Model save
Browse files- README.md +9 -9
- all_results.json +4 -4
- train_results.json +4 -4
- trainer_state.json +947 -947
README.md
CHANGED
|
@@ -18,15 +18,15 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
-
- Loss:
|
| 22 |
-
- Rewards/chosen: -
|
| 23 |
-
- Rewards/rejected: -
|
| 24 |
- Rewards/accuracies: 0.7320
|
| 25 |
-
- Rewards/margins: 1.
|
| 26 |
-
- Logps/rejected: -3.
|
| 27 |
-
- Logps/chosen: -2.
|
| 28 |
-
- Logits/rejected: -
|
| 29 |
-
- Logits/chosen: -
|
| 30 |
|
| 31 |
## Model description
|
| 32 |
|
|
@@ -63,7 +63,7 @@ The following hyperparameters were used during training:
|
|
| 63 |
|
| 64 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
-
|
|
| 67 |
|
| 68 |
|
| 69 |
### Framework versions
|
|
|
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 1.2093
|
| 22 |
+
- Rewards/chosen: -5.8139
|
| 23 |
+
- Rewards/rejected: -7.6630
|
| 24 |
- Rewards/accuracies: 0.7320
|
| 25 |
+
- Rewards/margins: 1.8490
|
| 26 |
+
- Logps/rejected: -3.8315
|
| 27 |
+
- Logps/chosen: -2.9070
|
| 28 |
+
- Logits/rejected: -1.8947
|
| 29 |
+
- Logits/chosen: -1.8796
|
| 30 |
|
| 31 |
## Model description
|
| 32 |
|
|
|
|
| 63 |
|
| 64 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 1.267 | 0.8375 | 400 | 1.2093 | -5.8139 | -7.6630 | 0.7320 | 1.8490 | -3.8315 | -2.9070 | -1.8947 | -1.8796 |
|
| 67 |
|
| 68 |
|
| 69 |
### Framework versions
|
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 0.998691442030882,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 1.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 61135,
|
| 7 |
-
"train_samples_per_second": 2.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 0.998691442030882,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.3686470238167785,
|
| 5 |
+
"train_runtime": 22217.0339,
|
| 6 |
"train_samples": 61135,
|
| 7 |
+
"train_samples_per_second": 2.752,
|
| 8 |
+
"train_steps_per_second": 0.021
|
| 9 |
}
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 0.998691442030882,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 1.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 61135,
|
| 7 |
-
"train_samples_per_second": 2.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 0.998691442030882,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 1.3686470238167785,
|
| 5 |
+
"train_runtime": 22217.0339,
|
| 6 |
"train_samples": 61135,
|
| 7 |
+
"train_samples_per_second": 2.752,
|
| 8 |
+
"train_steps_per_second": 0.021
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -10,13 +10,13 @@
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.002093692750588851,
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 1.0416666666666666e-08,
|
| 15 |
"logits/chosen": -2.8280014991760254,
|
| 16 |
"logits/rejected": -2.8466408252716064,
|
| 17 |
"logps/chosen": -1.1081000566482544,
|
| 18 |
"logps/rejected": -1.146370530128479,
|
| 19 |
-
"loss": 1.
|
| 20 |
"rewards/accuracies": 0.46875,
|
| 21 |
"rewards/chosen": -2.216200113296509,
|
| 22 |
"rewards/margins": 0.0765407383441925,
|
|
@@ -25,1453 +25,1453 @@
|
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.010468463752944255,
|
| 28 |
-
"grad_norm":
|
| 29 |
"learning_rate": 5.208333333333333e-08,
|
| 30 |
-
"logits/chosen": -2.
|
| 31 |
-
"logits/rejected": -2.
|
| 32 |
-
"logps/chosen": -1.
|
| 33 |
-
"logps/rejected": -1.
|
| 34 |
-
"loss": 1.
|
| 35 |
"rewards/accuracies": 0.5390625,
|
| 36 |
-
"rewards/chosen": -2.
|
| 37 |
-
"rewards/margins": 0.
|
| 38 |
-
"rewards/rejected": -2.
|
| 39 |
"step": 5
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.02093692750588851,
|
| 43 |
-
"grad_norm":
|
| 44 |
"learning_rate": 1.0416666666666667e-07,
|
| 45 |
-
"logits/chosen": -2.
|
| 46 |
-
"logits/rejected": -2.
|
| 47 |
-
"logps/chosen": -1.
|
| 48 |
-
"logps/rejected": -1.
|
| 49 |
-
"loss": 1.
|
| 50 |
"rewards/accuracies": 0.581250011920929,
|
| 51 |
-
"rewards/chosen": -2.
|
| 52 |
-
"rewards/margins": 0.
|
| 53 |
-
"rewards/rejected": -2.
|
| 54 |
"step": 10
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 0.031405391258832765,
|
| 58 |
-
"grad_norm":
|
| 59 |
"learning_rate": 1.5624999999999999e-07,
|
| 60 |
-
"logits/chosen": -2.
|
| 61 |
-
"logits/rejected": -2.
|
| 62 |
-
"logps/chosen": -1.
|
| 63 |
-
"logps/rejected": -1.
|
| 64 |
-
"loss": 1.
|
| 65 |
"rewards/accuracies": 0.5375000238418579,
|
| 66 |
-
"rewards/chosen": -2.
|
| 67 |
-
"rewards/margins": 0.
|
| 68 |
-
"rewards/rejected": -2.
|
| 69 |
"step": 15
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.04187385501177702,
|
| 73 |
-
"grad_norm":
|
| 74 |
"learning_rate": 2.0833333333333333e-07,
|
| 75 |
-
"logits/chosen": -2.
|
| 76 |
-
"logits/rejected": -2.
|
| 77 |
-
"logps/chosen": -
|
| 78 |
-
"logps/rejected": -1.
|
| 79 |
-
"loss": 1.
|
| 80 |
-
"rewards/accuracies": 0.
|
| 81 |
-
"rewards/chosen": -
|
| 82 |
-
"rewards/margins": 0.
|
| 83 |
-
"rewards/rejected": -2.
|
| 84 |
"step": 20
|
| 85 |
},
|
| 86 |
{
|
| 87 |
"epoch": 0.05234231876472128,
|
| 88 |
-
"grad_norm":
|
| 89 |
"learning_rate": 2.604166666666667e-07,
|
| 90 |
-
"logits/chosen": -2.
|
| 91 |
-
"logits/rejected": -2.
|
| 92 |
-
"logps/chosen": -0.
|
| 93 |
-
"logps/rejected": -1.
|
| 94 |
-
"loss": 1.
|
| 95 |
-
"rewards/accuracies": 0.
|
| 96 |
-
"rewards/chosen": -1.
|
| 97 |
-
"rewards/margins": 0.
|
| 98 |
-
"rewards/rejected": -2.
|
| 99 |
"step": 25
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.06281078251766553,
|
| 103 |
-
"grad_norm":
|
| 104 |
"learning_rate": 3.1249999999999997e-07,
|
| 105 |
-
"logits/chosen": -2.
|
| 106 |
-
"logits/rejected": -2.
|
| 107 |
-
"logps/chosen": -0.
|
| 108 |
-
"logps/rejected": -0.
|
| 109 |
-
"loss": 1.
|
| 110 |
"rewards/accuracies": 0.550000011920929,
|
| 111 |
-
"rewards/chosen": -1.
|
| 112 |
-
"rewards/margins": 0.
|
| 113 |
-
"rewards/rejected": -1.
|
| 114 |
"step": 30
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.07327924627060979,
|
| 118 |
-
"grad_norm":
|
| 119 |
"learning_rate": 3.645833333333333e-07,
|
| 120 |
-
"logits/chosen": -2.
|
| 121 |
-
"logits/rejected": -2.
|
| 122 |
-
"logps/chosen": -0.
|
| 123 |
-
"logps/rejected": -1.
|
| 124 |
-
"loss": 1.
|
| 125 |
-
"rewards/accuracies": 0.
|
| 126 |
-
"rewards/chosen": -1.
|
| 127 |
-
"rewards/margins": 0.
|
| 128 |
-
"rewards/rejected": -2.
