/rejected": -0.
|
| 874 |
"step": 480
|
| 875 |
},
|
| 876 |
{
|
| 877 |
"epoch": 0.632258064516129,
|
| 878 |
-
"grad_norm":
|
| 879 |
-
"learning_rate":
|
| 880 |
-
"log_odds_chosen":
|
| 881 |
-
"log_odds_ratio": -0.
|
| 882 |
-
"logits/chosen": -
|
| 883 |
-
"logits/rejected": 15.
|
| 884 |
-
"logps/chosen": -
|
| 885 |
-
"logps/rejected": -
|
| 886 |
-
"loss":
|
| 887 |
-
"nll_loss":
|
| 888 |
-
"rewards/accuracies": 0.
|
| 889 |
-
"rewards/chosen": -0.
|
| 890 |
-
"rewards/margins": 0.
|
| 891 |
-
"rewards/rejected": -0.
|
| 892 |
"step": 490
|
| 893 |
},
|
| 894 |
{
|
| 895 |
"epoch": 0.6451612903225806,
|
| 896 |
-
"grad_norm":
|
| 897 |
-
"learning_rate":
|
| 898 |
-
"log_odds_chosen":
|
| 899 |
-
"log_odds_ratio": -0.
|
| 900 |
-
"logits/chosen": -2.
|
| 901 |
-
"logits/rejected": 15.
|
| 902 |
-
"logps/chosen": -
|
| 903 |
-
"logps/rejected": -
|
| 904 |
-
"loss": 1.
|
| 905 |
-
"nll_loss": 1.
|
| 906 |
-
"rewards/accuracies": 0.
|
| 907 |
-
"rewards/chosen": -0.
|
| 908 |
-
"rewards/margins": 0.
|
| 909 |
-
"rewards/rejected": -0.
|
| 910 |
"step": 500
|
| 911 |
},
|
| 912 |
{
|
| 913 |
"epoch": 0.6580645161290323,
|
| 914 |
-
"grad_norm":
|
| 915 |
-
"learning_rate":
|
| 916 |
-
"log_odds_chosen":
|
| 917 |
-
"log_odds_ratio": -0.
|
| 918 |
-
"logits/chosen": -
|
| 919 |
-
"logits/rejected": 15.
|
| 920 |
-
"logps/chosen": -0.
|
| 921 |
-
"logps/rejected": -
|
| 922 |
-
"loss":
|
| 923 |
-
"nll_loss":
|
| 924 |
"rewards/accuracies": 0.987500011920929,
|
| 925 |
-
"rewards/chosen": -0.
|
| 926 |
-
"rewards/margins": 0.
|
| 927 |
-
"rewards/rejected": -0.
|
| 928 |
"step": 510
|
| 929 |
},
|
| 930 |
{
|
| 931 |
"epoch": 0.6709677419354839,
|
| 932 |
-
"grad_norm":
|
| 933 |
-
"learning_rate":
|
| 934 |
-
"log_odds_chosen":
|
| 935 |
-
"log_odds_ratio": -0.
|
| 936 |
-
"logits/chosen": -
|
| 937 |
-
"logits/rejected": 15.
|
| 938 |
-
"logps/chosen": -0.
|
| 939 |
-
"logps/rejected": -
|
| 940 |
-
"loss":
|
| 941 |
-
"nll_loss": 0.
|
| 942 |
"rewards/accuracies": 0.9937499761581421,
|
| 943 |
-
"rewards/chosen": -0.
|
| 944 |
-
"rewards/margins": 0.
|
| 945 |
-
"rewards/rejected": -0.
|
| 946 |
"step": 520
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 0.6838709677419355,
|
| 950 |
-
"grad_norm":
|
| 951 |
-
"learning_rate":
|
| 952 |
-
"log_odds_chosen":
|
| 953 |
-
"log_odds_ratio": -0.
|
| 954 |
-
"logits/chosen": -
|
| 955 |
-
"logits/rejected":
|
| 956 |
-
"logps/chosen": -0.
|
| 957 |
-
"logps/rejected": -
|
| 958 |
-
"loss": 0.
|
| 959 |
-
"nll_loss": 0.
|
| 960 |
-
"rewards/accuracies": 0.
|
| 961 |
-
"rewards/chosen": -0.
|
| 962 |
-
"rewards/margins": 0.
|
| 963 |
-
"rewards/rejected": -0.
|
| 964 |
"step": 530
|
| 965 |
},
|
| 966 |
{
|
| 967 |
"epoch": 0.6967741935483871,
|
| 968 |
-
"grad_norm":
|
| 969 |
-
"learning_rate":
|
| 970 |
-
"log_odds_chosen":
|
| 971 |
-
"log_odds_ratio": -0.
|
| 972 |
-
"logits/chosen": -1.
|
| 973 |
-
"logits/rejected":
|
| 974 |
-
"logps/chosen": -
|
| 975 |
-
"logps/rejected": -
|
| 976 |
-
"loss":
|
| 977 |
-
"nll_loss":
|
| 978 |
-
"rewards/accuracies": 0.
|
| 979 |
-
"rewards/chosen": -0.
|
| 980 |
-
"rewards/margins": 0.
|
| 981 |
-
"rewards/rejected": -0.
|
| 982 |
"step": 540
|
| 983 |
},
|
| 984 |
{
|
| 985 |
"epoch": 0.7096774193548387,
|
| 986 |
-
"grad_norm":
|
| 987 |
-
"learning_rate":
|
| 988 |
-
"log_odds_chosen":
|
| 989 |
-
"log_odds_ratio": -0.
|
| 990 |
-
"logits/chosen": -2.
|
| 991 |
-
"logits/rejected": 15.
|
| 992 |
-
"logps/chosen": -
|
| 993 |
-
"logps/rejected": -
|
| 994 |
-
"loss":
|
| 995 |
-
"nll_loss":
|
| 996 |
"rewards/accuracies": 0.987500011920929,
|
| 997 |
-
"rewards/chosen": -0.
|
| 998 |
-
"rewards/margins": 0.
|
| 999 |
-
"rewards/rejected": -0.
|
| 1000 |
"step": 550
|
| 1001 |
},
|
| 1002 |
{
|
| 1003 |
"epoch": 0.7225806451612903,
|
| 1004 |
-
"grad_norm":
|
| 1005 |
-
"learning_rate":
|
| 1006 |
-
"log_odds_chosen":
|
| 1007 |
-
"log_odds_ratio": -0.
|
| 1008 |
-
"logits/chosen": -
|
| 1009 |
-
"logits/rejected": 15.
|
| 1010 |
-
"logps/chosen": -
|
| 1011 |
-
"logps/rejected": -
|
| 1012 |
-
"loss":
|
| 1013 |
-
"nll_loss":
|
| 1014 |
-
"rewards/accuracies": 0.
|
| 1015 |
-
"rewards/chosen": -0.
|
| 1016 |
-
"rewards/margins": 0.
|
| 1017 |
-
"rewards/rejected": -0.
|
| 1018 |
"step": 560
|
| 1019 |
},
|
| 1020 |
{
|
| 1021 |
"epoch": 0.7354838709677419,
|
| 1022 |
-
"grad_norm":
|
| 1023 |
-
"learning_rate":
|
| 1024 |
-
"log_odds_chosen":
|
| 1025 |
-
"log_odds_ratio": -0.
|
| 1026 |
-
"logits/chosen": -2.
|
| 1027 |
-
"logits/rejected": 15.
|
| 1028 |
-
"logps/chosen": -0.
|
| 1029 |
-
"logps/rejected": -
|
| 1030 |
-
"loss": 0.
|
| 1031 |
-
"nll_loss": 0.
|
| 1032 |
"rewards/accuracies": 0.9937499761581421,
|
| 1033 |
-
"rewards/chosen": -0.
|
| 1034 |
-
"rewards/margins": 0.
|
| 1035 |
-
"rewards/rejected": -0.
|
| 1036 |
"step": 570
|
| 1037 |
},
|
| 1038 |
{
|
| 1039 |
"epoch": 0.7483870967741936,
|
| 1040 |
-
"grad_norm":
|
| 1041 |
-
"learning_rate":
|
| 1042 |
-
"log_odds_chosen":
|
| 1043 |
-
"log_odds_ratio": -0.
|
| 1044 |
-
"logits/chosen": -
|
| 1045 |
-
"logits/rejected": 14.
|
| 1046 |
-
"logps/chosen": -0.
|
| 1047 |
-
"logps/rejected": -
|
| 1048 |
-
"loss": 0.
|
| 1049 |
-
"nll_loss": 0.
|
| 1050 |
"rewards/accuracies": 0.981249988079071,
|
| 1051 |
-
"rewards/chosen": -0.
|
| 1052 |
-
"rewards/margins": 0.
|
| 1053 |
-
"rewards/rejected": -0.
|
| 1054 |
"step": 580
|
| 1055 |
},
|
| 1056 |
{
|
| 1057 |
"epoch": 0.7612903225806451,
|
| 1058 |
-
"grad_norm":
|
| 1059 |
-
"learning_rate":
|
| 1060 |
-
"log_odds_chosen":
|
| 1061 |
-
"log_odds_ratio": -0.
|
| 1062 |
-
"logits/chosen": -
|
| 1063 |
-
"logits/rejected": 15.
|
| 1064 |
-
"logps/chosen": -0.
|
| 1065 |
-
"logps/rejected": -
|
| 1066 |
-
"loss": 0.
|
| 1067 |
-
"nll_loss": 0.
|
| 1068 |
-
"rewards/accuracies": 0.
|
| 1069 |
-
"rewards/chosen": -0.
|
| 1070 |
-
"rewards/margins": 0.
|
| 1071 |
-
"rewards/rejected": -0.
|
| 1072 |
"step": 590
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 0.7741935483870968,
|
| 1076 |
-
"grad_norm":
|
| 1077 |
-
"learning_rate":
|
| 1078 |
-
"log_odds_chosen":
|
| 1079 |
-
"log_odds_ratio": -0.
|
| 1080 |
-
"logits/chosen": -2.
|
| 1081 |
-
"logits/rejected": 15.
|
| 1082 |
-
"logps/chosen": -
|
| 1083 |
-
"logps/rejected": -
|
| 1084 |
-
"loss":
|
| 1085 |
-
"nll_loss":
|
| 1086 |
"rewards/accuracies": 0.9937499761581421,
|
| 1087 |
-
"rewards/chosen": -0.
|
| 1088 |
-
"rewards/margins": 0.
|
| 1089 |
-
"rewards/rejected": -0.
|
| 1090 |
"step": 600
|
| 1091 |
},
|
| 1092 |
{
|
| 1093 |
"epoch": 0.7870967741935484,
|
| 1094 |
-
"grad_norm":
|
| 1095 |
-
"learning_rate":
|
| 1096 |
-
"log_odds_chosen":
|
| 1097 |
-
"log_odds_ratio": -0.
|
| 1098 |
-
"logits/chosen": -3.
|
| 1099 |
-
"logits/rejected": 15.
|
| 1100 |
-
"logps/chosen": -
|
| 1101 |
-
"logps/rejected": -
|
| 1102 |
-
"loss":
|
| 1103 |
-
"nll_loss":
|
| 1104 |
-
"rewards/accuracies": 0.
|
| 1105 |
-
"rewards/chosen": -0.
|
| 1106 |
-
"rewards/margins": 0.
|
| 1107 |
-
"rewards/rejected": -0.
|
| 1108 |
"step": 610
|
| 1109 |
},
|
| 1110 |
{
|
| 1111 |
"epoch": 0.8,
|
| 1112 |
-
"grad_norm":
|
| 1113 |
-
"learning_rate":
|
| 1114 |
-
"log_odds_chosen":
|
| 1115 |
-
"log_odds_ratio": -0.
|
| 1116 |
-
"logits/chosen": -
|
| 1117 |
-
"logits/rejected": 15.
|
| 1118 |
-
"logps/chosen": -0.
|
| 1119 |
-
"logps/rejected": -
|
| 1120 |
-
"loss":
|
| 1121 |
-
"nll_loss":
|
| 1122 |
"rewards/accuracies": 0.981249988079071,
|
| 1123 |
-
"rewards/chosen": -0.
|
| 1124 |
-
"rewards/margins": 0.
|
| 1125 |
-
"rewards/rejected": -0.
|
| 1126 |
"step": 620
|
| 1127 |
},
|
| 1128 |
{
|
| 1129 |
"epoch": 0.8129032258064516,
|
| 1130 |
-
"grad_norm":
|
| 1131 |
-
"learning_rate":
|
| 1132 |
-
"log_odds_chosen":
|
| 1133 |
-
"log_odds_ratio": -0.
|
| 1134 |
-
"logits/chosen": -
|
| 1135 |
-
"logits/rejected": 15.
|
| 1136 |
-
"logps/chosen": -0.
|
| 1137 |
-
"logps/rejected": -
|
| 1138 |
-
"loss": 0.
|
| 1139 |
-
"nll_loss": 0.
|
| 1140 |
-
"rewards/accuracies": 0
|
| 1141 |
-
"rewards/chosen": -0.
|
| 1142 |
-
"rewards/margins": 0.
|
| 1143 |
-
"rewards/rejected": -0.
|
| 1144 |
"step": 630
|
| 1145 |
},
|
| 1146 |
{
|
| 1147 |
"epoch": 0.8258064516129032,
|
| 1148 |
-
"grad_norm":
|
| 1149 |
-
"learning_rate":
|
| 1150 |
-
"log_odds_chosen":
|
| 1151 |
-
"log_odds_ratio": -0.
|
| 1152 |
-
"logits/chosen": -2.
|
| 1153 |
-
"logits/rejected": 15.
|
| 1154 |
-
"logps/chosen": -
|
| 1155 |
-
"logps/rejected": -
|
| 1156 |
-
"loss":
|
| 1157 |
-
"nll_loss":
|
| 1158 |
-
"rewards/accuracies": 0
|
| 1159 |
-
"rewards/chosen": -0.