|
| 129 |
"step": 35
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.08374771002355404,
|
| 133 |
-
"grad_norm":
|
| 134 |
"learning_rate": 4.1666666666666667e-07,
|
| 135 |
-
"logits/chosen": -2.
|
| 136 |
-
"logits/rejected": -2.
|
| 137 |
-
"logps/chosen": -0.
|
| 138 |
-
"logps/rejected": -1.
|
| 139 |
-
"loss": 1.
|
| 140 |
-
"rewards/accuracies": 0.
|
| 141 |
-
"rewards/chosen": -1.
|
| 142 |
-
"rewards/margins": 0.
|
| 143 |
-
"rewards/rejected": -2.
|
| 144 |
"step": 40
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.0942161737764983,
|
| 148 |
-
"grad_norm":
|
| 149 |
"learning_rate": 4.6874999999999996e-07,
|
| 150 |
-
"logits/chosen": -2.
|
| 151 |
-
"logits/rejected": -2.
|
| 152 |
-
"logps/chosen": -0.
|
| 153 |
-
"logps/rejected": -1.
|
| 154 |
-
"loss": 1.
|
| 155 |
-
"rewards/accuracies": 0.
|
| 156 |
-
"rewards/chosen": -1.
|
| 157 |
-
"rewards/margins": 0.
|
| 158 |
-
"rewards/rejected": -2.
|
| 159 |
"step": 45
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.10468463752944256,
|
| 163 |
-
"grad_norm":
|
| 164 |
"learning_rate": 4.999731868769026e-07,
|
| 165 |
-
"logits/chosen": -2.
|
| 166 |
-
"logits/rejected": -2.
|
| 167 |
-
"logps/chosen": -0.
|
| 168 |
-
"logps/rejected": -1.
|
| 169 |
-
"loss": 1.
|
| 170 |
-
"rewards/accuracies": 0.
|
| 171 |
-
"rewards/chosen": -1.
|
| 172 |
-
"rewards/margins": 0.
|
| 173 |
-
"rewards/rejected": -2.
|
| 174 |
"step": 50
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.11515310128238682,
|
| 178 |
-
"grad_norm":
|
| 179 |
"learning_rate": 4.996716052911017e-07,
|
| 180 |
-
"logits/chosen": -2.
|
| 181 |
-
"logits/rejected": -2.
|
| 182 |
-
"logps/chosen": -1.
|
| 183 |
-
"logps/rejected": -1.
|
| 184 |
-
"loss": 1.
|
| 185 |
-
"rewards/accuracies": 0.
|
| 186 |
-
"rewards/chosen": -2.
|
| 187 |
-
"rewards/margins": 0.
|
| 188 |
-
"rewards/rejected": -2.
|
| 189 |
"step": 55
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.12562156503533106,
|
| 193 |
-
"grad_norm":
|
| 194 |
"learning_rate": 4.990353313429303e-07,
|
| 195 |
-
"logits/chosen": -2.
|
| 196 |
-
"logits/rejected": -2.
|
| 197 |
-
"logps/chosen": -1.
|
| 198 |
-
"logps/rejected": -1.
|
| 199 |
-
"loss": 1.
|
| 200 |
-
"rewards/accuracies": 0.
|
| 201 |
-
"rewards/chosen": -2.
|
| 202 |
-
"rewards/margins": 0.
|
| 203 |
-
"rewards/rejected": -2.
|
| 204 |
"step": 60
|
| 205 |
},
|
| 206 |
{
|
| 207 |
"epoch": 0.1360900287882753,
|
| 208 |
-
"grad_norm":
|
| 209 |
"learning_rate": 4.980652179769217e-07,
|
| 210 |
-
"logits/chosen": -2.
|
| 211 |
-
"logits/rejected": -2.
|
| 212 |
-
"logps/chosen": -
|
| 213 |
-
"logps/rejected": -1.
|
| 214 |
-
"loss": 1.
|
| 215 |
"rewards/accuracies": 0.59375,
|
| 216 |
-
"rewards/chosen": -
|
| 217 |
-
"rewards/margins": 0.
|
| 218 |
-
"rewards/rejected": -2.
|
| 219 |
"step": 65
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.14655849254121958,
|
| 223 |
-
"grad_norm":
|
| 224 |
"learning_rate": 4.967625656594781e-07,
|
| 225 |
-
"logits/chosen": -2.
|
| 226 |
-
"logits/rejected": -2.
|
| 227 |
-
"logps/chosen": -1.
|
| 228 |
-
"logps/rejected": -1.
|
| 229 |
-
"loss": 1.
|
| 230 |
-
"rewards/accuracies": 0.
|
| 231 |
-
"rewards/chosen": -2.
|
| 232 |
-
"rewards/margins": 0.
|
| 233 |
-
"rewards/rejected": -2.
|
| 234 |
"step": 70
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.15702695629416383,
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 4.951291206355559e-07,
|
| 240 |
-
"logits/chosen": -2.
|
| 241 |
-
"logits/rejected": -2.
|
| 242 |
-
"logps/chosen": -1.
|
| 243 |
-
"logps/rejected": -1.
|
| 244 |
-
"loss": 1.
|
| 245 |
-
"rewards/accuracies": 0.
|
| 246 |
-
"rewards/chosen": -2.
|
| 247 |
-
"rewards/margins": 0.
|
| 248 |
-
"rewards/rejected": -2.
|
| 249 |
"step": 75
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.16749542004710807,
|
| 253 |
-
"grad_norm":
|
| 254 |
"learning_rate": 4.93167072587771e-07,
|
| 255 |
-
"logits/chosen": -2.
|
| 256 |
-
"logits/rejected": -2.
|
| 257 |
-
"logps/chosen": -1.
|
| 258 |
-
"logps/rejected": -1.
|
| 259 |
-
"loss": 1.
|
| 260 |
"rewards/accuracies": 0.6875,
|
| 261 |
-
"rewards/chosen": -2.
|
| 262 |
-
"rewards/margins": 0.
|
| 263 |
-
"rewards/rejected": -2.
|
| 264 |
"step": 80
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 0.17796388380005235,
|
| 268 |
-
"grad_norm":
|
| 269 |
"learning_rate": 4.908790517010636e-07,
|
| 270 |
-
"logits/chosen": -2.
|
| 271 |
-
"logits/rejected": -2.
|
| 272 |
-
"logps/chosen": -1.
|
| 273 |
-
"logps/rejected": -1.
|
| 274 |
-
"loss": 1.
|
| 275 |
-
"rewards/accuracies": 0.
|
| 276 |
-
"rewards/chosen": -2.
|
| 277 |
-
"rewards/margins": 0.
|
| 278 |
-
"rewards/rejected": -2.
|
| 279 |
"step": 85
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.1884323475529966,
|
| 283 |
-
"grad_norm":
|
| 284 |
"learning_rate": 4.882681251368548e-07,
|
| 285 |
-
"logits/chosen": -2.
|
| 286 |
-
"logits/rejected": -2.
|
| 287 |
-
"logps/chosen": -1.
|
| 288 |
-
"logps/rejected": -1.
|
| 289 |
-
"loss": 1.
|
| 290 |
-
"rewards/accuracies": 0.
|
| 291 |
-
"rewards/chosen": -2.
|
| 292 |
-
"rewards/margins": 0.
|
| 293 |
-
"rewards/rejected": -
|
| 294 |
"step": 90
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 0.19890081130594087,
|
| 298 |
-
"grad_norm":
|
| 299 |
"learning_rate": 4.853377929214243e-07,
|
| 300 |
-
"logits/chosen": -2.
|
| 301 |
-
"logits/rejected": -2.
|
| 302 |
-
"logps/chosen": -1.
|
| 303 |
-
"logps/rejected": -1.
|
| 304 |
-
"loss": 1.
|
| 305 |
-
"rewards/accuracies": 0.
|
| 306 |
-
"rewards/chosen": -2.
|
| 307 |
-
"rewards/margins": 0.
|
| 308 |
-
"rewards/rejected": -
|
| 309 |
"step": 95
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.2093692750588851,
|
| 313 |
-
"grad_norm":
|
| 314 |
"learning_rate": 4.820919832540181e-07,
|
| 315 |
-
"logits/chosen": -2.
|
| 316 |
-
"logits/rejected": -2.
|
| 317 |
-
"logps/chosen": -1.
|
| 318 |
-
"logps/rejected": -1.
|
| 319 |
-
"loss": 1.
|
| 320 |
-
"rewards/accuracies": 0.
|
| 321 |
-
"rewards/chosen": -2.
|
| 322 |
-
"rewards/margins": 0.
|
| 323 |
-
"rewards/rejected": -2.
|
| 324 |
"step": 100
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 0.21983773881182936,
|
| 328 |
-
"grad_norm":
|
| 329 |
"learning_rate": 4.785350472409791e-07,
|
| 330 |
-
"logits/chosen": -2.
|
| 331 |
-
"logits/rejected": -2.
|
| 332 |
-
"logps/chosen": -1.
|
| 333 |
-
"logps/rejected": -1.
|
| 334 |
-
"loss": 1.
|
| 335 |
-
"rewards/accuracies": 0.
|
| 336 |
-
"rewards/chosen": -2.
|
| 337 |
-
"rewards/margins": 0.
|
| 338 |
-
"rewards/rejected": -
|
| 339 |
"step": 105
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.23030620256477363,
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 4.7467175306295647e-07,
|
| 345 |
-
"logits/chosen": -2.
|
| 346 |
-
"logits/rejected": -2.
|
| 347 |
-
"logps/chosen": -1.
|
| 348 |
-
"logps/rejected": -1.
|
| 349 |
-
"loss": 1.
|
| 350 |
-
"rewards/accuracies": 0.
|
| 351 |
-
"rewards/chosen": -2.
|
| 352 |
-
"rewards/margins": 0.
|
| 353 |
-
"rewards/rejected": -2.
|
| 354 |
"step": 110
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.24077466631771788,
|
| 358 |
-
"grad_norm":
|
| 359 |
"learning_rate": 4.70507279583015e-07,
|
| 360 |
-
"logits/chosen": -2.
|
| 361 |
-
"logits/rejected": -2.
|
| 362 |
-
"logps/chosen": -1.
|
| 363 |
-
"logps/rejected": -1.
|
| 364 |
-
"loss": 1.
|
| 365 |
-
"rewards/accuracies": 0.
|
| 366 |
-
"rewards/chosen": -2.
|
| 367 |
-
"rewards/margins": 0.
|
| 368 |
-
"rewards/rejected": -2.
|
| 369 |
"step": 115
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.2512431300706621,
|
| 373 |
-
"grad_norm":
|
| 374 |
"learning_rate": 4.6604720940421207e-07,
|
| 375 |
-
"logits/chosen": -2.
|
| 376 |
-
"logits/rejected": -2.
|
| 377 |
-
"logps/chosen": -1.
|
| 378 |
-
"logps/rejected": -1.
|
| 379 |
-
"loss": 1.
|
| 380 |
-
"rewards/accuracies": 0.
|
| 381 |
-
"rewards/chosen": -2.
|
| 382 |
-
"rewards/margins": 0.
|
| 383 |
-
"rewards/rejected": -
|
| 384 |
"step": 120
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 0.26171159382360637,
|
| 388 |
-
"grad_norm":
|
| 389 |
"learning_rate": 4.612975213859487e-07,
|
| 390 |
-
"logits/chosen": -2.
|
| 391 |
-
"logits/rejected": -2.
|
| 392 |
-
"logps/chosen": -1.
|
| 393 |
-
"logps/rejected": -1.
|
| 394 |
-
"loss": 1.
|
| 395 |
-
"rewards/accuracies": 0.
|
| 396 |
-
"rewards/chosen": -2.
|
| 397 |
-
"rewards/margins": 0.
|
| 398 |
-
"rewards/rejected": -
|
| 399 |
"step": 125
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.2721800575765506,
|
| 403 |
-
"grad_norm":
|
| 404 |
"learning_rate": 4.5626458262912735e-07,
|
| 405 |
-
"logits/chosen": -2.
|
| 406 |
-
"logits/rejected": -2.
|
| 407 |
-
"logps/chosen": -1.
|
| 408 |
-
"logps/rejected": -1.
|
| 409 |
-
"loss": 1.
|
| 410 |
-
"rewards/accuracies": 0.
|
| 411 |
-
"rewards/chosen": -2.
|
| 412 |
-
"rewards/margins": 0.
|
| 413 |
-
"rewards/rejected": -
|
| 414 |
"step": 130
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 0.2826485213294949,
|
| 418 |
-
"grad_norm":
|
| 419 |
"learning_rate": 4.5095513994085974e-07,
|
| 420 |
-
"logits/chosen": -2.
|
| 421 |
-
"logits/rejected": -2.
|
| 422 |
-
"logps/chosen": -1.
|
| 423 |
-
"logps/rejected": -1.
|
| 424 |
-
"loss": 1.
|
| 425 |
-
"rewards/accuracies": 0.
|
| 426 |
-
"rewards/chosen": -2.
|
| 427 |
-
"rewards/margins": 0.
|
| 428 |
-
"rewards/rejected": -
|
| 429 |
"step": 135
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.29311698508243916,
|
| 433 |
-
"grad_norm":
|
| 434 |
"learning_rate": 4.453763107901675e-07,
|
| 435 |
-
"logits/chosen": -2.
|
| 436 |
-
"logits/rejected": -2.
|
| 437 |
-
"logps/chosen": -1.
|
| 438 |
-
"logps/rejected": -1.
|
| 439 |
-
"loss": 1.
|
| 440 |
-
"rewards/accuracies": 0.
|
| 441 |
-
"rewards/chosen": -2.
|
| 442 |
-
"rewards/margins": 0.
|
| 443 |
-
"rewards/rejected": -3.
|
| 444 |
"step": 140
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.3035854488353834,
|
| 448 |
-
"grad_norm":
|
| 449 |
"learning_rate": 4.395355737667985e-07,
|
| 450 |
-
"logits/chosen": -2.
|
| 451 |
-
"logits/rejected": -2.
|
| 452 |
-
"logps/chosen": -1.
|
| 453 |
-
"logps/rejected": -1.
|
| 454 |
-
"loss": 1.
|
| 455 |
"rewards/accuracies": 0.675000011920929,
|
| 456 |
-
"rewards/chosen": -2.
|
| 457 |
-
"rewards/margins": 0.
|
| 458 |
-
"rewards/rejected": -3.
|
| 459 |
"step": 145
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.31405391258832765,
|
| 463 |
-
"grad_norm":
|
| 464 |
"learning_rate": 4.3344075855595097e-07,
|
| 465 |
-
"logits/chosen": -2.
|
| 466 |
-
"logits/rejected": -2.
|
| 467 |
-
"logps/chosen": -1.
|
| 468 |
-
"logps/rejected": -1.
|
| 469 |
-
"loss": 1.
|
| 470 |
-
"rewards/accuracies": 0.
|
| 471 |
-
"rewards/chosen": -2.
|
| 472 |
-
"rewards/margins": 0.
|
| 473 |
-
"rewards/rejected": -
|
| 474 |
"step": 150
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 0.3245223763412719,
|
| 478 |
-
"grad_norm":
|
| 479 |
"learning_rate": 4.271000354423425e-07,
|
| 480 |
-
"logits/chosen": -2.
|
| 481 |
-
"logits/rejected": -2.
|
| 482 |
-
"logps/chosen": -1.
|
| 483 |
-
"logps/rejected": -1.
|
| 484 |
-
"loss": 1.
|
| 485 |
"rewards/accuracies": 0.699999988079071,
|
| 486 |
-
"rewards/chosen": -2.
|
| 487 |
-
"rewards/margins":
|
| 488 |
-
"rewards/rejected": -3.
|
| 489 |
"step": 155
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.33499084009421615,
|
| 493 |
-
"grad_norm":
|
| 494 |
"learning_rate": 4.2052190435769554e-07,
|
| 495 |
-
"logits/chosen": -2.
|
| 496 |
-
"logits/rejected": -2.