|
| 1160 |
-
"rewards/margins": 0.
|
| 1161 |
-
"rewards/rejected": -0.
|
| 1162 |
"step": 640
|
| 1163 |
},
|
| 1164 |
{
|
| 1165 |
"epoch": 0.8387096774193549,
|
| 1166 |
-
"grad_norm":
|
| 1167 |
-
"learning_rate":
|
| 1168 |
-
"log_odds_chosen":
|
| 1169 |
-
"log_odds_ratio": -0.
|
| 1170 |
-
"logits/chosen": -
|
| 1171 |
-
"logits/rejected": 16.
|
| 1172 |
-
"logps/chosen": -
|
| 1173 |
-
"logps/rejected": -
|
| 1174 |
-
"loss":
|
| 1175 |
-
"nll_loss":
|
| 1176 |
-
"rewards/accuracies": 0.
|
| 1177 |
-
"rewards/chosen": -0.
|
| 1178 |
-
"rewards/margins": 0.
|
| 1179 |
-
"rewards/rejected": -0.
|
| 1180 |
"step": 650
|
| 1181 |
},
|
| 1182 |
{
|
| 1183 |
"epoch": 0.8516129032258064,
|
| 1184 |
-
"grad_norm":
|
| 1185 |
-
"learning_rate":
|
| 1186 |
-
"log_odds_chosen":
|
| 1187 |
-
"log_odds_ratio": -0.
|
| 1188 |
-
"logits/chosen": -
|
| 1189 |
-
"logits/rejected": 15.
|
| 1190 |
-
"logps/chosen": -
|
| 1191 |
-
"logps/rejected": -
|
| 1192 |
-
"loss":
|
| 1193 |
-
"nll_loss":
|
| 1194 |
-
"rewards/accuracies": 0.
|
| 1195 |
-
"rewards/chosen": -0.
|
| 1196 |
-
"rewards/margins": 0.
|
| 1197 |
-
"rewards/rejected": -0.
|
| 1198 |
"step": 660
|
| 1199 |
},
|
| 1200 |
{
|
| 1201 |
"epoch": 0.864516129032258,
|
| 1202 |
-
"grad_norm":
|
| 1203 |
-
"learning_rate":
|
| 1204 |
-
"log_odds_chosen":
|
| 1205 |
-
"log_odds_ratio": -0.
|
| 1206 |
-
"logits/chosen": -
|
| 1207 |
-
"logits/rejected": 15.
|
| 1208 |
-
"logps/chosen": -0.
|
| 1209 |
-
"logps/rejected": -
|
| 1210 |
-
"loss":
|
| 1211 |
-
"nll_loss":
|
| 1212 |
-
"rewards/accuracies": 0.
|
| 1213 |
-
"rewards/chosen": -0.
|
| 1214 |
-
"rewards/margins": 0.
|
| 1215 |
-
"rewards/rejected": -0.
|
| 1216 |
"step": 670
|
| 1217 |
},
|
| 1218 |
{
|
| 1219 |
"epoch": 0.8774193548387097,
|
| 1220 |
-
"grad_norm":
|
| 1221 |
-
"learning_rate":
|
| 1222 |
-
"log_odds_chosen":
|
| 1223 |
-
"log_odds_ratio": -0.
|
| 1224 |
-
"logits/chosen": -2.
|
| 1225 |
-
"logits/rejected": 15.
|
| 1226 |
-
"logps/chosen": -
|
| 1227 |
-
"logps/rejected": -
|
| 1228 |
-
"loss":
|
| 1229 |
-
"nll_loss":
|
| 1230 |
-
"rewards/accuracies": 0
|
| 1231 |
-
"rewards/chosen": -0.
|
| 1232 |
-
"rewards/margins": 0.
|
| 1233 |
-
"rewards/rejected": -0.
|
| 1234 |
"step": 680
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
"epoch": 0.8903225806451613,
|
| 1238 |
-
"grad_norm":
|
| 1239 |
-
"learning_rate":
|
| 1240 |
-
"log_odds_chosen":
|
| 1241 |
-
"log_odds_ratio": -0.
|
| 1242 |
-
"logits/chosen": -2.
|
| 1243 |
-
"logits/rejected":
|
| 1244 |
-
"logps/chosen": -
|
| 1245 |
-
"logps/rejected": -
|
| 1246 |
-
"loss":
|
| 1247 |
-
"nll_loss":
|
| 1248 |
"rewards/accuracies": 0.9750000238418579,
|
| 1249 |
-
"rewards/chosen": -0.
|
| 1250 |
-
"rewards/margins": 0.
|
| 1251 |
-
"rewards/rejected": -0.
|
| 1252 |
"step": 690
|
| 1253 |
},
|
| 1254 |
{
|
| 1255 |
"epoch": 0.9032258064516129,
|
| 1256 |
-
"grad_norm":
|
| 1257 |
-
"learning_rate":
|
| 1258 |
-
"log_odds_chosen":
|
| 1259 |
-
"log_odds_ratio": -0.
|
| 1260 |
-
"logits/chosen": -2.
|
| 1261 |
-
"logits/rejected": 15.
|
| 1262 |
-
"logps/chosen": -0.
|
| 1263 |
-
"logps/rejected": -
|
| 1264 |
-
"loss": 0.
|
| 1265 |
-
"nll_loss": 0.
|
| 1266 |
-
"rewards/accuracies": 0.
|
| 1267 |
-
"rewards/chosen": -0.
|
| 1268 |
-
"rewards/margins": 0.
|
| 1269 |
-
"rewards/rejected": -0.
|
| 1270 |
"step": 700
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 0.9161290322580645,
|
| 1274 |
-
"grad_norm":
|
| 1275 |
-
"learning_rate":
|
| 1276 |
-
"log_odds_chosen":
|
| 1277 |
-
"log_odds_ratio": -0.
|
| 1278 |
-
"logits/chosen": -
|
| 1279 |
-
"logits/rejected": 15.
|
| 1280 |
-
"logps/chosen": -0.
|
| 1281 |
-
"logps/rejected": -
|
| 1282 |
-
"loss":
|
| 1283 |
-
"nll_loss":
|
| 1284 |
"rewards/accuracies": 0.9750000238418579,
|
| 1285 |
-
"rewards/chosen": -0.
|
| 1286 |
-
"rewards/margins": 0.
|
| 1287 |
-
"rewards/rejected": -0.
|
| 1288 |
"step": 710
|
| 1289 |
},
|
| 1290 |
{
|
| 1291 |
"epoch": 0.9290322580645162,
|
| 1292 |
-
"grad_norm":
|
| 1293 |
-
"learning_rate":
|
| 1294 |
-
"log_odds_chosen":
|
| 1295 |
-
"log_odds_ratio": -0.
|
| 1296 |
-
"logits/chosen": -2.
|
| 1297 |
-
"logits/rejected": 15.
|
| 1298 |
-
"logps/chosen": -0.
|
| 1299 |
-
"logps/rejected": -
|
| 1300 |
-
"loss": 0.
|
| 1301 |
-
"nll_loss": 0.
|
| 1302 |
"rewards/accuracies": 1.0,
|
| 1303 |
-
"rewards/chosen": -0.
|
| 1304 |
-
"rewards/margins": 0.
|
| 1305 |
-
"rewards/rejected": -0.
|
| 1306 |
"step": 720
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
"epoch": 0.9419354838709677,
|
| 1310 |
-
"grad_norm":
|
| 1311 |
-
"learning_rate":
|
| 1312 |
-
"log_odds_chosen":
|
| 1313 |
-
"log_odds_ratio": -0.
|
| 1314 |
-
"logits/chosen": -
|
| 1315 |
-
"logits/rejected": 15.
|
| 1316 |
-
"logps/chosen": -0.
|
| 1317 |
-
"logps/rejected": -
|
| 1318 |
-
"loss":
|
| 1319 |
-
"nll_loss":
|
| 1320 |
-
"rewards/accuracies": 0.
|
| 1321 |
-
"rewards/chosen": -0.
|
| 1322 |
-
"rewards/margins": 0.
|
| 1323 |
-
"rewards/rejected": -0.
|
| 1324 |
"step": 730
|
| 1325 |
},
|
| 1326 |
{
|
| 1327 |
"epoch": 0.9548387096774194,
|
| 1328 |
-
"grad_norm":
|
| 1329 |
-
"learning_rate":
|
| 1330 |
-
"log_odds_chosen":
|
| 1331 |
-
"log_odds_ratio": -0.
|
| 1332 |
-
"logits/chosen": -
|
| 1333 |
-
"logits/rejected": 14.
|
| 1334 |
-
"logps/chosen": -0.
|
| 1335 |
-
"logps/rejected": -
|
| 1336 |
-
"loss": 0.
|
| 1337 |
-
"nll_loss": 0.
|
| 1338 |
-
"rewards/accuracies": 0.
|
| 1339 |
-
"rewards/chosen": -0.
|
| 1340 |
-
"rewards/margins": 0.
|
| 1341 |
-
"rewards/rejected": -0.
|
| 1342 |
"step": 740
|
| 1343 |
},
|
| 1344 |
{
|
| 1345 |
"epoch": 0.967741935483871,
|
| 1346 |
-
"grad_norm":
|
| 1347 |
-
"learning_rate":
|
| 1348 |
-
"log_odds_chosen":
|
| 1349 |
-
"log_odds_ratio": -0.
|
| 1350 |
-
"logits/chosen": -
|
| 1351 |
-
"logits/rejected": 15.
|
| 1352 |
-
"logps/chosen": -
|
| 1353 |
-
"logps/rejected": -
|
| 1354 |
-
"loss":
|
| 1355 |
-
"nll_loss":
|
| 1356 |
-
"rewards/accuracies": 0.
|
| 1357 |
-
"rewards/chosen": -0.
|
| 1358 |
-
"rewards/margins": 0.
|
| 1359 |
-
"rewards/rejected": -0.
|
| 1360 |
"step": 750
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
"epoch": 0.9806451612903225,
|
| 1364 |
-
"grad_norm":
|
| 1365 |
-
"learning_rate":
|
| 1366 |
-
"log_odds_chosen":
|
| 1367 |
-
"log_odds_ratio": -0.
|
| 1368 |
-
"logits/chosen": -2.
|
| 1369 |
-
"logits/rejected": 15.
|
| 1370 |
-
"logps/chosen": -0.
|
| 1371 |
-
"logps/rejected": -
|
| 1372 |
-
"loss":
|
| 1373 |
-
"nll_loss":
|
| 1374 |
"rewards/accuracies": 0.9750000238418579,
|
| 1375 |
-
"rewards/chosen": -0.
|
| 1376 |
-
"rewards/margins": 0.
|
| 1377 |
-
"rewards/rejected": -0.
|
| 1378 |
"step": 760
|
| 1379 |
},
|
| 1380 |
{
|
| 1381 |
"epoch": 0.9935483870967742,
|
| 1382 |
-
"grad_norm":
|
| 1383 |
-
"learning_rate":
|
| 1384 |
-
"log_odds_chosen":
|
| 1385 |
-
"log_odds_ratio": -0.
|
| 1386 |
-
"logits/chosen": -
|
| 1387 |
-
"logits/rejected": 15.
|
| 1388 |
-
"logps/chosen": -0.
|
| 1389 |
-
"logps/rejected": -
|
| 1390 |
-
"loss":
|
| 1391 |
-
"nll_loss":
|
| 1392 |
-
"rewards/accuracies": 0.
|
| 1393 |
-
"rewards/chosen": -0.
|
| 1394 |
-
"rewards/margins": 0.
|
| 1395 |
-
"rewards/rejected": -0.
|
| 1396 |
"step": 770
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 1.0,
|
| 1400 |
"step": 775,
|
| 1401 |
"total_flos": 0.0,
|
| 1402 |
-
"train_loss": 1.
|
| 1403 |
-
"train_runtime":
|
| 1404 |
-
"train_samples_per_second":
|
| 1405 |
-
"train_steps_per_second": 0.
|
| 1406 |
}
|
| 1407 |
],
|
| 1408 |
"logging_steps": 10,
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.012903225806451613,
|
| 14 |
+
"grad_norm": 35.01725185776426,
|
| 15 |
+
"learning_rate": 4.5000000000000003e-07,
|
| 16 |
+
"log_odds_chosen": 1.0187499523162842,
|
| 17 |
+
"log_odds_ratio": -0.34882813692092896,
|
| 18 |
+
"logits/chosen": -0.8787597417831421,
|
| 19 |
+
"logits/rejected": 15.768750190734863,
|
| 20 |
+
"logps/chosen": -1.615625023841858,
|
| 21 |
+
"logps/rejected": -2.5015625953674316,
|
| 22 |
+
"loss": 1.9536,
|
| 23 |
+
"nll_loss": 1.9367187023162842,
|
| 24 |
"rewards/accuracies": 0.949999988079071,
|
| 25 |
+
"rewards/chosen": -0.08076171576976776,
|
| 26 |
+
"rewards/margins": 0.0443115234375,
|
| 27 |
+
"rewards/rejected": -0.12509766221046448,
|
| 28 |
"step": 10
|
| 29 |
},
|
| 30 |
{
|
| 31 |
"epoch": 0.025806451612903226,
|
| 32 |
+
"grad_norm": 13.48550724489305,
|
| 33 |
+
"learning_rate": 9.500000000000001e-07,
|
| 34 |
+
"log_odds_chosen": 0.690234363079071,
|
| 35 |
+
"log_odds_ratio": -0.4419921934604645,
|
| 36 |
+
"logits/chosen": 0.51593017578125,
|
| 37 |
+
"logits/rejected": 17.450000762939453,
|
| 38 |
+
"logps/chosen": -1.4140625,
|
| 39 |
+
"logps/rejected": -1.9765625,
|
| 40 |
+
"loss": 1.6515,
|
| 41 |
+
"nll_loss": 1.6281249523162842,
|
| 42 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 43 |
+
"rewards/chosen": -0.07062987983226776,
|
| 44 |
+
"rewards/margins": 0.02830810472369194,
|
| 45 |
+
"rewards/rejected": -0.09902343899011612,
|
| 46 |
"step": 20
|
| 47 |
},
|
| 48 |
{
|
| 49 |
"epoch": 0.03870967741935484,
|
| 50 |
+
"grad_norm": 11.199724125486222,
|
| 51 |
+
"learning_rate": 1.45e-06,
|
| 52 |
+
"log_odds_chosen": 0.3243164122104645,
|
| 53 |
+
"log_odds_ratio": -0.5767577886581421,
|
| 54 |
+
"logits/chosen": 0.98828125,
|
| 55 |
+
"logits/rejected": 17.287500381469727,
|
| 56 |
+
"logps/chosen": -1.2507812976837158,
|
| 57 |
+
"logps/rejected": -1.490625023841858,
|
| 58 |
+
"loss": 1.4296,
|
| 59 |
+
"nll_loss": 1.400781273841858,
|
| 60 |
+
"rewards/accuracies": 0.737500011920929,
|
| 61 |
+
"rewards/chosen": -0.0625,
|
| 62 |
+
"rewards/margins": 0.012020873837172985,
|
| 63 |
+
"rewards/rejected": -0.07456054538488388,
|
| 64 |
"step": 30
|
| 65 |
},
|
| 66 |
{
|
| 67 |
"epoch": 0.05161290322580645,
|
| 68 |
+
"grad_norm": 9.337042800085895,
|
| 69 |
+
"learning_rate": 1.9500000000000004e-06,
|
| 70 |
+
"log_odds_chosen": 0.2557617127895355,
|
| 71 |
+
"log_odds_ratio": -0.603515625,
|
| 72 |
+
"logits/chosen": 0.38398438692092896,
|
| 73 |
+
"logits/rejected": 17.081249237060547,
|
| 74 |
+
"logps/chosen": -1.111718773841858,
|
| 75 |
+
"logps/rejected": -1.2937500476837158,
|
| 76 |
+
"loss": 1.2509,
|
| 77 |
+
"nll_loss": 1.220312476158142,
|
| 78 |
+
"rewards/accuracies": 0.699999988079071,
|
| 79 |
+
"rewards/chosen": -0.05561523512005806,
|
| 80 |
+
"rewards/margins": 0.009097290225327015,
|
| 81 |
+
"rewards/rejected": -0.06474609673023224,
|
| 82 |
"step": 40
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 0.06451612903225806,
|
| 86 |
+
"grad_norm": 8.845956747236016,
|
| 87 |
+
"learning_rate": 2.4500000000000003e-06,
|
| 88 |
+
"log_odds_chosen": 0.4603515565395355,
|
| 89 |
+
"log_odds_ratio": -0.525390625,
|
| 90 |
+
"logits/chosen": -0.30903321504592896,
|
| 91 |
+
"logits/rejected": 16.443750381469727,
|
| 92 |
+
"logps/chosen": -1.074609398841858,
|
| 93 |
+
"logps/rejected": -1.40625,
|
| 94 |
+
"loss": 1.1469,
|
| 95 |
+
"nll_loss": 1.119531273841858,
|
| 96 |
+
"rewards/accuracies": 0.793749988079071,
|
| 97 |
+
"rewards/chosen": -0.05373535305261612,
|
| 98 |
+
"rewards/margins": 0.01662902906537056,
|
| 99 |
+
"rewards/rejected": -0.07036133110523224,
|
| 100 |
"step": 50
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.07741935483870968,
|
| 104 |
+
"grad_norm": 11.864656308978198,
|
| 105 |
+
"learning_rate": 2.95e-06,
|
| 106 |
+
"log_odds_chosen": 0.994140625,
|
| 107 |
+
"log_odds_ratio": -0.3656249940395355,
|
| 108 |
+
"logits/chosen": -0.944531261920929,
|
| 109 |
+
"logits/rejected": 15.512499809265137,
|
| 110 |
+
"logps/chosen": -1.09375,
|
| 111 |
+
"logps/rejected": -1.865625023841858,
|
| 112 |
+
"loss": 1.1414,
|
| 113 |
+
"nll_loss": 1.1242187023162842,
|
| 114 |
+
"rewards/accuracies": 0.925000011920929,
|
| 115 |
+
"rewards/chosen": -0.0546875,
|
| 116 |
+
"rewards/margins": 0.03852539137005806,
|
| 117 |
+
"rewards/rejected": -0.09321288764476776,
|
| 118 |
"step": 60
|
| 119 |
},
|
| 120 |
{
|
| 121 |
"epoch": 0.09032258064516129,
|
| 122 |
+
"grad_norm": 8.085176699285602,
|
| 123 |
+
"learning_rate": 3.45e-06,
|
| 124 |
+
"log_odds_chosen": 1.61328125,
|
| 125 |
+
"log_odds_ratio": -0.24765625596046448,
|
| 126 |
+
"logits/chosen": -1.749609351158142,
|
| 127 |
+
"logits/rejected": 15.681249618530273,
|
| 128 |
+
"logps/chosen": -1.0242187976837158,
|
| 129 |
+
"logps/rejected": -2.328906297683716,
|
| 130 |
+
"loss": 1.0588,
|
| 131 |
+
"nll_loss": 1.0457031726837158,
|
| 132 |
+
"rewards/accuracies": 0.981249988079071,
|
| 133 |
+
"rewards/chosen": -0.05122070387005806,
|
| 134 |
+
"rewards/margins": 0.065185546875,
|
| 135 |
+
"rewards/rejected": -0.116455078125,
|
| 136 |
"step": 70
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.1032258064516129,
|
| 140 |
+
"grad_norm": 8.293601800252382,
|
| 141 |
+
"learning_rate": 3.95e-06,
|
| 142 |
+
"log_odds_chosen": 1.795312523841858,
|
| 143 |
+
"log_odds_ratio": -0.2099609375,
|
| 144 |
+
"logits/chosen": -1.5556640625,
|
| 145 |
+
"logits/rejected": 15.699999809265137,
|
| 146 |
+
"logps/chosen": -1.0593750476837158,
|
| 147 |
+
"logps/rejected": -2.520312547683716,
|
| 148 |
+
"loss": 1.0778,
|
| 149 |
+
"nll_loss": 1.0675780773162842,
|
| 150 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 151 |
+
"rewards/chosen": -0.052978515625,
|
| 152 |
+
"rewards/margins": 0.07309570163488388,
|
| 153 |
+
"rewards/rejected": -0.12607422471046448,
|
| 154 |
"step": 80
|
| 155 |
},
|
| 156 |
{
|
| 157 |
"epoch": 0.11612903225806452,
|
| 158 |
+
"grad_norm": 10.024940843527872,
|
| 159 |
+
"learning_rate": 4.450000000000001e-06,
|
| 160 |
+
"log_odds_chosen": 1.907812476158142,
|
| 161 |
+
"log_odds_ratio": -0.21826171875,
|
| 162 |
+
"logits/chosen": -1.687109351158142,
|
| 163 |
+
"logits/rejected": 15.618749618530273,
|
| 164 |
+
"logps/chosen": -1.0261719226837158,
|
| 165 |
+
"logps/rejected": -2.5835938453674316,
|
| 166 |
+
"loss": 1.0509,
|
| 167 |
+
"nll_loss": 1.0398437976837158,
|
| 168 |
+
"rewards/accuracies": 0.956250011920929,
|
| 169 |
+
"rewards/chosen": -0.05126953125,
|
| 170 |
+
"rewards/margins": 0.078125,
|
| 171 |
+
"rewards/rejected": -0.12929686903953552,
|
| 172 |
"step": 90
|
| 173 |
},
|
| 174 |
{
|
| 175 |
"epoch": 0.12903225806451613,
|
| 176 |
+
"grad_norm": 9.561260138128471,
|
| 177 |
+
"learning_rate": 4.95e-06,
|
| 178 |
+
"log_odds_chosen": 2.039843797683716,
|
| 179 |
+
"log_odds_ratio": -0.19287109375,
|
| 180 |
+
"logits/chosen": -1.9093749523162842,
|
| 181 |
+
"logits/rejected": 15.118749618530273,
|
| 182 |
+
"logps/chosen": -1.058984398841858,
|
| 183 |
+
"logps/rejected": -2.7578125,
|
| 184 |
+
"loss": 1.0856,
|
| 185 |
+
"nll_loss": 1.075781226158142,
|
| 186 |
+
"rewards/accuracies": 0.9375,
|
| 187 |
+
"rewards/chosen": -0.05295410007238388,
|
| 188 |
+
"rewards/margins": 0.08500976860523224,
|
| 189 |
+
"rewards/rejected": -0.137939453125,
|
| 190 |
"step": 100
|
| 191 |
},
|
| 192 |
{
|
| 193 |
"epoch": 0.14193548387096774,
|
| 194 |
+
"grad_norm": 7.99980022257946,
|
| 195 |
+
"learning_rate": 4.789131426105757e-06,
|
| 196 |
+
"log_odds_chosen": 2.422656297683716,
|
| 197 |
+
"log_odds_ratio": -0.15205077826976776,
|
| 198 |
+
"logits/chosen": -1.204187035560608,
|
| 199 |
+
"logits/rejected": 15.850000381469727,
|
| 200 |
+
"logps/chosen": -1.05078125,
|
| 201 |
+
"logps/rejected": -3.0999999046325684,
|
| 202 |
+
"loss": 1.0735,
|
| 203 |
+
"nll_loss": 1.064453125,
|
| 204 |
+
"rewards/accuracies": 0.981249988079071,
|
| 205 |
+
"rewards/chosen": -0.052490234375,
|
| 206 |
+
"rewards/margins": 0.10239257663488388,
|
| 207 |
+
"rewards/rejected": -0.15498046576976776,
|
| 208 |
"step": 110
|
| 209 |
},
|
| 210 |
{
|
| 211 |
"epoch": 0.15483870967741936,
|
| 212 |
+
"grad_norm": 7.301318748826927,
|
| 213 |
+
"learning_rate": 4.583492485141056e-06,
|
| 214 |
+
"log_odds_chosen": 2.557812452316284,
|
| 215 |
+
"log_odds_ratio": -0.16196289658546448,
|
| 216 |
+
"logits/chosen": -2.0648436546325684,
|
| 217 |
+
"logits/rejected": 16.181249618530273,
|
| 218 |
+
"logps/chosen": -1.025390625,
|
| 219 |
+
"logps/rejected": -3.1953125,
|
| 220 |
+
"loss": 1.0461,
|
| 221 |
+
"nll_loss": 1.0378906726837158,
|
| 222 |
+
"rewards/accuracies": 0.949999988079071,
|
| 223 |
+
"rewards/chosen": -0.05124511569738388,
|
| 224 |
+
"rewards/margins": 0.10859374701976776,
|
| 225 |
+
"rewards/rejected": -0.15976563096046448,
|
| 226 |
"step": 120
|
| 227 |
},
|
| 228 |
{
|
| 229 |
"epoch": 0.16774193548387098,
|
| 230 |
+
"grad_norm": 7.9997556459437735,
|
| 231 |
+
"learning_rate": 4.4022545316281195e-06,
|
| 232 |
+
"log_odds_chosen": 2.9625000953674316,
|
| 233 |
+
"log_odds_ratio": -0.11276855319738388,
|
| 234 |
+
"logits/chosen": -1.30615234375,
|
| 235 |
+
"logits/rejected": 15.875,
|
| 236 |
+
"logps/chosen": -1.056640625,
|
| 237 |
+
"logps/rejected": -3.6171875,
|
| 238 |
+
"loss": 1.071,
|
| 239 |
+
"nll_loss": 1.06640625,
|
| 240 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 241 |
+
"rewards/chosen": -0.05278320237994194,
|
| 242 |
+
"rewards/margins": 0.12812499701976776,
|
| 243 |
+
"rewards/rejected": -0.18085937201976776,
|
| 244 |
"step": 130
|
| 245 |
},
|
| 246 |
{
|
| 247 |
"epoch": 0.18064516129032257,
|
| 248 |
+