|
| 497 |
-
"logps/chosen": -1.
|
| 498 |
-
"logps/rejected": -1.
|
| 499 |
-
"loss": 1.
|
| 500 |
-
"rewards/accuracies": 0.
|
| 501 |
-
"rewards/chosen": -2.
|
| 502 |
-
"rewards/margins": 1.
|
| 503 |
-
"rewards/rejected": -3.
|
| 504 |
"step": 160
|
| 505 |
},
|
| 506 |
{
|
| 507 |
"epoch": 0.34545930384716045,
|
| 508 |
-
"grad_norm":
|
| 509 |
"learning_rate": 4.137151834863213e-07,
|
| 510 |
-
"logits/chosen": -2.
|
| 511 |
-
"logits/rejected": -2.
|
| 512 |
-
"logps/chosen": -1.
|
| 513 |
-
"logps/rejected": -1.
|
| 514 |
-
"loss": 1.
|
| 515 |
-
"rewards/accuracies": 0.
|
| 516 |
-
"rewards/chosen": -2.
|
| 517 |
-
"rewards/margins": 0.
|
| 518 |
-
"rewards/rejected": -3.
|
| 519 |
"step": 165
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.3559277676001047,
|
| 523 |
-
"grad_norm":
|
| 524 |
"learning_rate": 4.0668899744407567e-07,
|
| 525 |
-
"logits/chosen": -2.
|
| 526 |
-
"logits/rejected": -2.
|
| 527 |
-
"logps/chosen": -1.
|
| 528 |
-
"logps/rejected": -1.
|
| 529 |
-
"loss": 1.
|
| 530 |
-
"rewards/accuracies": 0.
|
| 531 |
-
"rewards/chosen": -2.
|
| 532 |
-
"rewards/margins": 0.
|
| 533 |
-
"rewards/rejected": -3.
|
| 534 |
"step": 170
|
| 535 |
},
|
| 536 |
{
|
| 537 |
"epoch": 0.36639623135304894,
|
| 538 |
-
"grad_norm":
|
| 539 |
"learning_rate": 3.994527650465352e-07,
|
| 540 |
-
"logits/chosen": -2.
|
| 541 |
-
"logits/rejected": -2.
|
| 542 |
-
"logps/chosen": -1.
|
| 543 |
-
"logps/rejected": -1.
|
| 544 |
-
"loss": 1.
|
| 545 |
-
"rewards/accuracies": 0.
|
| 546 |
-
"rewards/chosen": -2.
|
| 547 |
-
"rewards/margins":
|
| 548 |
-
"rewards/rejected": -3.
|
| 549 |
"step": 175
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.3768646951059932,
|
| 553 |
-
"grad_norm":
|
| 554 |
"learning_rate": 3.920161866827889e-07,
|
| 555 |
-
"logits/chosen": -2.
|
| 556 |
-
"logits/rejected": -2.
|
| 557 |
-
"logps/chosen": -1.
|
| 558 |
-
"logps/rejected": -1.
|
| 559 |
-
"loss": 1.
|
| 560 |
-
"rewards/accuracies": 0.
|
| 561 |
-
"rewards/chosen": -2.
|
| 562 |
-
"rewards/margins":
|
| 563 |
-
"rewards/rejected": -3.
|
| 564 |
"step": 180
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.38733315885893743,
|
| 568 |
-
"grad_norm":
|
| 569 |
"learning_rate": 3.8438923131177237e-07,
|
| 570 |
-
"logits/chosen": -2.
|
| 571 |
-
"logits/rejected": -2.
|
| 572 |
-
"logps/chosen": -1.
|
| 573 |
-
"logps/rejected": -1.
|
| 574 |
-
"loss":
|
| 575 |
-
"rewards/accuracies": 0.
|
| 576 |
-
"rewards/chosen": -2.
|
| 577 |
-
"rewards/margins":
|
| 578 |
-
"rewards/rejected": -3.
|
| 579 |
"step": 185
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.39780162261188173,
|
| 583 |
-
"grad_norm":
|
| 584 |
"learning_rate": 3.765821230985757e-07,
|
| 585 |
-
"logits/chosen": -2.
|
| 586 |
-
"logits/rejected": -2.
|
| 587 |
-
"logps/chosen": -1.
|
| 588 |
-
"logps/rejected": -
|
| 589 |
-
"loss": 1.
|
| 590 |
-
"rewards/accuracies": 0.
|
| 591 |
-
"rewards/chosen": -2.
|
| 592 |
-
"rewards/margins":
|
| 593 |
-
"rewards/rejected": -
|
| 594 |
"step": 190
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.408270086364826,
|
| 598 |
-
"grad_norm":
|
| 599 |
"learning_rate": 3.6860532770864005e-07,
|
| 600 |
-
"logits/chosen": -2.
|
| 601 |
-
"logits/rejected": -2.
|
| 602 |
-
"logps/chosen": -1.
|
| 603 |
-
"logps/rejected": -
|
| 604 |
-
"loss": 1.
|
| 605 |
-
"rewards/accuracies": 0.
|
| 606 |
-
"rewards/chosen": -
|
| 607 |
-
"rewards/margins": 1.
|
| 608 |
-
"rewards/rejected": -
|
| 609 |
"step": 195
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.4187385501177702,
|
| 613 |
-
"grad_norm":
|
| 614 |
"learning_rate": 3.604695382782159e-07,
|
| 615 |
-
"logits/chosen": -2.
|
| 616 |
-
"logits/rejected": -2.
|
| 617 |
-
"logps/chosen": -1.
|
| 618 |
-
"logps/rejected": -
|
| 619 |
-
"loss": 1.
|
| 620 |
-
"rewards/accuracies": 0.
|
| 621 |
-
"rewards/chosen": -
|
| 622 |
-
"rewards/margins":
|
| 623 |
-
"rewards/rejected": -
|
| 624 |
"step": 200
|
| 625 |
},
|
| 626 |
{
|
| 627 |
"epoch": 0.42920701387071447,
|
| 628 |
-
"grad_norm":
|
| 629 |
"learning_rate": 3.5218566107988867e-07,
|
| 630 |
-
"logits/chosen": -2.
|
| 631 |
-
"logits/rejected": -2.
|
| 632 |
-
"logps/chosen": -1.
|
| 633 |
-
"logps/rejected": -2.
|
| 634 |
-
"loss": 1.
|
| 635 |
-
"rewards/accuracies": 0.
|
| 636 |
-
"rewards/chosen": -
|
| 637 |
-
"rewards/margins": 1.
|
| 638 |
-
"rewards/rejected": -
|
| 639 |
"step": 205
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.4396754776236587,
|
| 643 |
-
"grad_norm":
|
| 644 |
"learning_rate": 3.4376480090239047e-07,
|
| 645 |
-
"logits/chosen": -2.
|
| 646 |
-
"logits/rejected": -2.
|
| 647 |
-
"logps/chosen": -1.
|
| 648 |
-
"logps/rejected": -
|
| 649 |
-
"loss": 1.
|
| 650 |
"rewards/accuracies": 0.699999988079071,
|
| 651 |
-
"rewards/chosen": -3.
|
| 652 |
-
"rewards/margins": 0.
|
| 653 |
-
"rewards/rejected": -
|
| 654 |
"step": 210
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.45014394137660296,
|
| 658 |
-
"grad_norm":
|
| 659 |
"learning_rate": 3.3521824616429284e-07,
|
| 660 |
-
"logits/chosen": -2.
|
| 661 |
-
"logits/rejected": -2.
|
| 662 |
-
"logps/chosen": -1.
|
| 663 |
-
"logps/rejected": -2.
|
| 664 |
-
"loss": 1.
|
| 665 |
-
"rewards/accuracies": 0.
|
| 666 |
-
"rewards/chosen": -3.
|
| 667 |
-
"rewards/margins":
|
| 668 |
-
"rewards/rejected": -4.
|
| 669 |
"step": 215
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.46061240512954726,
|
| 673 |
-
"grad_norm": 49.
|
| 674 |
"learning_rate": 3.265574537815398e-07,
|
| 675 |
-
"logits/chosen": -2.
|
| 676 |
-
"logits/rejected": -2.