"grad_norm": 8.787059338175244,
|
| 249 |
+
"learning_rate": 4.240944648399854e-06,
|
| 250 |
+
"log_odds_chosen": 2.53515625,
|
| 251 |
+
"log_odds_ratio": -0.14907225966453552,
|
| 252 |
+
"logits/chosen": -1.6693603992462158,
|
| 253 |
+
"logits/rejected": 15.34375,
|
| 254 |
+
"logps/chosen": -1.0421874523162842,
|
| 255 |
+
"logps/rejected": -3.192187547683716,
|
| 256 |
+
"loss": 1.0614,
|
| 257 |
+
"nll_loss": 1.054296851158142,
|
| 258 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 259 |
+
"rewards/chosen": -0.05207519605755806,
|
| 260 |
+
"rewards/margins": 0.10761718451976776,
|
| 261 |
+
"rewards/rejected": -0.15966796875,
|
| 262 |
"step": 140
|
| 263 |
},
|
| 264 |
{
|
| 265 |
"epoch": 0.1935483870967742,
|
| 266 |
+
"grad_norm": 7.326908745174855,
|
| 267 |
+
"learning_rate": 4.096159602595203e-06,
|
| 268 |
+
"log_odds_chosen": 2.682812452316284,
|
| 269 |
+
"log_odds_ratio": -0.16826172173023224,
|
| 270 |
+
"logits/chosen": -2.536914110183716,
|
| 271 |
+
"logits/rejected": 15.175000190734863,
|
| 272 |
+
"logps/chosen": -1.026953101158142,
|
| 273 |
+
"logps/rejected": -3.34375,
|
| 274 |
+
"loss": 1.0456,
|
| 275 |
+
"nll_loss": 1.0378906726837158,
|
| 276 |
"rewards/accuracies": 0.981249988079071,
|
| 277 |
+
"rewards/chosen": -0.05136718600988388,
|
| 278 |
+
"rewards/margins": 0.11567382514476776,
|
| 279 |
+
"rewards/rejected": -0.1669921875,
|
| 280 |
"step": 150
|
| 281 |
},
|
| 282 |
{
|
| 283 |
"epoch": 0.2064516129032258,
|
| 284 |
+
"grad_norm": 9.22122535363528,
|
| 285 |
+
"learning_rate": 3.965257928590721e-06,
|
| 286 |
+
"log_odds_chosen": 2.785937547683716,
|
| 287 |
+
"log_odds_ratio": -0.18808594346046448,
|
| 288 |
+
"logits/chosen": -2.75390625,
|
| 289 |
+
"logits/rejected": 14.600000381469727,
|
| 290 |
+
"logps/chosen": -0.991015613079071,
|
| 291 |
+
"logps/rejected": -3.4000000953674316,
|
| 292 |
+
"loss": 1.0212,
|
| 293 |
+
"nll_loss": 1.011328101158142,
|
| 294 |
"rewards/accuracies": 0.9750000238418579,
|
| 295 |
+
"rewards/chosen": -0.04953613132238388,
|
| 296 |
+
"rewards/margins": 0.12050781399011612,
|
| 297 |
+
"rewards/rejected": -0.17011718451976776,
|
| 298 |
"step": 160
|
| 299 |
},
|
| 300 |
{
|
| 301 |
"epoch": 0.21935483870967742,
|
| 302 |
+
"grad_norm": 7.10853327954206,
|
| 303 |
+
"learning_rate": 3.846153846153846e-06,
|
| 304 |
+
"log_odds_chosen": 3.2984375953674316,
|
| 305 |
+
"log_odds_ratio": -0.1068115234375,
|
| 306 |
+
"logits/chosen": -2.6039061546325684,
|
| 307 |
+
"logits/rejected": 15.818750381469727,
|
| 308 |
+
"logps/chosen": -0.961718738079071,
|
| 309 |
+
"logps/rejected": -3.8218750953674316,
|
| 310 |
+
"loss": 0.9891,
|
| 311 |
+
"nll_loss": 0.9839843511581421,
|
| 312 |
+
"rewards/accuracies": 0.987500011920929,
|
| 313 |
+
"rewards/chosen": -0.04804687574505806,
|
| 314 |
+
"rewards/margins": 0.14321288466453552,
|
| 315 |
+
"rewards/rejected": -0.19121094048023224,
|
| 316 |
"step": 170
|
| 317 |
},
|
| 318 |
{
|
| 319 |
"epoch": 0.23225806451612904,
|
| 320 |
+
"grad_norm": 7.942817891766315,
|
| 321 |
+
"learning_rate": 3.7371754637596797e-06,
|
| 322 |
+
"log_odds_chosen": 3.356250047683716,
|
| 323 |
+
"log_odds_ratio": -0.09636230766773224,
|
| 324 |
+
"logits/chosen": -2.181640625,
|
| 325 |
+
"logits/rejected": 15.274999618530273,
|
| 326 |
+
"logps/chosen": -0.979296863079071,
|
| 327 |
+
"logps/rejected": -3.8890624046325684,
|
| 328 |
+
"loss": 0.9928,
|
| 329 |
+
"nll_loss": 0.98828125,
|
| 330 |
+
"rewards/accuracies": 0.987500011920929,
|
| 331 |
+
"rewards/chosen": -0.0489501953125,
|
| 332 |
+
"rewards/margins": 0.14541015028953552,
|
| 333 |
+
"rewards/rejected": -0.19443359971046448,
|
| 334 |
"step": 180
|
| 335 |
},
|
| 336 |
{
|
| 337 |
"epoch": 0.24516129032258063,
|
| 338 |
+
"grad_norm": 15.366307517097038,
|
| 339 |
+
"learning_rate": 3.63696483726654e-06,
|
| 340 |
+
"log_odds_chosen": 3.3984375,
|
| 341 |
+
"log_odds_ratio": -0.10626220703125,
|
| 342 |
+
"logits/chosen": -1.9557616710662842,
|
| 343 |
+
"logits/rejected": 15.681249618530273,
|
| 344 |
+
"logps/chosen": -0.9742187261581421,
|
| 345 |
+
"logps/rejected": -3.940624952316284,
|
| 346 |
+
"loss": 1.0004,
|
| 347 |
+
"nll_loss": 0.99609375,
|
| 348 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 349 |
+
"rewards/chosen": -0.04877929762005806,
|
| 350 |
+
"rewards/margins": 0.14833983778953552,
|
| 351 |
+
"rewards/rejected": -0.19707031548023224,
|
| 352 |
"step": 190
|
| 353 |
},
|
| 354 |
{
|
| 355 |
"epoch": 0.25806451612903225,
|
| 356 |
+
"grad_norm": 6.891451760355594,
|
| 357 |
+
"learning_rate": 3.5444060250416796e-06,
|
| 358 |
+
"log_odds_chosen": 3.645312547683716,
|
| 359 |
+
"log_odds_ratio": -0.0811767578125,
|
| 360 |
+
"logits/chosen": -2.288281202316284,
|
| 361 |
+
"logits/rejected": 15.893750190734863,
|
| 362 |
+
"logps/chosen": -1.0066406726837158,
|
| 363 |
+
"logps/rejected": -4.215624809265137,
|
| 364 |
+
"loss": 1.0215,
|
| 365 |
+
"nll_loss": 1.017187476158142,
|
| 366 |
+
"rewards/accuracies": 0.987500011920929,
|
| 367 |
+
"rewards/chosen": -0.05036621168255806,
|
| 368 |
+
"rewards/margins": 0.16035155951976776,
|
| 369 |
+
"rewards/rejected": -0.21074219048023224,
|
| 370 |
"step": 200
|
| 371 |
},
|
| 372 |
{
|
| 373 |
"epoch": 0.2709677419354839,
|
| 374 |
+
"grad_norm": 8.135582655341244,
|
| 375 |
+
"learning_rate": 3.458572319330373e-06,
|
| 376 |
+
"log_odds_chosen": 3.387500047683716,
|
| 377 |
+
"log_odds_ratio": -0.14912109076976776,
|
| 378 |
+
"logits/chosen": -2.5718750953674316,
|
| 379 |
+
"logits/rejected": 15.149999618530273,
|
| 380 |
+
"logps/chosen": -0.981249988079071,
|
| 381 |
+
"logps/rejected": -3.9453125,
|
| 382 |
+
"loss": 0.998,
|
| 383 |
+
"nll_loss": 0.991015613079071,
|
| 384 |
+
"rewards/accuracies": 0.9375,
|
| 385 |
+
"rewards/chosen": -0.04902343824505806,
|
| 386 |
+
"rewards/margins": 0.148193359375,
|
| 387 |
+
"rewards/rejected": -0.197265625,
|
| 388 |
"step": 210
|
| 389 |
},
|
| 390 |
{
|
| 391 |
"epoch": 0.2838709677419355,
|
| 392 |
+
"grad_norm": 7.518297474091935,
|
| 393 |
+
"learning_rate": 3.3786868919974304e-06,
|
| 394 |
+
"log_odds_chosen": 2.4242186546325684,
|
| 395 |
+
"log_odds_ratio": -0.17851562798023224,
|
| 396 |
+
"logits/chosen": -1.6906249523162842,
|
| 397 |
+
"logits/rejected": 15.84375,
|
| 398 |
+
"logps/chosen": -1.036718726158142,
|
| 399 |
+
"logps/rejected": -3.0859375,
|
| 400 |
+
"loss": 1.0572,
|
| 401 |
+
"nll_loss": 1.0480468273162842,
|
| 402 |
+
"rewards/accuracies": 0.956250011920929,
|
| 403 |
+
"rewards/chosen": -0.0518798828125,
|
| 404 |
+
"rewards/margins": 0.10249023139476776,
|
| 405 |
+
"rewards/rejected": -0.15439453721046448,
|
| 406 |
"step": 220
|
| 407 |
},
|
| 408 |
{
|
| 409 |
"epoch": 0.2967741935483871,
|
| 410 |
+
"grad_norm": 7.499198073553291,
|
| 411 |
+
"learning_rate": 3.3040930022754493e-06,
|
| 412 |
+
"log_odds_chosen": 2.407031297683716,
|
| 413 |
+
"log_odds_ratio": -0.145263671875,
|
| 414 |
+
"logits/chosen": -1.3737304210662842,
|
| 415 |
+
"logits/rejected": 15.581250190734863,
|
| 416 |
+
"logps/chosen": -1.0304687023162842,
|
| 417 |
+
"logps/rejected": -3.043750047683716,
|
| 418 |
+
"loss": 1.0455,
|
| 419 |
+
"nll_loss": 1.0382812023162842,
|
| 420 |
+
"rewards/accuracies": 0.981249988079071,
|
| 421 |
+
"rewards/chosen": -0.05156249925494194,
|
| 422 |
+
"rewards/margins": 0.10073242336511612,
|
| 423 |
+
"rewards/rejected": -0.15224608778953552,
|
| 424 |
"step": 230
|
| 425 |
},
|
| 426 |
{
|
| 427 |
"epoch": 0.3096774193548387,
|
| 428 |
+
"grad_norm": 7.726811415559842,
|
| 429 |
+
"learning_rate": 3.234231136765755e-06,
|
| 430 |
+
"log_odds_chosen": 2.612499952316284,
|
| 431 |
+
"log_odds_ratio": -0.14296874403953552,
|
| 432 |
+
"logits/chosen": -2.7847657203674316,
|
| 433 |
+
"logits/rejected": 14.0625,
|
| 434 |
+
"logps/chosen": -0.987500011920929,
|
| 435 |
+
"logps/rejected": -3.1937499046325684,
|
| 436 |
+
"loss": 1.0104,
|
| 437 |
+
"nll_loss": 1.00390625,
|
| 438 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 439 |
+
"rewards/chosen": -0.04938964918255806,
|
| 440 |
+
"rewards/margins": 0.11020507663488388,
|
| 441 |
+
"rewards/rejected": -0.15957030653953552,
|
| 442 |
"step": 240
|
| 443 |
},
|
| 444 |
{
|
| 445 |
"epoch": 0.3225806451612903,
|
| 446 |
+
"grad_norm": 7.365705719642815,
|
| 447 |
+
"learning_rate": 3.1686212526223897e-06,
|
| 448 |
+
"log_odds_chosen": 3.2109375,
|
| 449 |
+
"log_odds_ratio": -0.09316406399011612,
|
| 450 |
+
"logits/chosen": -3.19921875,
|
| 451 |
+
"logits/rejected": 15.037500381469727,
|
| 452 |
+
"logps/chosen": -0.948046863079071,
|
| 453 |
+
"logps/rejected": -3.6968750953674316,
|
| 454 |
+
"loss": 0.9679,
|
| 455 |
+
"nll_loss": 0.964062511920929,
|
| 456 |
"rewards/accuracies": 0.987500011920929,
|
| 457 |
+
"rewards/chosen": -0.04741210862994194,
|
| 458 |
+
"rewards/margins": 0.13754883408546448,
|
| 459 |
+
"rewards/rejected": -0.18476562201976776,
|
| 460 |
"step": 250
|
| 461 |
},
|
| 462 |
{
|
| 463 |
"epoch": 0.33548387096774196,
|
| 464 |
+
"grad_norm": 6.518152237715758,
|
| 465 |
+
"learning_rate": 3.1068488300060003e-06,
|
| 466 |
+
"log_odds_chosen": 2.875,
|
| 467 |
+
"log_odds_ratio": -0.1068115234375,
|
| 468 |
+
"logits/chosen": -2.6175780296325684,
|
| 469 |
+
"logits/rejected": 15.149999618530273,
|
| 470 |
+
"logps/chosen": -0.97265625,
|
| 471 |
+
"logps/rejected": -3.4078125953674316,
|
| 472 |
+
"loss": 0.9938,
|
| 473 |
+
"nll_loss": 0.98828125,
|
| 474 |
+
"rewards/accuracies": 0.987500011920929,
|
| 475 |
+
"rewards/chosen": -0.04868163913488388,
|
| 476 |
+
"rewards/margins": 0.12158203125,
|
| 477 |
+
"rewards/rejected": -0.17031249403953552,
|
| 478 |
"step": 260
|
| 479 |
},
|
| 480 |
{
|
| 481 |
"epoch": 0.34838709677419355,
|
| 482 |
+
"grad_norm": 7.237801132096072,