|
| 677 |
-
"logps/chosen": -1.
|
| 678 |
-
"logps/rejected": -2.
|
| 679 |
-
"loss":
|
| 680 |
-
"rewards/accuracies": 0.
|
| 681 |
-
"rewards/chosen": -3.
|
| 682 |
-
"rewards/margins": 1.
|
| 683 |
-
"rewards/rejected": -4.
|
| 684 |
"step": 220
|
| 685 |
},
|
| 686 |
{
|
| 687 |
"epoch": 0.4710808688824915,
|
| 688 |
-
"grad_norm": 55.
|
| 689 |
"learning_rate": 3.1779403380910425e-07,
|
| 690 |
-
"logits/chosen": -2.
|
| 691 |
-
"logits/rejected": -2.
|
| 692 |
-
"logps/chosen": -
|
| 693 |
-
"logps/rejected": -2.
|
| 694 |
-
"loss": 1.
|
| 695 |
-
"rewards/accuracies": 0.
|
| 696 |
-
"rewards/chosen": -
|
| 697 |
-
"rewards/margins":
|
| 698 |
-
"rewards/rejected": -
|
| 699 |
"step": 225
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.48154933263543576,
|
| 703 |
-
"grad_norm":
|
| 704 |
"learning_rate": 3.0893973387735683e-07,
|
| 705 |
-
"logits/chosen": -2.
|
| 706 |
-
"logits/rejected": -2.
|
| 707 |
-
"logps/chosen": -
|
| 708 |
-
"logps/rejected": -2.
|
| 709 |
-
"loss":
|
| 710 |
-
"rewards/accuracies": 0.
|
| 711 |
-
"rewards/chosen": -
|
| 712 |
-
"rewards/margins": 1.
|
| 713 |
-
"rewards/rejected": -
|
| 714 |
"step": 230
|
| 715 |
},
|
| 716 |
{
|
| 717 |
"epoch": 0.49201779638838,
|
| 718 |
-
"grad_norm":
|
| 719 |
"learning_rate": 3.000064234440111e-07,
|
| 720 |
-
"logits/chosen": -2.
|
| 721 |
-
"logits/rejected": -2.
|
| 722 |
-
"logps/chosen": -
|
| 723 |
-
"logps/rejected": -2.
|
| 724 |
-
"loss":
|
| 725 |
"rewards/accuracies": 0.706250011920929,
|
| 726 |
-
"rewards/chosen": -
|
| 727 |
-
"rewards/margins":
|
| 728 |
-
"rewards/rejected": -
|
| 729 |
"step": 235
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.5024862601413242,
|
| 733 |
-
"grad_norm":
|
| 734 |
"learning_rate": 2.910060778827554e-07,
|
| 735 |
-
"logits/chosen": -2.
|
| 736 |
-
"logits/rejected": -2.
|
| 737 |
-
"logps/chosen": -
|
| 738 |
-
"logps/rejected": -2.
|
| 739 |
-
"loss": 1.
|
| 740 |
"rewards/accuracies": 0.71875,
|
| 741 |
-
"rewards/chosen": -
|
| 742 |
-
"rewards/margins": 1.
|
| 743 |
-
"rewards/rejected": -
|
| 744 |
"step": 240
|
| 745 |
},
|
| 746 |
{
|
| 747 |
"epoch": 0.5129547238942685,
|
| 748 |
-
"grad_norm":
|
| 749 |
"learning_rate": 2.8195076242990116e-07,
|
| 750 |
-
"logits/chosen": -2.
|
| 751 |
-
"logits/rejected": -2.
|
| 752 |
-
"logps/chosen": -
|
| 753 |
-
"logps/rejected": -2.
|
| 754 |
-
"loss":
|
| 755 |
-
"rewards/accuracies": 0.
|
| 756 |
-
"rewards/chosen": -
|
| 757 |
-
"rewards/margins": 1.
|
| 758 |
-
"rewards/rejected": -
|
| 759 |
"step": 245
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.5234231876472127,
|
| 763 |
-
"grad_norm":
|
| 764 |
"learning_rate": 2.7285261601056697e-07,
|
| 765 |
-
"logits/chosen": -
|
| 766 |
-
"logits/rejected": -2.
|
| 767 |
-
"logps/chosen": -
|
| 768 |
-
"logps/rejected": -
|
| 769 |
-
"loss": 1.
|
| 770 |
-
"rewards/accuracies": 0.
|
| 771 |
-
"rewards/chosen": -
|
| 772 |
-
"rewards/margins": 1.
|
| 773 |
-
"rewards/rejected": -
|
| 774 |
"step": 250
|
| 775 |
},
|
| 776 |
{
|
| 777 |
"epoch": 0.533891651400157,
|
| 778 |
-
"grad_norm":
|
| 779 |
"learning_rate": 2.6372383496608186e-07,
|
| 780 |
-
"logits/chosen": -2.
|
| 781 |
-
"logits/rejected": -2.
|
| 782 |
-
"logps/chosen": -
|
| 783 |
-
"logps/rejected": -
|
| 784 |
-
"loss": 1.
|
| 785 |
-
"rewards/accuracies": 0.
|
| 786 |
-
"rewards/chosen": -
|
| 787 |
-
"rewards/margins": 1.
|
| 788 |
-
"rewards/rejected": -
|
| 789 |
"step": 255
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 0.5443601151531012,
|
| 793 |
-
"grad_norm":
|
| 794 |
"learning_rate": 2.5457665670441937e-07,
|
| 795 |
-
"logits/chosen": -
|
| 796 |
-
"logits/rejected": -2.
|
| 797 |
-
"logps/chosen": -
|
| 798 |
-
"logps/rejected": -
|
| 799 |
-
"loss":
|
| 800 |
-
"rewards/accuracies": 0.
|
| 801 |
-
"rewards/chosen": -
|
| 802 |
-
"rewards/margins": 1.
|
| 803 |
-
"rewards/rejected": -
|
| 804 |
"step": 260
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 0.5548285789060455,
|
| 808 |
-
"grad_norm":
|
| 809 |
"learning_rate": 2.454233432955807e-07,
|
| 810 |
-
"logits/chosen": -
|
| 811 |
-
"logits/rejected": -
|
| 812 |
-
"logps/chosen": -
|
| 813 |
-
"logps/rejected": -
|
| 814 |
-
"loss":
|
| 815 |
-
"rewards/accuracies": 0.
|
| 816 |
-
"rewards/chosen": -
|
| 817 |
-
"rewards/margins": 1.
|
| 818 |
-
"rewards/rejected": -
|
| 819 |
"step": 265
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 0.5652970426589898,
|
| 823 |
-
"grad_norm":
|
| 824 |
"learning_rate": 2.3627616503391812e-07,
|
| 825 |
-
"logits/chosen": -2.
|
| 826 |
-
"logits/rejected": -2.
|
| 827 |
-
"logps/chosen": -2.
|
| 828 |
-
"logps/rejected": -
|
| 829 |
-
"loss":
|
| 830 |
-
"rewards/accuracies": 0.
|
| 831 |
-
"rewards/chosen": -
|
| 832 |
-
"rewards/margins":
|
| 833 |
-
"rewards/rejected": -
|
| 834 |
"step": 270
|
| 835 |
},
|
| 836 |
{
|
| 837 |
"epoch": 0.575765506411934,
|
| 838 |
-
"grad_norm":
|
| 839 |
"learning_rate": 2.2714738398943308e-07,
|
| 840 |
-
"logits/chosen": -
|
| 841 |
-
"logits/rejected": -
|
| 842 |
-
"logps/chosen": -
|
| 843 |
-
"logps/rejected": -
|
| 844 |
-
"loss":
|
| 845 |
-
"rewards/accuracies": 0.
|
| 846 |
-
"rewards/chosen": -
|
| 847 |
-
"rewards/margins": 1.
|
| 848 |
-
"rewards/rejected": -
|
| 849 |
"step": 275
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 0.5862339701648783,
|
| 853 |
-
"grad_norm":
|
| 854 |
"learning_rate": 2.1804923757009882e-07,
|
| 855 |
-
"logits/chosen": -
|
| 856 |
-
"logits/rejected": -
|
| 857 |
-
"logps/chosen": -
|
| 858 |
-
"logps/rejected": -
|
| 859 |
-
"loss": 1.