|
| 483 |
+
"learning_rate": 3.0485538042484623e-06,
|
| 484 |
+
"log_odds_chosen": 3.1875,
|
| 485 |
+
"log_odds_ratio": -0.10446777194738388,
|
| 486 |
+
"logits/chosen": -2.614062547683716,
|
| 487 |
+
"logits/rejected": 14.987500190734863,
|
| 488 |
+
"logps/chosen": -0.9515625238418579,
|
| 489 |
+
"logps/rejected": -3.6890625953674316,
|
| 490 |
+
"loss": 0.9738,
|
| 491 |
+
"nll_loss": 0.9691406488418579,
|
| 492 |
+
"rewards/accuracies": 1.0,
|
| 493 |
+
"rewards/chosen": -0.047607421875,
|
| 494 |
+
"rewards/margins": 0.13681641221046448,
|
| 495 |
+
"rewards/rejected": -0.1845703125,
|
| 496 |
"step": 270
|
| 497 |
},
|
| 498 |
{
|
| 499 |
"epoch": 0.36129032258064514,
|
| 500 |
+
"grad_norm": 7.797308798589523,
|
| 501 |
+
"learning_rate": 2.9934217004462487e-06,
|
| 502 |
+
"log_odds_chosen": 3.8890624046325684,
|
| 503 |
+
"log_odds_ratio": -0.0784912109375,
|
| 504 |
+
"logits/chosen": -2.3812499046325684,
|
| 505 |
+
"logits/rejected": 15.181249618530273,
|
| 506 |
+
"logps/chosen": -0.993359386920929,
|
| 507 |
+
"logps/rejected": -4.421875,
|
| 508 |
+
"loss": 0.9999,
|
| 509 |
+
"nll_loss": 0.9976562261581421,
|
| 510 |
"rewards/accuracies": 0.981249988079071,
|
| 511 |
+
"rewards/chosen": -0.04965820163488388,
|
| 512 |
+
"rewards/margins": 0.17138671875,
|
| 513 |
+
"rewards/rejected": -0.22119140625,
|
| 514 |
"step": 280
|
| 515 |
},
|
| 516 |
{
|
| 517 |
"epoch": 0.3741935483870968,
|
| 518 |
+
"grad_norm": 30.586016864966307,
|
| 519 |
+
"learning_rate": 2.9411764705882355e-06,
|
| 520 |
+
"log_odds_chosen": 4.060937404632568,
|
| 521 |
+
"log_odds_ratio": -0.06879882514476776,
|
| 522 |
+
"logits/chosen": -2.482421875,
|
| 523 |
+
"logits/rejected": 15.481249809265137,
|
| 524 |
+
"logps/chosen": -0.962109386920929,
|
| 525 |
+
"logps/rejected": -4.540625095367432,
|
| 526 |
+
"loss": 0.9723,
|
| 527 |
+
"nll_loss": 0.9691406488418579,
|
| 528 |
"rewards/accuracies": 0.981249988079071,
|
| 529 |
+
"rewards/chosen": -0.04816894605755806,
|
| 530 |
+
"rewards/margins": 0.17919921875,
|
| 531 |
+
"rewards/rejected": -0.22734375298023224,
|
| 532 |
"step": 290
|
| 533 |
},
|
| 534 |
{
|
| 535 |
"epoch": 0.3870967741935484,
|
| 536 |
+
"grad_norm": 8.595204860528385,
|
| 537 |
+
"learning_rate": 2.891574659831201e-06,
|
| 538 |
+
"log_odds_chosen": 4.564062595367432,
|
| 539 |
+
"log_odds_ratio": -0.07656250149011612,
|
| 540 |
+
"logits/chosen": -2.651562452316284,
|
| 541 |
+
"logits/rejected": 15.306249618530273,
|
| 542 |
+
"logps/chosen": -0.9839843511581421,
|
| 543 |
+
"logps/rejected": -5.090624809265137,
|
| 544 |
+
"loss": 0.9966,
|
| 545 |
+
"nll_loss": 0.992968738079071,
|
| 546 |
+
"rewards/accuracies": 0.981249988079071,
|
| 547 |
+
"rewards/chosen": -0.04916992038488388,
|
| 548 |
+
"rewards/margins": 0.20488281548023224,
|
| 549 |
+
"rewards/rejected": -0.25439453125,
|
| 550 |
"step": 300
|
| 551 |
},
|
| 552 |
{
|
| 553 |
"epoch": 0.4,
|
| 554 |
+
"grad_norm": 8.703268967809453,
|
| 555 |
+
"learning_rate": 2.844400619942872e-06,
|
| 556 |
+
"log_odds_chosen": 4.431250095367432,
|
| 557 |
+
"log_odds_ratio": -0.07321777194738388,
|
| 558 |
+
"logits/chosen": -2.5328125953674316,
|
| 559 |
+
"logits/rejected": 14.806249618530273,
|
| 560 |
+
"logps/chosen": -0.96875,
|
| 561 |
+
"logps/rejected": -4.932812690734863,
|
| 562 |
+
"loss": 0.9852,
|
| 563 |
+
"nll_loss": 0.982421875,
|
| 564 |
+
"rewards/accuracies": 0.981249988079071,
|
| 565 |
+
"rewards/chosen": -0.04848632961511612,
|
| 566 |
+
"rewards/margins": 0.1982421875,
|
| 567 |
+
"rewards/rejected": -0.24677734076976776,
|
| 568 |
"step": 310
|
| 569 |
},
|
| 570 |
{
|
| 571 |
"epoch": 0.4129032258064516,
|
| 572 |
+
"grad_norm": 7.421491749584791,
|
| 573 |
+
"learning_rate": 2.799462554779272e-06,
|
| 574 |
+
"log_odds_chosen": 4.453125,
|
| 575 |
+
"log_odds_ratio": -0.07643432915210724,
|
| 576 |
+
"logits/chosen": -2.30078125,
|
| 577 |
+
"logits/rejected": 15.4375,
|
| 578 |
+
"logps/chosen": -0.954296886920929,
|
| 579 |
+
"logps/rejected": -4.931250095367432,
|
| 580 |
+
"loss": 0.9638,
|
| 581 |
+
"nll_loss": 0.959765613079071,
|
| 582 |
+
"rewards/accuracies": 0.981249988079071,
|
| 583 |
+
"rewards/chosen": -0.04775390774011612,
|
| 584 |
+
"rewards/margins": 0.19873046875,
|
| 585 |
+
"rewards/rejected": -0.24628905951976776,
|
| 586 |
"step": 320
|
| 587 |
},
|
| 588 |
{
|
| 589 |
"epoch": 0.4258064516129032,
|
| 590 |
+
"grad_norm": 8.918312407814545,
|
| 591 |
+
"learning_rate": 2.7565892320998566e-06,
|
| 592 |
+
"log_odds_chosen": 4.756249904632568,
|
| 593 |
+
"log_odds_ratio": -0.059814453125,
|
| 594 |
+
"logits/chosen": -2.6878905296325684,
|
| 595 |
+
"logits/rejected": 15.5,
|
| 596 |
+
"logps/chosen": -0.982421875,
|
| 597 |
+
"logps/rejected": -5.262499809265137,
|
| 598 |
+
"loss": 1.011,
|
| 599 |
+
"nll_loss": 1.0078125,
|
| 600 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 601 |
+
"rewards/chosen": -0.04909668117761612,
|
| 602 |
+
"rewards/margins": 0.21425780653953552,
|
| 603 |
+
"rewards/rejected": -0.2630859315395355,
|
| 604 |
"step": 330
|
| 605 |
},
|
| 606 |
{
|
| 607 |
"epoch": 0.43870967741935485,
|
| 608 |
+
"grad_norm": 7.528569200319769,
|
| 609 |
+
"learning_rate": 2.7156272329678423e-06,
|
| 610 |
+
"log_odds_chosen": 4.699999809265137,
|
| 611 |
+
"log_odds_ratio": -0.04852294921875,
|
| 612 |
+
"logits/chosen": -2.184375047683716,
|
| 613 |
+
"logits/rejected": 15.287500381469727,
|
| 614 |
+
"logps/chosen": -0.987500011920929,
|
| 615 |
+
"logps/rejected": -5.215624809265137,
|
| 616 |
+
"loss": 0.9846,
|
| 617 |
+
"nll_loss": 0.9820312261581421,
|
| 618 |
+
"rewards/accuracies": 1.0,
|
| 619 |
+
"rewards/chosen": -0.04936523362994194,
|
| 620 |
+
"rewards/margins": 0.21123047173023224,
|
| 621 |
+
"rewards/rejected": -0.2607421875,
|
| 622 |
"step": 340
|
| 623 |
},
|
| 624 |
{
|
| 625 |
"epoch": 0.45161290322580644,
|
| 626 |
+
"grad_norm": 7.297861319360668,
|
| 627 |
+
"learning_rate": 2.6764386378609465e-06,
|
| 628 |
+
"log_odds_chosen": 4.709374904632568,
|
| 629 |
+
"log_odds_ratio": -0.057373046875,
|
| 630 |
+
"logits/chosen": -2.99609375,
|
| 631 |
+
"logits/rejected": 15.09375,
|
| 632 |
+
"logps/chosen": -0.9339843988418579,
|
| 633 |
+
"logps/rejected": -5.134375095367432,
|
| 634 |
+
"loss": 0.9631,
|
| 635 |
+
"nll_loss": 0.9605468511581421,
|
| 636 |
"rewards/accuracies": 0.9937499761581421,
|
| 637 |
+
"rewards/chosen": -0.04672851413488388,
|
| 638 |
+
"rewards/margins": 0.21025390923023224,
|
| 639 |
+
"rewards/rejected": -0.2569335997104645,
|
| 640 |
"step": 350
|
| 641 |
},
|
| 642 |
{
|
| 643 |
"epoch": 0.4645161290322581,
|
| 644 |
+
"grad_norm": 8.07548290053001,
|
| 645 |
+
"learning_rate": 2.638899069846298e-06,
|
| 646 |
+
"log_odds_chosen": 4.639062404632568,
|
| 647 |
+
"log_odds_ratio": -0.06218261644244194,
|
| 648 |
+
"logits/chosen": -2.604687452316284,
|
| 649 |
+
"logits/rejected": 15.231249809265137,
|
| 650 |
+
"logps/chosen": -0.969531238079071,
|
| 651 |
+
"logps/rejected": -5.131249904632568,
|
| 652 |
+
"loss": 0.9815,
|
| 653 |
+
"nll_loss": 0.979687511920929,
|
| 654 |
+
"rewards/accuracies": 0.987500011920929,
|
| 655 |
+
"rewards/chosen": -0.04843749850988388,
|
| 656 |
+
"rewards/margins": 0.2080078125,
|
| 657 |
+
"rewards/rejected": -0.2564453184604645,
|
| 658 |
"step": 360
|
| 659 |
},
|
| 660 |
{
|
| 661 |
"epoch": 0.4774193548387097,
|
| 662 |
+
"grad_norm": 7.439409049540854,
|
| 663 |
+
"learning_rate": 2.6028960314767675e-06,
|
| 664 |
+
"log_odds_chosen": 4.8125,
|
| 665 |
+
"log_odds_ratio": -0.06487731635570526,
|
| 666 |
+
"logits/chosen": -3.2523436546325684,
|
| 667 |
+
"logits/rejected": 15.425000190734863,
|
| 668 |
+
"logps/chosen": -0.981249988079071,
|
| 669 |
+
"logps/rejected": -5.306250095367432,
|
| 670 |
+
"loss": 0.984,
|
| 671 |
+
"nll_loss": 0.981640636920929,
|
| 672 |
+
"rewards/accuracies": 0.981249988079071,
|
| 673 |
+
"rewards/chosen": -0.049072265625,
|
| 674 |
+
"rewards/margins": 0.216796875,
|
| 675 |
+
"rewards/rejected": -0.2657226622104645,
|
| 676 |
"step": 370
|
| 677 |
},
|
| 678 |
{
|
| 679 |
"epoch": 0.49032258064516127,
|
| 680 |
+
"grad_norm": 6.717751304060122,
|
| 681 |
+
"learning_rate": 2.568327484687571e-06,
|
| 682 |
+
"log_odds_chosen": 4.300000190734863,
|
| 683 |
+
"log_odds_ratio": -0.09389648586511612,
|
| 684 |
+
"logits/chosen": -2.848437547683716,
|
| 685 |
+
"logits/rejected": 15.28125,
|
| 686 |
+
"logps/chosen": -0.934765636920929,
|
| 687 |
+
"logps/rejected": -4.760937690734863,
|
| 688 |
+
"loss": 0.9459,
|
| 689 |
+
"nll_loss": 0.94140625,
|
| 690 |
"rewards/accuracies": 0.9750000238418579,
|
| 691 |
+
"rewards/chosen": -0.0467529296875,
|
| 692 |
+
"rewards/margins": 0.19101563096046448,
|
| 693 |
+
"rewards/rejected": -0.23779296875,
|
| 694 |
"step": 380
|
| 695 |
},
|
| 696 |
{
|
| 697 |
"epoch": 0.5032258064516129,
|
| 698 |
+
"grad_norm": 7.456596504746751,
|
| 699 |
+
"learning_rate": 2.535100632816969e-06,
|
| 700 |
+
"log_odds_chosen": 4.435937404632568,
|
| 701 |
+
"log_odds_ratio": -0.07844237983226776,
|
| 702 |
+
"logits/chosen": -2.9749999046325684,
|
| 703 |
+
"logits/rejected": 15.487500190734863,
|
| 704 |
+
"logps/chosen": -0.935546875,
|
| 705 |
+
"logps/rejected": -4.892187595367432,
|
| 706 |
+
"loss": 0.9631,
|
| 707 |
+
"nll_loss": 0.9593750238418579,
|
| 708 |
"rewards/accuracies": 0.96875,
|
| 709 |
+
"rewards/chosen": -0.0467529296875,
|
| 710 |
+
"rewards/margins": 0.19755859673023224,
|
| 711 |
+
"rewards/rejected": -0.24462890625,
|
| 712 |
"step": 390
|
| 713 |
},
|
| 714 |
{
|
| 715 |
"epoch": 0.5161290322580645,
|
| 716 |
+
"grad_norm": 6.423921141623107,
|
| 717 |
+
"learning_rate": 2.5031308716087944e-06,
|
| 718 |
+
"log_odds_chosen": 4.626562595367432,
|
| 719 |
+