|
| 860 |
-
"rewards/accuracies": 0.
|
| 861 |
-
"rewards/chosen": -
|
| 862 |
-
"rewards/margins": 1.
|
| 863 |
-
"rewards/rejected": -
|
| 864 |
"step": 280
|
| 865 |
},
|
| 866 |
{
|
| 867 |
"epoch": 0.5967024339178225,
|
| 868 |
-
"grad_norm":
|
| 869 |
"learning_rate": 2.089939221172446e-07,
|
| 870 |
-
"logits/chosen": -
|
| 871 |
-
"logits/rejected": -
|
| 872 |
-
"logps/chosen": -
|
| 873 |
-
"logps/rejected": -
|
| 874 |
-
"loss": 1.
|
| 875 |
-
"rewards/accuracies": 0.
|
| 876 |
-
"rewards/chosen": -
|
| 877 |
-
"rewards/margins": 1.
|
| 878 |
-
"rewards/rejected": -
|
| 879 |
"step": 285
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 0.6071708976707668,
|
| 883 |
-
"grad_norm":
|
| 884 |
"learning_rate": 1.9999357655598891e-07,
|
| 885 |
-
"logits/chosen": -
|
| 886 |
-
"logits/rejected": -
|
| 887 |
-
"logps/chosen": -
|
| 888 |
-
"logps/rejected": -2.
|
| 889 |
-
"loss":
|
| 890 |
-
"rewards/accuracies": 0.
|
| 891 |
-
"rewards/chosen": -
|
| 892 |
-
"rewards/margins":
|
| 893 |
-
"rewards/rejected": -
|
| 894 |
"step": 290
|
| 895 |
},
|
| 896 |
{
|
| 897 |
"epoch": 0.6176393614237111,
|
| 898 |
-
"grad_norm":
|
| 899 |
"learning_rate": 1.9106026612264315e-07,
|
| 900 |
-
"logits/chosen": -2.
|
| 901 |
-
"logits/rejected": -
|
| 902 |
-
"logps/chosen": -
|
| 903 |
-
"logps/rejected": -2.
|
| 904 |
-
"loss":
|
| 905 |
-
"rewards/accuracies": 0.
|
| 906 |
-
"rewards/chosen": -
|
| 907 |
-
"rewards/margins":
|
| 908 |
-
"rewards/rejected": -
|
| 909 |
"step": 295
|
| 910 |
},
|
| 911 |
{
|
| 912 |
"epoch": 0.6281078251766553,
|
| 913 |
-
"grad_norm":
|
| 914 |
"learning_rate": 1.8220596619089573e-07,
|
| 915 |
-
"logits/chosen": -
|
| 916 |
-
"logits/rejected": -
|
| 917 |
-
"logps/chosen": -2.
|
| 918 |
-
"logps/rejected": -
|
| 919 |
-
"loss": 1.
|
| 920 |
-
"rewards/accuracies": 0.
|
| 921 |
-
"rewards/chosen": -
|
| 922 |
-
"rewards/margins": 1.
|
| 923 |
-
"rewards/rejected": -
|
| 924 |
"step": 300
|
| 925 |
},
|
| 926 |
{
|
| 927 |
"epoch": 0.6385762889295996,
|
| 928 |
-
"grad_norm":
|
| 929 |
"learning_rate": 1.7344254621846017e-07,
|
| 930 |
-
"logits/chosen": -
|
| 931 |
-
"logits/rejected": -
|
| 932 |
-
"logps/chosen": -
|
| 933 |
-
"logps/rejected": -
|
| 934 |
-
"loss":
|
| 935 |
-
"rewards/accuracies": 0.
|
| 936 |
-
"rewards/chosen": -
|
| 937 |
-
"rewards/margins":
|
| 938 |
-
"rewards/rejected": -
|
| 939 |
"step": 305
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 0.6490447526825438,
|
| 943 |
-
"grad_norm":
|
| 944 |
"learning_rate": 1.647817538357072e-07,
|
| 945 |
-
"logits/chosen": -
|
| 946 |
-
"logits/rejected": -
|
| 947 |
-
"logps/chosen": -2.
|
| 948 |
-
"logps/rejected": -
|
| 949 |
-
"loss":
|
| 950 |
-
"rewards/accuracies": 0.
|
| 951 |
-
"rewards/chosen": -
|
| 952 |
-
"rewards/margins": 1.
|
| 953 |
-
"rewards/rejected": -
|
| 954 |
"step": 310
|
| 955 |
},
|
| 956 |
{
|
| 957 |
"epoch": 0.6595132164354881,
|
| 958 |
-
"grad_norm":
|
| 959 |
"learning_rate": 1.562351990976095e-07,
|
| 960 |
-
"logits/chosen": -
|
| 961 |
-
"logits/rejected": -
|
| 962 |
-
"logps/chosen": -2.
|
| 963 |
-
"logps/rejected": -
|
| 964 |
-
"loss":
|
| 965 |
"rewards/accuracies": 0.6937500238418579,
|
| 966 |
-
"rewards/chosen": -
|
| 967 |
-
"rewards/margins": 1.
|
| 968 |
-
"rewards/rejected": -
|
| 969 |
"step": 315
|
| 970 |
},
|
| 971 |
{
|
| 972 |
"epoch": 0.6699816801884323,
|
| 973 |
-
"grad_norm":
|
| 974 |
"learning_rate": 1.478143389201113e-07,
|
| 975 |
-
"logits/chosen": -
|
| 976 |
-
"logits/rejected": -
|
| 977 |
-
"logps/chosen": -2.
|
| 978 |
-
"logps/rejected": -
|
| 979 |
-
"loss":
|
| 980 |
-
"rewards/accuracies": 0.
|
| 981 |
-
"rewards/chosen": -4.
|
| 982 |
-
"rewards/margins": 1.
|
| 983 |
-
"rewards/rejected": -
|
| 984 |
"step": 320
|
| 985 |
},
|
| 986 |
{
|
| 987 |
"epoch": 0.6804501439413766,
|
| 988 |
-
"grad_norm":
|
| 989 |
"learning_rate": 1.3953046172178413e-07,
|
| 990 |
-
"logits/chosen": -
|
| 991 |
-
"logits/rejected": -
|
| 992 |
-
"logps/chosen": -2.
|
| 993 |
-
"logps/rejected": -
|
| 994 |
-
"loss": 1.
|
| 995 |
-
"rewards/accuracies": 0.
|
| 996 |
-
"rewards/chosen": -
|
| 997 |
-
"rewards/margins": 1.
|
| 998 |
-
"rewards/rejected": -
|
| 999 |
"step": 325
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
"epoch": 0.6909186076943209,
|
| 1003 |
-
"grad_norm":
|
| 1004 |
"learning_rate": 1.3139467229135998e-07,
|
| 1005 |
-
"logits/chosen": -
|
| 1006 |
-
"logits/rejected": -
|
| 1007 |
-
"logps/chosen": -2.
|
| 1008 |
-
"logps/rejected": -3.
|
| 1009 |
-
"loss":
|
| 1010 |
"rewards/accuracies": 0.7437499761581421,
|
| 1011 |
-
"rewards/chosen": -
|
| 1012 |
-
"rewards/margins": 1.
|
| 1013 |
-
"rewards/rejected": -
|
| 1014 |
"step": 330
|
| 1015 |
},
|
| 1016 |
{
|
| 1017 |
"epoch": 0.7013870714472651,
|
| 1018 |
-
"grad_norm":
|
| 1019 |
"learning_rate": 1.2341787690142435e-07,
|
| 1020 |
-
"logits/chosen": -
|
| 1021 |
-
"logits/rejected": -
|
| 1022 |
-
"logps/chosen": -
|
| 1023 |
-
"logps/rejected": -3.
|
| 1024 |
-
"loss":
|
| 1025 |
-
"rewards/accuracies": 0.
|
| 1026 |
-
"rewards/chosen": -
|
| 1027 |
-
"rewards/margins": 1.
|
| 1028 |
-
"rewards/rejected": -
|
| 1029 |
"step": 335
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
"epoch": 0.7118555352002094,
|
| 1033 |
-
"grad_norm":
|
| 1034 |
"learning_rate": 1.1561076868822755e-07,
|
| 1035 |
-
"logits/chosen": -
|
| 1036 |
-
"logits/rejected": -
|
| 1037 |
-
"logps/chosen": -2.
|
| 1038 |
-
"logps/rejected": -
|
| 1039 |
-
"loss":
|
| 1040 |
-
"rewards/accuracies": 0.
|
| 1041 |
-
"rewards/chosen": -
|
| 1042 |
-
"rewards/margins":
|
| 1043 |
-
"rewards/rejected": -
|
| 1044 |
"step": 340
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 0.7223239989531536,
|
| 1048 |
-
"grad_norm":
|
| 1049 |
"learning_rate": 1.0798381331721107e-07,
|
| 1050 |
-
"logits/chosen": -
|
| 1051 |
-
"logits/rejected": -
|
| 1052 |
-
"logps/chosen": -2.
|
| 1053 |
-
"logps/rejected": -
|
| 1054 |
-
"loss":
|
| 1055 |
-
"rewards/accuracies": 0.
|
| 1056 |
-
"rewards/chosen": -
|
| 1057 |
-
"rewards/margins": 1.
|
| 1058 |
-
"rewards/rejected": -
|
| 1059 |
"step": 345
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
"epoch": 0.7327924627060979,
|
| 1063 |
-
"grad_norm":
|
| 1064 |
"learning_rate": 1.0054723495346482e-07,
|
| 1065 |
-
"logits/chosen": -
|
| 1066 |
-
"logits/rejected": -
|
| 1067 |
-
"logps/chosen": -2.
|
| 1068 |
-
"logps/rejected": -
|
| 1069 |
-
"loss":
|
| 1070 |
-
"rewards/accuracies": 0.
|
| 1071 |
-
"rewards/chosen": -
|
| 1072 |
-
"rewards/margins": 1.
|
| 1073 |
-
"rewards/rejected": -
|
| 1074 |
"step": 350
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
"epoch": 0.7432609264590422,
|
| 1078 |
-
"grad_norm":
|
| 1079 |
"learning_rate": 9.331100255592436e-08,
|
| 1080 |
-
"logits/chosen": -
|
| 1081 |
-
"logits/rejected": -
|
| 1082 |
-
"logps/chosen": -
|
| 1083 |
-
"logps/rejected": -
|
| 1084 |
-
"loss": 1.
|
| 1085 |
-
"rewards/accuracies": 0.
|
| 1086 |
-
"rewards/chosen": -
|
| 1087 |
-
"rewards/margins": 1.
|
| 1088 |
-
"rewards/rejected": -
|
| 1089 |
"step": 355
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
"epoch": 0.7537293902119864,
|
| 1093 |
-
"grad_norm":
|
| 1094 |
"learning_rate": 8.628481651367875e-08,
|
| 1095 |
-
"logits/chosen": -
|
| 1096 |
-
"logits/rejected": -
|
| 1097 |
-
"logps/chosen": -
|
| 1098 |
-
"logps/rejected": -
|
| 1099 |
-
"loss":
|
| 1100 |
-
"rewards/accuracies": 0.
|
| 1101 |
-
"rewards/chosen": -
|
| 1102 |
-
"rewards/margins": 1.
|
| 1103 |
-
"rewards/rejected": -
|
| 1104 |
"step": 360
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
"epoch": 0.7641978539649307,
|
| 1108 |
-
"grad_norm":
|
| 1109 |
"learning_rate": 7.947809564230445e-08,
|
| 1110 |
-
"logits/chosen": -
|
| 1111 |
-
"logits/rejected": -
|
| 1112 |
-
"logps/chosen": -2.
|
| 1113 |
-
"logps/rejected": -
|
| 1114 |
-
"loss":
|
| 1115 |
-
"rewards/accuracies": 0.
|
| 1116 |
-
"rewards/chosen": -4.
|
| 1117 |
-
"rewards/margins": 1.
|
| 1118 |
-
"rewards/rejected": -
|
| 1119 |
"step": 365
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
"epoch": 0.7746663177178749,
|
| 1123 |
-
"grad_norm":
|
| 1124 |
"learning_rate": 7.289996455765748e-08,
|
| 1125 |
-
"logits/chosen": -
|
| 1126 |
-
"logits/rejected": -
|
| 1127 |
-
"logps/chosen": -
|
| 1128 |
-
"logps/rejected": -
|
| 1129 |
-
"loss":
|
| 1130 |
-
"rewards/accuracies": 0.
|
| 1131 |
-
"rewards/chosen": -
|
| 1132 |
-
"rewards/margins": 1.
|
| 1133 |
-
"rewards/rejected": -
|
| 1134 |
"step": 370
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
"epoch": 0.7851347814708192,
|
| 1138 |
-
"grad_norm":
|
| 1139 |
"learning_rate": 6.655924144404906e-08,
|
| 1140 |
-
"logits/chosen": -
|
| 1141 |
-
"logits/rejected": -
|
| 1142 |
-
"logps/chosen": -2.
|
| 1143 |
-
"logps/rejected": -
|
| 1144 |
-
"loss":
|
| 1145 |
-
"rewards/accuracies": 0.
|
| 1146 |
-
"rewards/chosen": -
|
| 1147 |
-
"rewards/margins": 1.
|
| 1148 |
-
"rewards/rejected": -
|
| 1149 |
"step": 375
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 0.7956032452237635,
|
| 1153 |
-
"grad_norm":
|
| 1154 |
"learning_rate": 6.046442623320145e-08,
|
| 1155 |
-
"logits/chosen": -
|
| 1156 |
-
"logits/rejected": -
|
| 1157 |
-
"logps/chosen": -2.
|
| 1158 |
-
"logps/rejected": -
|
| 1159 |
-
"loss":
|
| 1160 |
-
"rewards/accuracies": 0.
|
| 1161 |
-
"rewards/chosen": -
|
| 1162 |
-
"rewards/margins": 1.
|
| 1163 |
-
"rewards/rejected": -
|
| 1164 |
"step": 380
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
"epoch": 0.8060717089767077,
|
| 1168 |
-
"grad_norm":
|
| 1169 |
"learning_rate": 5.4623689209832484e-08,
|
| 1170 |
-
"logits/chosen": -
|
| 1171 |
-
"logits/rejected": -
|
| 1172 |
-
"logps/chosen": -2.
|
| 1173 |
-
"logps/rejected": -
|
| 1174 |
-
"loss":
|
| 1175 |
"rewards/accuracies": 0.7562500238418579,
|
| 1176 |
-
"rewards/chosen": -
|
| 1177 |
-
"rewards/margins": 1.
|
| 1178 |
-
"rewards/rejected": -
|
| 1179 |
"step": 385
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 0.816540172729652,
|
| 1183 |
-
"grad_norm":
|
| 1184 |
"learning_rate": 4.904486005914027e-08,
|
| 1185 |
-
"logits/chosen": -
|
| 1186 |
-
"logits/rejected": -
|
| 1187 |
-
"logps/chosen": -2.
|
| 1188 |
-
"logps/rejected": -
|
| 1189 |
-
"loss":
|
| 1190 |
"rewards/accuracies": 0.731249988079071,
|
| 1191 |
-
"rewards/chosen": -
|
| 1192 |
-
"rewards/margins": 1.
|
| 1193 |
-
"rewards/rejected": -
|
| 1194 |
"step": 390
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
"epoch": 0.8270086364825961,
|
| 1198 |
-
"grad_norm":
|
| 1199 |
"learning_rate": 4.373541737087263e-08,
|
| 1200 |
-
"logits/chosen": -
|
| 1201 |
-
"logits/rejected": -
|
| 1202 |
-
"logps/chosen": -2.
|
| 1203 |
-
"logps/rejected": -
|
| 1204 |
-
"loss":
|
| 1205 |
-
"rewards/accuracies": 0.