"log_odds_ratio": -0.05325927585363388,
|
| 720 |
+
"logits/chosen": -3.141406297683716,
|
| 721 |
+
"logits/rejected": 15.106249809265137,
|
| 722 |
+
"logps/chosen": -0.951171875,
|
| 723 |
+
"logps/rejected": -5.078125,
|
| 724 |
+
"loss": 0.9649,
|
| 725 |
+
"nll_loss": 0.962109386920929,
|
| 726 |
+
"rewards/accuracies": 0.987500011920929,
|
| 727 |
+
"rewards/chosen": -0.04755859449505806,
|
| 728 |
+
"rewards/margins": 0.20634765923023224,
|
| 729 |
+
"rewards/rejected": -0.2538085877895355,
|
| 730 |
"step": 400
|
| 731 |
},
|
| 732 |
{
|
| 733 |
"epoch": 0.5290322580645161,
|
| 734 |
+
"grad_norm": 8.037211110808377,
|
| 735 |
+
"learning_rate": 2.472340882170744e-06,
|
| 736 |
+
"log_odds_chosen": 4.784375190734863,
|
| 737 |
+
"log_odds_ratio": -0.03513183444738388,
|
| 738 |
+
"logits/chosen": -2.914843797683716,
|
| 739 |
+
"logits/rejected": 15.162500381469727,
|
| 740 |
+
"logps/chosen": -0.950390636920929,
|
| 741 |
+
"logps/rejected": -5.237500190734863,
|
| 742 |
+
"loss": 0.957,
|
| 743 |
+
"nll_loss": 0.955078125,
|
| 744 |
+
"rewards/accuracies": 1.0,
|
| 745 |
+
"rewards/chosen": -0.04758300632238388,
|
| 746 |
+
"rewards/margins": 0.21406249701976776,
|
| 747 |
+
"rewards/rejected": -0.26142579317092896,
|
| 748 |
"step": 410
|
| 749 |
},
|
| 750 |
{
|
| 751 |
"epoch": 0.5419354838709678,
|
| 752 |
+
"grad_norm": 7.975874669678937,
|
| 753 |
+
"learning_rate": 2.4426598437301577e-06,
|
| 754 |
+
"log_odds_chosen": 4.809374809265137,
|
| 755 |
+
"log_odds_ratio": -0.06281737983226776,
|
| 756 |
+
"logits/chosen": -2.657031297683716,
|
| 757 |
+
"logits/rejected": 14.993749618530273,
|
| 758 |
+
"logps/chosen": -0.962109386920929,
|
| 759 |
+
"logps/rejected": -5.296875,
|
| 760 |
+
"loss": 0.9726,
|
| 761 |
+
"nll_loss": 0.969921886920929,
|
| 762 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 763 |
+
"rewards/chosen": -0.048095703125,
|
| 764 |
+
"rewards/margins": 0.21660156548023224,
|
| 765 |
+
"rewards/rejected": -0.2645507752895355,
|
| 766 |
"step": 420
|
| 767 |
},
|
| 768 |
{
|
| 769 |
"epoch": 0.5548387096774193,
|
| 770 |
+
"grad_norm": 7.980706154694817,
|
| 771 |
+
"learning_rate": 2.4140227479263382e-06,
|
| 772 |
+
"log_odds_chosen": 4.993750095367432,
|
| 773 |
+
"log_odds_ratio": -0.05314941331744194,
|
| 774 |
+
"logits/chosen": -1.9533202648162842,
|
| 775 |
+
"logits/rejected": 15.962499618530273,
|
| 776 |
+
"logps/chosen": -0.9574218988418579,
|
| 777 |
+
"logps/rejected": -5.465624809265137,
|
| 778 |
+
"loss": 0.9602,
|
| 779 |
+
"nll_loss": 0.957812488079071,
|
| 780 |
+
"rewards/accuracies": 0.981249988079071,
|
| 781 |
+
"rewards/chosen": -0.04787597805261612,
|
| 782 |
+
"rewards/margins": 0.22539062798023224,
|
| 783 |
+
"rewards/rejected": -0.2728515565395355,
|
| 784 |
"step": 430
|
| 785 |
},
|
| 786 |
{
|
| 787 |
"epoch": 0.567741935483871,
|
| 788 |
+
"grad_norm": 9.156473187948562,
|
| 789 |
+
"learning_rate": 2.3863697995167388e-06,
|
| 790 |
+
"log_odds_chosen": 4.903124809265137,
|
| 791 |
+
"log_odds_ratio": -0.04511108249425888,
|
| 792 |
+
"logits/chosen": -3.05126953125,
|
| 793 |
+
"logits/rejected": 15.243749618530273,
|
| 794 |
+
"logps/chosen": -0.9671875238418579,
|
| 795 |
+
"logps/rejected": -5.387499809265137,
|
| 796 |
+
"loss": 0.9845,
|
| 797 |
+
"nll_loss": 0.981640636920929,
|
| 798 |
"rewards/accuracies": 0.9937499761581421,
|
| 799 |
+
"rewards/chosen": -0.04831542819738388,
|
| 800 |
+
"rewards/margins": 0.22089843451976776,
|
| 801 |
+
"rewards/rejected": -0.26953125,
|
| 802 |
"step": 440
|
| 803 |
},
|
| 804 |
{
|
| 805 |
"epoch": 0.5806451612903226,
|
| 806 |
+
"grad_norm": 6.879386887391153,
|
| 807 |
+
"learning_rate": 2.359645890915044e-06,
|
| 808 |
+
"log_odds_chosen": 4.698437690734863,
|
| 809 |
+
"log_odds_ratio": -0.07050170749425888,
|
| 810 |
+
"logits/chosen": -2.684375047683716,
|
| 811 |
+
"logits/rejected": 14.993749618530273,
|
| 812 |
+
"logps/chosen": -0.983203113079071,
|
| 813 |
+
"logps/rejected": -5.199999809265137,
|
| 814 |
+
"loss": 1.0031,
|
| 815 |
+
"nll_loss": 0.9996093511581421,
|
| 816 |
+
"rewards/accuracies": 0.96875,
|
| 817 |
+
"rewards/chosen": -0.0491943359375,
|
| 818 |
+
"rewards/margins": 0.2109375,
|
| 819 |
+
"rewards/rejected": -0.26005858182907104,
|
| 820 |
"step": 450
|
| 821 |
},
|
| 822 |
{
|
| 823 |
"epoch": 0.5935483870967742,
|
| 824 |
+
"grad_norm": 9.291098134467267,
|
| 825 |
+
"learning_rate": 2.333800140046683e-06,
|
| 826 |
+
"log_odds_chosen": 4.96875,
|
| 827 |
+
"log_odds_ratio": -0.05540771409869194,
|
| 828 |
+
"logits/chosen": -2.6537108421325684,
|
| 829 |
+
"logits/rejected": 15.399999618530273,
|
| 830 |
+
"logps/chosen": -0.9515625238418579,
|
| 831 |
+
"logps/rejected": -5.418749809265137,
|
| 832 |
+
"loss": 0.9651,
|
| 833 |
+
"nll_loss": 0.962890625,
|
| 834 |
+
"rewards/accuracies": 0.981249988079071,
|
| 835 |
+
"rewards/chosen": -0.04755859449505806,
|
| 836 |
+
"rewards/margins": 0.22353515028953552,
|
| 837 |
+
"rewards/rejected": -0.2708984315395355,
|
| 838 |
"step": 460
|
| 839 |
},
|
| 840 |
{
|
| 841 |
"epoch": 0.6064516129032258,
|
| 842 |
+
"grad_norm": 8.260430320302163,
|
| 843 |
+
"learning_rate": 2.3087854826980507e-06,
|
| 844 |
+
"log_odds_chosen": 4.625,
|
| 845 |
+
"log_odds_ratio": -0.078857421875,
|
| 846 |
+
"logits/chosen": -3.047656297683716,
|
| 847 |
+
"logits/rejected": 14.675000190734863,
|
| 848 |
+
"logps/chosen": -0.994921863079071,
|
| 849 |
+
"logps/rejected": -5.173437595367432,
|
| 850 |
+
"loss": 0.9971,
|
| 851 |
+
"nll_loss": 0.9925781488418579,
|
| 852 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 853 |
+
"rewards/chosen": -0.04975586012005806,
|
| 854 |
+
"rewards/margins": 0.20878906548023224,
|
| 855 |
+
"rewards/rejected": -0.25859373807907104,
|
| 856 |
"step": 470
|
| 857 |
},
|
| 858 |
{
|
| 859 |
"epoch": 0.6193548387096774,
|
| 860 |
+
"grad_norm": 7.97198152793078,
|
| 861 |
+
"learning_rate": 2.2845583119247696e-06,
|
| 862 |
+
"log_odds_chosen": 5.15625,
|
| 863 |
+
"log_odds_ratio": -0.043426513671875,
|
| 864 |
+
"logits/chosen": -2.921093702316284,
|
| 865 |
+
"logits/rejected": 15.574999809265137,
|
| 866 |
+
"logps/chosen": -0.9105468988418579,
|
| 867 |
+
"logps/rejected": -5.540625095367432,
|
| 868 |
+
"loss": 0.9281,
|
| 869 |
+
"nll_loss": 0.926562488079071,
|
| 870 |
+
"rewards/accuracies": 0.987500011920929,
|
| 871 |
+
"rewards/chosen": -0.04550781100988388,
|
| 872 |
+
"rewards/margins": 0.23164062201976776,
|
| 873 |
+
"rewards/rejected": -0.27714842557907104,
|
| 874 |
"step": 480
|
| 875 |
},
|
| 876 |
{
|
| 877 |
"epoch": 0.632258064516129,
|
| 878 |
+
"grad_norm": 8.119188050464615,
|
| 879 |
+
"learning_rate": 2.2610781582306733e-06,
|
| 880 |
+
"log_odds_chosen": 4.90625,
|
| 881 |
+
"log_odds_ratio": -0.04886474460363388,
|
| 882 |
+
"logits/chosen": -3.275390625,
|
| 883 |
+
"logits/rejected": 15.618749618530273,
|
| 884 |
+
"logps/chosen": -0.9800781011581421,
|
| 885 |
+
"logps/rejected": -5.409375190734863,
|
| 886 |
+
"loss": 0.9917,
|
| 887 |
+
"nll_loss": 0.989453136920929,
|
| 888 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 889 |
+
"rewards/chosen": -0.04899902269244194,
|
| 890 |
+
"rewards/margins": 0.22128906846046448,
|
| 891 |
+
"rewards/rejected": -0.27031248807907104,
|
| 892 |
"step": 490
|
| 893 |
},
|
| 894 |
{
|
| 895 |
"epoch": 0.6451612903225806,
|
| 896 |
+
"grad_norm": 7.067779726362276,
|
| 897 |
+
"learning_rate": 2.238307405179226e-06,
|
| 898 |
+
"log_odds_chosen": 4.7109375,
|
| 899 |
+
"log_odds_ratio": -0.06611327826976776,
|
| 900 |
+
"logits/chosen": -2.866406202316284,
|
| 901 |
+
"logits/rejected": 15.162500381469727,
|
| 902 |
+
"logps/chosen": -0.9976562261581421,
|
| 903 |
+
"logps/rejected": -5.243750095367432,
|
| 904 |
+
"loss": 1.0073,
|
| 905 |
+
"nll_loss": 1.0046875476837158,
|
| 906 |
+
"rewards/accuracies": 0.981249988079071,
|
| 907 |
+
"rewards/chosen": -0.04990234225988388,
|
| 908 |
+
"rewards/margins": 0.21210937201976776,
|
| 909 |
+
"rewards/rejected": -0.26201170682907104,
|
| 910 |
"step": 500
|
| 911 |
},
|
| 912 |
{
|
| 913 |
"epoch": 0.6580645161290323,
|
| 914 |
+
"grad_norm": 7.627869931724449,
|
| 915 |
+
"learning_rate": 2.2162110358896814e-06,
|
| 916 |
+
"log_odds_chosen": 5.0625,
|
| 917 |
+
"log_odds_ratio": -0.04056396335363388,
|
| 918 |
+
"logits/chosen": -3.075000047683716,
|
| 919 |
+
"logits/rejected": 15.043749809265137,
|
| 920 |
+
"logps/chosen": -0.9144531488418579,
|
| 921 |
+
"logps/rejected": -5.449999809265137,
|
| 922 |
+
"loss": 0.9441,
|
| 923 |
+
"nll_loss": 0.943359375,
|
| 924 |
"rewards/accuracies": 0.987500011920929,
|
| 925 |
+
"rewards/chosen": -0.04575195163488388,
|
| 926 |
+
"rewards/margins": 0.22675780951976776,
|
| 927 |
+
"rewards/rejected": -0.27275389432907104,
|
| 928 |
"step": 510
|
| 929 |
},
|
| 930 |
{
|
| 931 |
"epoch": 0.6709677419354839,
|
| 932 |
+
"grad_norm": 7.44284861722793,
|
| 933 |
+
"learning_rate": 2.1947564065307354e-06,
|
| 934 |
+
"log_odds_chosen": 4.876562595367432,
|
| 935 |
+
"log_odds_ratio": -0.04053955152630806,
|
| 936 |
+
"logits/chosen": -3.2132811546325684,
|
| 937 |
+
"logits/rejected": 15.425000190734863,
|
| 938 |
+
"logps/chosen": -0.913281261920929,
|
| 939 |
+
"logps/rejected": -5.275000095367432,
|
| 940 |
+
"loss": 0.9314,
|
| 941 |
+
"nll_loss": 0.9292968511581421,
|
| 942 |
"rewards/accuracies": 0.9937499761581421,
|
| 943 |
+
"rewards/chosen": -0.045654296875,
|
| 944 |
+
"rewards/margins": 0.21787109971046448,
|
| 945 |
+
"rewards/rejected": -0.26347655057907104,
|
| 946 |
"step": 520
|
| 947 |
},
|
| 948 |
{
|
| 949 |
"epoch": 0.6838709677419355,
|
| 950 |
+
"grad_norm": 7.943793514721159,
|
| 951 |
+
"learning_rate": 2.173913043478261e-06,
|
| 952 |
+
"log_odds_chosen": 4.599999904632568,
|
| 953 |
+
"log_odds_ratio": -0.06662597507238388,
|
| 954 |
+
"logits/chosen": -2.092578172683716,
|
| 955 |
+
"logits/rejected": 14.931249618530273,
|
| 956 |
+
"logps/chosen": -0.911328136920929,
|
| 957 |
+
"logps/rejected": -5.003125190734863,
|
| 958 |
+
"loss": 0.9341,
|
| 959 |
+
"nll_loss": 0.9312499761581421,
|
| 960 |
+
"rewards/accuracies": 0.987500011920929,
|
| 961 |
+
"rewards/chosen": -0.04555664211511612,
|
| 962 |
+
"rewards/margins": 0.20468750596046448,
|
| 963 |
+
"rewards/rejected": -0.25029295682907104,
|
| 964 |
"step": 530
|
| 965 |
},
|
| 966 |
{
|
| 967 |
"epoch": 0.6967741935483871,
|
| 968 |
+
"grad_norm": 6.647015764786715,
|
| 969 |
+
"learning_rate": 2.1536524612697402e-06,
|
| 970 |
+
"log_odds_chosen": 4.793749809265137,
|
| 971 |
+
"log_odds_ratio": -0.05656128004193306,
|
| 972 |
+
"logits/chosen": -1.971093773841858,
|
| 973 |
+
"logits/rejected": 16.006250381469727,
|
| 974 |
+
"logps/chosen": -0.9496093988418579,
|
| 975 |
+
"logps/rejected": -5.231249809265137,
|
| 976 |
+
"loss": 0.9388,
|
| 977 |
+
"nll_loss": 0.9359375238418579,
|
| 978 |
+
"rewards/accuracies": 0.981249988079071,
|
| 979 |
+
"rewards/chosen": -0.0474853515625,
|
| 980 |
+
"rewards/margins": 0.21416015923023224,
|
| 981 |
+
"rewards/rejected": -0.26152342557907104,
|
| 982 |
"step": 540
|
| 983 |
},
|
| 984 |
{
|
| 985 |
"epoch": 0.7096774193548387,
|
| 986 |
+
"grad_norm": 8.078904372917599,
|
| 987 |
+
"learning_rate": 2.1339479988816e-06,
|
| 988 |
+
"log_odds_chosen": 4.9375,
|
| 989 |
+
"log_odds_ratio": -0.03520507737994194,
|
| 990 |
+
"logits/chosen": -2.5245604515075684,
|
| 991 |
+
"logits/rejected": 15.412500381469727,
|
| 992 |
+
"logps/chosen": -0.9507812261581421,
|
| 993 |
+
"logps/rejected": -5.384375095367432,
|
| 994 |
+
"loss": 0.9523,
|
| 995 |
+
"nll_loss": 0.9515625238418579,
|
| 996 |
"rewards/accuracies": 0.987500011920929,
|
| 997 |
+
"rewards/chosen": -0.04746093600988388,
|
| 998 |
+
"rewards/margins": 0.22158202528953552,
|
| 999 |
+
"rewards/rejected": -0.26923829317092896,
|
| 1000 |
"step": 550
|
| 1001 |
},
|
| 1002 |
{
|
| 1003 |
"epoch": 0.7225806451612903,
|
| 1004 |
+
"grad_norm": 7.439562671796703,
|
| 1005 |
+
"learning_rate": 2.114774672189068e-06,
|
| 1006 |
+
"log_odds_chosen": 4.993750095367432,
|
| 1007 |
+
"log_odds_ratio": -0.044708251953125,
|
| 1008 |
+
"logits/chosen": -3.035937547683716,
|
| 1009 |
+
"logits/rejected": 15.199999809265137,
|
| 1010 |
+
"logps/chosen": -0.9527343511581421,
|
| 1011 |
+
"logps/rejected": -5.446875095367432,
|
| 1012 |
+
"loss": 0.9692,
|
| 1013 |
+
"nll_loss": 0.966796875,
|
| 1014 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1015 |
+
"rewards/chosen": -0.04768066480755806,
|
| 1016 |
+
"rewards/margins": 0.22470703721046448,
|
| 1017 |
+
"rewards/rejected": -0.27226561307907104,
|
| 1018 |
"step": 560
|
| 1019 |
},
|
| 1020 |
{
|
| 1021 |
"epoch": 0.7354838709677419,
|
| 1022 |
+
"grad_norm": 6.850420581812698,
|
| 1023 |
+
"learning_rate": 2.0961090407515926e-06,
|
| 1024 |
+
"log_odds_chosen": 5.012499809265137,
|
| 1025 |
+
"log_odds_ratio": -0.03801269456744194,
|
| 1026 |
+
"logits/chosen": -2.707812547683716,
|
| 1027 |
+
"logits/rejected": 15.393750190734863,
|
| 1028 |
+
"logps/chosen": -0.926953136920929,
|
| 1029 |
+
"logps/rejected": -5.425000190734863,
|
| 1030 |
+
"loss": 0.9428,
|
| 1031 |
+
"nll_loss": 0.94140625,
|
| 1032 |
"rewards/accuracies": 0.9937499761581421,
|
| 1033 |
+
"rewards/chosen": -0.04636230319738388,
|
| 1034 |
+
"rewards/margins": 0.22451171278953552,
|
| 1035 |
+
"rewards/rejected": -0.2706054747104645,
|
| 1036 |
"step": 570
|
| 1037 |
},
|
| 1038 |
{
|
| 1039 |
"epoch": 0.7483870967741936,
|
| 1040 |
+
"grad_norm": 7.650991563757063,
|
| 1041 |
+
"learning_rate": 2.077929087308457e-06,
|
| 1042 |
+
"log_odds_chosen": 4.650000095367432,
|
| 1043 |
+
"log_odds_ratio": -0.04947509616613388,
|
| 1044 |
+
"logits/chosen": -3.3359375,
|
| 1045 |
+
"logits/rejected": 14.625,
|
| 1046 |
+
"logps/chosen": -0.8941406011581421,
|
| 1047 |
+
"logps/rejected": -5.009375095367432,
|
| 1048 |
+
"loss": 0.9181,
|
| 1049 |
+
"nll_loss": 0.916015625,
|
| 1050 |
"rewards/accuracies": 0.981249988079071,
|
| 1051 |
+
"rewards/chosen": -0.04475097730755806,
|
| 1052 |
+
"rewards/margins": 0.2060546875,
|
| 1053 |
+
"rewards/rejected": -0.25048828125,
|
| 1054 |
"step": 580
|
| 1055 |
},
|
| 1056 |
{
|
| 1057 |
"epoch": 0.7612903225806451,
|
| 1058 |
+
"grad_norm": 7.115648653428975,
|
| 1059 |
+
"learning_rate": 2.0602141085758233e-06,
|
| 1060 |
+
"log_odds_chosen": 4.795312404632568,
|
| 1061 |
+
"log_odds_ratio": -0.04981689527630806,
|
| 1062 |
+
"logits/chosen": -3.2984375953674316,
|
| 1063 |
+
"logits/rejected": 15.037500381469727,
|
| 1064 |
+
"logps/chosen": -0.912890613079071,
|
| 1065 |
+
"logps/rejected": -5.192187309265137,
|
| 1066 |
+
"loss": 0.9367,
|
| 1067 |
+
"nll_loss": 0.9339843988418579,
|
| 1068 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1069 |
+
"rewards/chosen": -0.045654296875,
|
| 1070 |
+
"rewards/margins": 0.21416015923023224,
|
| 1071 |
+
"rewards/rejected": -0.25947266817092896,
|
| 1072 |
"step": 590
|
| 1073 |
},
|
| 1074 |
{
|
| 1075 |
"epoch": 0.7741935483870968,
|
| 1076 |
+
"grad_norm": 6.8341435054086555,
|
| 1077 |
+
"learning_rate": 2.0429446161135922e-06,
|
| 1078 |
+
"log_odds_chosen": 4.859375,
|
| 1079 |
+
"log_odds_ratio": -0.04189453274011612,
|
| 1080 |
+
"logits/chosen": -2.752734422683716,
|
| 1081 |
+
"logits/rejected": 15.381250381469727,
|
| 1082 |
+
"logps/chosen": -0.94921875,
|
| 1083 |
+
"logps/rejected": -5.303124904632568,
|
| 1084 |
+
"loss": 0.9627,
|
| 1085 |
+
"nll_loss": 0.9609375,
|
| 1086 |
"rewards/accuracies": 0.9937499761581421,
|
| 1087 |
+
"rewards/chosen": -0.04743652418255806,
|
| 1088 |
+
"rewards/margins": 0.2177734375,
|
| 1089 |
+
"rewards/rejected": -0.26494139432907104,
|
| 1090 |
"step": 600
|
| 1091 |
},
|
| 1092 |
{
|
| 1093 |
"epoch": 0.7870967741935484,
|
| 1094 |
+
"grad_norm": 7.6844935408533495,
|
| 1095 |
+
"learning_rate": 2.0261022461827697e-06,
|
| 1096 |
+
"log_odds_chosen": 5.337500095367432,
|
| 1097 |
+
"log_odds_ratio": -0.03539428859949112,
|
| 1098 |
+
"logits/chosen": -3.785937547683716,
|
| 1099 |
+
"logits/rejected": 15.237500190734863,
|
| 1100 |
+
"logps/chosen": -0.9476562738418579,
|
| 1101 |
+
"logps/rejected": -5.771874904632568,
|
| 1102 |
+
"loss": 0.9636,
|
| 1103 |
+
"nll_loss": 0.962109386920929,
|
| 1104 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1105 |
+
"rewards/chosen": -0.04733886569738388,
|
| 1106 |
+
"rewards/margins": 0.2412109375,
|
| 1107 |
+
"rewards/rejected": -0.28857421875,
|
| 1108 |
"step": 610
|
| 1109 |
},
|
| 1110 |
{
|
| 1111 |
"epoch": 0.8,
|
| 1112 |
+
"grad_norm": 8.392646613992586,
|
| 1113 |
+
"learning_rate": 2.009669677645352e-06,
|
| 1114 |
+
"log_odds_chosen": 4.953125,
|
| 1115 |
+
"log_odds_ratio": -0.048187255859375,
|
| 1116 |
+
"logits/chosen": -3.5023436546325684,
|
| 1117 |
+
"logits/rejected": 15.518750190734863,
|
| 1118 |
+
"logps/chosen": -0.9253906011581421,
|
| 1119 |
+
"logps/rejected": -5.353125095367432,
|
| 1120 |
+
"loss": 0.9431,
|
| 1121 |
+
"nll_loss": 0.94140625,
|
| 1122 |
"rewards/accuracies": 0.981249988079071,
|
| 1123 |
+
"rewards/chosen": -0.04624023288488388,
|
| 1124 |
+
"rewards/margins": 0.22138671576976776,
|
| 1125 |
+
"rewards/rejected": -0.26777344942092896,
|
| 1126 |
"step": 620
|
| 1127 |
},
|
| 1128 |
{
|
| 1129 |
"epoch": 0.8129032258064516,
|
| 1130 |
+
"grad_norm": 7.4781250651760205,
|
| 1131 |
+
"learning_rate": 1.9936305570722503e-06,
|
| 1132 |
+
"log_odds_chosen": 4.815625190734863,
|
| 1133 |
+
"log_odds_ratio": -0.04139404371380806,
|
| 1134 |
+
"logits/chosen": -3.1523194313049316,
|
| 1135 |
+
"logits/rejected": 15.456250190734863,
|
| 1136 |
+
"logps/chosen": -0.9039062261581421,
|
| 1137 |
+
"logps/rejected": -5.181250095367432,
|
| 1138 |
+
"loss": 0.9141,
|
| 1139 |
+
"nll_loss": 0.912890613079071,
|
| 1140 |
+
"rewards/accuracies": 1.0,
|
| 1141 |
+
"rewards/chosen": -0.04514160007238388,
|
| 1142 |
+
"rewards/margins": 0.2138671875,
|
| 1143 |
+
"rewards/rejected": -0.2591796815395355,
|
| 1144 |
"step": 630
|
| 1145 |
},
|
| 1146 |
{
|
| 1147 |
"epoch": 0.8258064516129032,
|
| 1148 |
+
"grad_norm": 8.77906696889682,
|
| 1149 |
+
"learning_rate": 1.977969430323089e-06,
|
| 1150 |
+
"log_odds_chosen": 4.981249809265137,
|
| 1151 |
+
"log_odds_ratio": -0.035614013671875,
|
| 1152 |
+
"logits/chosen": -2.821337938308716,
|
| 1153 |
+
"logits/rejected": 15.518750190734863,
|
| 1154 |
+
"logps/chosen": -0.9605468511581421,
|
| 1155 |
+
"logps/rejected": -5.434374809265137,
|
| 1156 |
+
"loss": 0.973,
|
| 1157 |
+
"nll_loss": 0.97265625,
|
| 1158 |
+
"rewards/accuracies": 1.0,
|
| 1159 |
+
"rewards/chosen": -0.04794921725988388,
|
| 1160 |
+
"rewards/margins": 0.22382812201976776,
|
| 1161 |
+
"rewards/rejected": -0.27177733182907104,
|
| 1162 |
"step": 640
|
| 1163 |
},
|
| 1164 |
{
|
| 1165 |
"epoch": 0.8387096774193549,
|
| 1166 |
+
"grad_norm": 7.337971445424149,
|
| 1167 |
+
"learning_rate": 1.962671679947149e-06,
|
| 1168 |
+
"log_odds_chosen": 5.328125,
|
| 1169 |
+
"log_odds_ratio": -0.03910522535443306,
|
| 1170 |
+
"logits/chosen": -3.082812547683716,
|
| 1171 |
+
"logits/rejected": 16.075000762939453,
|
| 1172 |
+
"logps/chosen": -0.962109386920929,
|
| 1173 |
+
"logps/rejected": -5.787499904632568,
|
| 1174 |
+
"loss": 0.9758,
|
| 1175 |
+
"nll_loss": 0.974609375,
|
| 1176 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1177 |
+
"rewards/chosen": -0.04814453050494194,
|
| 1178 |
+
"rewards/margins": 0.2412109375,
|
| 1179 |
+
"rewards/rejected": -0.2894531190395355,
|
| 1180 |
"step": 650
|
| 1181 |
},
|
| 1182 |
{
|
| 1183 |
"epoch": 0.8516129032258064,
|
| 1184 |
+
"grad_norm": 7.439098375728625,
|
| 1185 |
+
"learning_rate": 1.9477234678290497e-06,
|
| 1186 |
+
"log_odds_chosen": 4.9140625,
|
| 1187 |
+
"log_odds_ratio": -0.04149780422449112,
|
| 1188 |
+
"logits/chosen": -2.4859375953674316,
|
| 1189 |
+
"logits/rejected": 15.40625,
|
| 1190 |
+
"logps/chosen": -0.982421875,
|
| 1191 |
+
"logps/rejected": -5.409375190734863,
|
| 1192 |
+
"loss": 0.9789,
|
| 1193 |
+
"nll_loss": 0.977343738079071,
|
| 1194 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1195 |
+
"rewards/chosen": -0.049072265625,
|
| 1196 |
+
"rewards/margins": 0.22138671576976776,
|
| 1197 |
+
"rewards/rejected": -0.2705078125,
|
| 1198 |
"step": 660
|
| 1199 |
},
|
| 1200 |
{
|
| 1201 |
"epoch": 0.864516129032258,
|
| 1202 |
+
"grad_norm": 7.0935668026210585,
|
| 1203 |
+
"learning_rate": 1.9331116825676476e-06,
|
| 1204 |
+
"log_odds_chosen": 4.962500095367432,
|
| 1205 |
+
"log_odds_ratio": -0.04017334058880806,
|
| 1206 |
+
"logits/chosen": -3.327343702316284,
|
| 1207 |
+
"logits/rejected": 15.162500381469727,
|
| 1208 |
+
"logps/chosen": -0.940625011920929,
|
| 1209 |
+
"logps/rejected": -5.387499809265137,
|
| 1210 |
+
"loss": 0.9549,
|
| 1211 |
+
"nll_loss": 0.953125,
|
| 1212 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1213 |
+
"rewards/chosen": -0.04702148586511612,
|
| 1214 |
+
"rewards/margins": 0.22285155951976776,
|
| 1215 |
+
"rewards/rejected": -0.2696289122104645,
|
| 1216 |
"step": 670
|
| 1217 |
},
|
| 1218 |
{
|
| 1219 |
"epoch": 0.8774193548387097,
|
| 1220 |
+
"grad_norm": 8.300635787093205,
|
| 1221 |
+
"learning_rate": 1.9188238911333405e-06,
|
| 1222 |
+
"log_odds_chosen": 5.215624809265137,
|
| 1223 |
+
"log_odds_ratio": -0.02827148512005806,
|
| 1224 |
+
"logits/chosen": -2.7085938453674316,
|
| 1225 |
+
"logits/rejected": 15.381250381469727,
|
| 1226 |
+
"logps/chosen": -0.9664062261581421,
|
| 1227 |
+
"logps/rejected": -5.675000190734863,
|
| 1228 |
+
"loss": 0.9702,
|
| 1229 |
+
"nll_loss": 0.9691406488418579,
|
| 1230 |
+
"rewards/accuracies": 1.0,
|
| 1231 |
+
"rewards/chosen": -0.04838867112994194,
|
| 1232 |
+
"rewards/margins": 0.23544922471046448,
|
| 1233 |
+
"rewards/rejected": -0.28369140625,
|
| 1234 |
"step": 680
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
"epoch": 0.8903225806451613,
|
| 1238 |
+
"grad_norm": 8.215237709287992,
|
| 1239 |
+
"learning_rate": 1.9048482943986487e-06,
|
| 1240 |
+
"log_odds_chosen": 4.873437404632568,
|
| 1241 |
+
"log_odds_ratio": -0.06497802585363388,
|
| 1242 |
+
"logits/chosen": -2.8578124046325684,
|
| 1243 |
+
"logits/rejected": 14.987500190734863,
|
| 1244 |
+
"logps/chosen": -0.9417968988418579,
|
| 1245 |
+
"logps/rejected": -5.318749904632568,
|
| 1246 |
+
"loss": 0.9407,
|
| 1247 |
+
"nll_loss": 0.938281238079071,
|
| 1248 |
"rewards/accuracies": 0.9750000238418579,
|
| 1249 |
+
"rewards/chosen": -0.04704589769244194,
|
| 1250 |
+
"rewards/margins": 0.21845702826976776,
|
| 1251 |
+
"rewards/rejected": -0.265625,
|
| 1252 |
"step": 690
|
| 1253 |
},
|
| 1254 |
{
|
| 1255 |
"epoch": 0.9032258064516129,
|
| 1256 |
+
"grad_norm": 7.758338411265055,
|
| 1257 |
+
"learning_rate": 1.8911736861805847e-06,
|
| 1258 |
+
"log_odds_chosen": 4.796875,
|
| 1259 |
+
"log_odds_ratio": -0.04442138597369194,
|
| 1260 |
+
"logits/chosen": -2.7476563453674316,
|
| 1261 |
+
"logits/rejected": 15.206250190734863,
|
| 1262 |
+
"logps/chosen": -0.9046875238418579,
|
| 1263 |
+
"logps/rejected": -5.175000190734863,
|
| 1264 |
+
"loss": 0.918,
|
| 1265 |
+
"nll_loss": 0.9164062738418579,
|
| 1266 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1267 |
+
"rewards/chosen": -0.0452880859375,
|
| 1268 |
+
"rewards/margins": 0.21328124403953552,
|
| 1269 |
+
"rewards/rejected": -0.25859373807907104,
|
| 1270 |
"step": 700
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 0.9161290322580645,
|
| 1274 |
+
"grad_norm": 7.757899950179007,
|
| 1275 |
+
"learning_rate": 1.8777894154716994e-06,
|
| 1276 |
+
"log_odds_chosen": 5.040625095367432,
|
| 1277 |
+
"log_odds_ratio": -0.05682983249425888,
|
| 1278 |
+
"logits/chosen": -3.3187499046325684,
|
| 1279 |
+
"logits/rejected": 15.112500190734863,
|
| 1280 |
+
"logps/chosen": -0.9300781488418579,
|
| 1281 |
+
"logps/rejected": -5.46875,
|
| 1282 |
+
"loss": 0.9523,
|
| 1283 |
+
"nll_loss": 0.949999988079071,
|
| 1284 |
"rewards/accuracies": 0.9750000238418579,
|
| 1285 |
+
"rewards/chosen": -0.04653320461511612,
|
| 1286 |
+
"rewards/margins": 0.22666016221046448,
|
| 1287 |
+
"rewards/rejected": -0.2730468809604645,
|
| 1288 |
"step": 710
|
| 1289 |
},
|
| 1290 |
{
|
| 1291 |
"epoch": 0.9290322580645162,
|
| 1292 |
+
"grad_norm": 6.925223393286465,
|
| 1293 |
+
"learning_rate": 1.864685351570504e-06,
|
| 1294 |
+
"log_odds_chosen": 5.334374904632568,
|
| 1295 |
+
"log_odds_ratio": -0.03135986253619194,
|
| 1296 |
+
"logits/chosen": -2.9644532203674316,
|
| 1297 |
+
"logits/rejected": 15.293749809265137,
|
| 1298 |
+
"logps/chosen": -0.904296875,
|
| 1299 |
+
"logps/rejected": -5.706250190734863,
|
| 1300 |
+
"loss": 0.9248,
|
| 1301 |
+
"nll_loss": 0.9242187738418579,
|
| 1302 |
"rewards/accuracies": 1.0,
|
| 1303 |
+
"rewards/chosen": -0.04521484300494194,
|
| 1304 |
+
"rewards/margins": 0.24013671278953552,
|
| 1305 |
+
"rewards/rejected": -0.2855468690395355,
|
| 1306 |
"step": 720
|
| 1307 |
},
|
| 1308 |
{
|
| 1309 |
"epoch": 0.9419354838709677,
|
| 1310 |
+
"grad_norm": 8.560013543994318,
|
| 1311 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 1312 |
+
"log_odds_chosen": 4.846875190734863,
|
| 1313 |
+
"log_odds_ratio": -0.04491577297449112,
|
| 1314 |
+
"logits/chosen": -3.2242188453674316,
|
| 1315 |
+
"logits/rejected": 15.106249809265137,
|
| 1316 |
+
"logps/chosen": -0.938671886920929,
|
| 1317 |
+
"logps/rejected": -5.278124809265137,
|
| 1318 |
+
"loss": 0.9527,
|
| 1319 |
+
"nll_loss": 0.951953113079071,
|
| 1320 |
+
"rewards/accuracies": 0.981249988079071,
|
| 1321 |
+
"rewards/chosen": -0.04694824293255806,
|
| 1322 |
+
"rewards/margins": 0.21738281846046448,
|
| 1323 |
+
"rewards/rejected": -0.26435548067092896,
|
| 1324 |
"step": 730
|
| 1325 |
},
|
| 1326 |
{
|
| 1327 |
"epoch": 0.9548387096774194,
|
| 1328 |
+
"grad_norm": 7.44645859310936,
|
| 1329 |
+
"learning_rate": 1.8392797319442742e-06,
|
| 1330 |
+
"log_odds_chosen": 4.715624809265137,
|
| 1331 |
+
"log_odds_ratio": -0.05336914211511612,
|
| 1332 |
+
"logits/chosen": -3.6304688453674316,
|
| 1333 |
+
"logits/rejected": 14.675000190734863,
|
| 1334 |
+
"logps/chosen": -0.896484375,
|
| 1335 |
+
"logps/rejected": -5.090624809265137,
|
| 1336 |
+
"loss": 0.903,
|
| 1337 |
+
"nll_loss": 0.9007812738418579,
|
| 1338 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 1339 |
+
"rewards/chosen": -0.0447998046875,
|
| 1340 |
+
"rewards/margins": 0.20957031846046448,
|
| 1341 |
+
"rewards/rejected": -0.25439453125,
|
| 1342 |
"step": 740
|
| 1343 |
},
|
| 1344 |
{
|
| 1345 |
"epoch": 0.967741935483871,
|
| 1346 |
+
"grad_norm": 7.033644961041026,
|
| 1347 |
+
"learning_rate": 1.8269602381046742e-06,
|
| 1348 |
+
"log_odds_chosen": 5.084374904632568,
|
| 1349 |
+
"log_odds_ratio": -0.04151611402630806,
|
| 1350 |
+
"logits/chosen": -3.184375047683716,
|
| 1351 |
+
"logits/rejected": 15.412500381469727,
|
| 1352 |
+
"logps/chosen": -0.9359375238418579,
|
| 1353 |
+
"logps/rejected": -5.5,
|
| 1354 |
+
"loss": 0.9548,
|
| 1355 |
+
"nll_loss": 0.9535156488418579,
|
| 1356 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1357 |
+
"rewards/chosen": -0.04672851413488388,
|
| 1358 |
+
"rewards/margins": 0.22841796278953552,
|
| 1359 |
+
"rewards/rejected": -0.27509766817092896,
|
| 1360 |
"step": 750
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
"epoch": 0.9806451612903225,
|
| 1364 |
+
"grad_norm": 7.302442214130208,
|
| 1365 |
+
"learning_rate": 1.8148850216015693e-06,
|
| 1366 |
+
"log_odds_chosen": 4.909375190734863,
|
| 1367 |
+
"log_odds_ratio": -0.05724792554974556,
|
| 1368 |
+
"logits/chosen": -2.991406202316284,
|
| 1369 |
+
"logits/rejected": 15.149999618530273,
|
| 1370 |
+
"logps/chosen": -0.930859386920929,
|
| 1371 |
+
"logps/rejected": -5.34375,
|
| 1372 |
+
"loss": 0.9487,
|
| 1373 |
+
"nll_loss": 0.9453125,
|
| 1374 |
"rewards/accuracies": 0.9750000238418579,
|
| 1375 |
+
"rewards/chosen": -0.04653320461511612,
|
| 1376 |
+
"rewards/margins": 0.22050781548023224,
|
| 1377 |
+
"rewards/rejected": -0.2671875059604645,
|
| 1378 |
"step": 760
|
| 1379 |
},
|
| 1380 |
{
|
| 1381 |
"epoch": 0.9935483870967742,
|
| 1382 |
+
"grad_norm": 25.02421316510306,
|
| 1383 |
+
"learning_rate": 1.803046114936547e-06,
|
| 1384 |
+
"log_odds_chosen": 5.053124904632568,
|
| 1385 |
+
"log_odds_ratio": -0.03767089918255806,
|
| 1386 |
+
"logits/chosen": -3.510937452316284,
|
| 1387 |
+
"logits/rejected": 15.0,
|
| 1388 |
+
"logps/chosen": -0.9378906488418579,
|
| 1389 |
+
"logps/rejected": -5.478125095367432,
|
| 1390 |
+
"loss": 0.9543,
|
| 1391 |
+
"nll_loss": 0.953906238079071,
|
| 1392 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1393 |
+
"rewards/chosen": -0.04685058444738388,
|
| 1394 |
+
"rewards/margins": 0.22675780951976776,
|
| 1395 |
+
"rewards/rejected": -0.27363282442092896,
|
| 1396 |
"step": 770
|
| 1397 |
},
|
| 1398 |
{
|
| 1399 |
"epoch": 1.0,
|
| 1400 |
"step": 775,
|
| 1401 |
"total_flos": 0.0,
|
| 1402 |
+
"train_loss": 1.0152948588709678,
|
| 1403 |
+
"train_runtime": 1543.1546,
|
| 1404 |
+
"train_samples_per_second": 8.035,
|
| 1405 |
+
"train_steps_per_second": 0.502
|
| 1406 |
}
|
| 1407 |
],
|
| 1408 |
"logging_steps": 10,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7032
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:db399f9f0b92b45825201b771756fe8762aa1a42b5166da039f07be2557212c4
|
| 3 |
size 7032
|