|
| 1206 |
-
"rewards/chosen": -
|
| 1207 |
-
"rewards/margins": 1.
|
| 1208 |
-
"rewards/rejected": -
|
| 1209 |
"step": 395
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
"epoch": 0.8374771002355405,
|
| 1213 |
-
"grad_norm":
|
| 1214 |
"learning_rate": 3.8702478614051345e-08,
|
| 1215 |
-
"logits/chosen": -
|
| 1216 |
-
"logits/rejected": -
|
| 1217 |
-
"logps/chosen": -2.
|
| 1218 |
-
"logps/rejected": -
|
| 1219 |
-
"loss":
|
| 1220 |
-
"rewards/accuracies": 0.
|
| 1221 |
-
"rewards/chosen": -
|
| 1222 |
-
"rewards/margins": 1.
|
| 1223 |
-
"rewards/rejected": -
|
| 1224 |
"step": 400
|
| 1225 |
},
|
| 1226 |
{
|
| 1227 |
"epoch": 0.8374771002355405,
|
| 1228 |
-
"eval_logits/chosen": -
|
| 1229 |
-
"eval_logits/rejected": -
|
| 1230 |
-
"eval_logps/chosen": -2.
|
| 1231 |
-
"eval_logps/rejected": -3.
|
| 1232 |
-
"eval_loss":
|
| 1233 |
"eval_rewards/accuracies": 0.7319999933242798,
|
| 1234 |
-
"eval_rewards/chosen": -
|
| 1235 |
-
"eval_rewards/margins": 1.
|
| 1236 |
-
"eval_rewards/rejected": -
|
| 1237 |
-
"eval_runtime": 171.
|
| 1238 |
-
"eval_samples_per_second": 11.
|
| 1239 |
"eval_steps_per_second": 0.729,
|
| 1240 |
"step": 400
|
| 1241 |
},
|
| 1242 |
{
|
| 1243 |
"epoch": 0.8479455639884846,
|
| 1244 |
-
"grad_norm":
|
| 1245 |
"learning_rate": 3.3952790595787986e-08,
|
| 1246 |
-
"logits/chosen": -
|
| 1247 |
-
"logits/rejected": -
|
| 1248 |
-
"logps/chosen": -
|
| 1249 |
-
"logps/rejected": -
|
| 1250 |
-
"loss":
|
| 1251 |
-
"rewards/accuracies": 0.
|
| 1252 |
-
"rewards/chosen": -
|
| 1253 |
-
"rewards/margins": 1.
|
| 1254 |
-
"rewards/rejected": -
|
| 1255 |
"step": 405
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
"epoch": 0.8584140277414289,
|
| 1259 |
-
"grad_norm":
|
| 1260 |
"learning_rate": 2.9492720416985e-08,
|
| 1261 |
-
"logits/chosen": -
|
| 1262 |
-
"logits/rejected": -
|
| 1263 |
-
"logps/chosen": -2.
|
| 1264 |
-
"logps/rejected": -
|
| 1265 |
-
"loss":
|
| 1266 |
-
"rewards/accuracies": 0.
|
| 1267 |
-
"rewards/chosen": -
|
| 1268 |
-
"rewards/margins": 1.
|
| 1269 |
-
"rewards/rejected": -
|
| 1270 |
"step": 410
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 0.8688824914943732,
|
| 1274 |
-
"grad_norm":
|
| 1275 |
"learning_rate": 2.5328246937043525e-08,
|
| 1276 |
-
"logits/chosen": -
|
| 1277 |
-
"logits/rejected": -
|
| 1278 |
-
"logps/chosen": -
|
| 1279 |
-
"logps/rejected": -
|
| 1280 |
-
"loss":
|
| 1281 |
-
"rewards/accuracies": 0.
|
| 1282 |
-
"rewards/chosen": -
|
| 1283 |
-
"rewards/margins": 1.
|
| 1284 |
-
"rewards/rejected": -
|
| 1285 |
"step": 415
|
| 1286 |
},
|
| 1287 |
{
|
| 1288 |
"epoch": 0.8793509552473174,
|
| 1289 |
-
"grad_norm":
|
| 1290 |
"learning_rate": 2.1464952759020856e-08,
|
| 1291 |
-
"logits/chosen": -
|
| 1292 |
-
"logits/rejected": -
|
| 1293 |
-
"logps/chosen": -2.
|
| 1294 |
-
"logps/rejected": -3.
|
| 1295 |
-
"loss":
|
| 1296 |
"rewards/accuracies": 0.699999988079071,
|
| 1297 |
-
"rewards/chosen": -
|
| 1298 |
-
"rewards/margins": 1.
|
| 1299 |
-
"rewards/rejected": -
|
| 1300 |
"step": 420
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
"epoch": 0.8898194190002617,
|
| 1304 |
-
"grad_norm":
|
| 1305 |
"learning_rate": 1.7908016745981856e-08,
|
| 1306 |
-
"logits/chosen": -
|
| 1307 |
-
"logits/rejected": -
|
| 1308 |
-
"logps/chosen": -2.
|
| 1309 |
-
"logps/rejected": -
|
| 1310 |
-
"loss":
|
| 1311 |
-
"rewards/accuracies": 0.
|
| 1312 |
-
"rewards/chosen": -
|
| 1313 |
-
"rewards/margins":
|
| 1314 |
-
"rewards/rejected": -
|
| 1315 |
"step": 425
|
| 1316 |
},
|
| 1317 |
{
|
| 1318 |
"epoch": 0.9002878827532059,
|
| 1319 |
-
"grad_norm":
|
| 1320 |
"learning_rate": 1.4662207078575684e-08,
|
| 1321 |
-
"logits/chosen": -
|
| 1322 |
-
"logits/rejected": -
|
| 1323 |
-
"logps/chosen": -2.
|
| 1324 |
-
"logps/rejected": -
|
| 1325 |
-
"loss":
|
| 1326 |
-
"rewards/accuracies": 0.
|
| 1327 |
-
"rewards/chosen": -
|
| 1328 |
-
"rewards/margins": 1.
|
| 1329 |
-
"rewards/rejected": -
|
| 1330 |
"step": 430
|
| 1331 |
},
|
| 1332 |
{
|
| 1333 |
"epoch": 0.9107563465061502,
|
| 1334 |
-
"grad_norm":
|
| 1335 |
"learning_rate": 1.1731874863145142e-08,
|
| 1336 |
-
"logits/chosen": -
|
| 1337 |
-
"logits/rejected": -
|
| 1338 |
-
"logps/chosen": -2.
|
| 1339 |
-
"logps/rejected": -
|
| 1340 |
-
"loss":
|
| 1341 |
"rewards/accuracies": 0.668749988079071,
|
| 1342 |
-
"rewards/chosen": -
|
| 1343 |
-
"rewards/margins": 1.
|
| 1344 |
-
"rewards/rejected": -
|
| 1345 |
"step": 435
|
| 1346 |
},
|
| 1347 |
{
|
| 1348 |
"epoch": 0.9212248102590945,
|
| 1349 |
-
"grad_norm":
|
| 1350 |
"learning_rate": 9.12094829893642e-09,
|
| 1351 |
-
"logits/chosen": -
|
| 1352 |
-
"logits/rejected": -
|
| 1353 |
-
"logps/chosen": -2.
|
| 1354 |
-
"logps/rejected": -
|
| 1355 |
-
"loss":
|
| 1356 |
-
"rewards/accuracies": 0.
|
| 1357 |
-
"rewards/chosen": -
|
| 1358 |
-
"rewards/margins": 1.
|
| 1359 |
-
"rewards/rejected": -
|
| 1360 |
"step": 440
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
"epoch": 0.9316932740120387,
|
| 1364 |
-
"grad_norm":
|
| 1365 |
"learning_rate": 6.832927412229017e-09,
|
| 1366 |
-
"logits/chosen": -1.
|
| 1367 |
-
"logits/rejected": -
|
| 1368 |
-
"logps/chosen": -2.
|
| 1369 |
-
"logps/rejected": -3.
|
| 1370 |
-
"loss":
|
| 1371 |
"rewards/accuracies": 0.737500011920929,
|
| 1372 |
-
"rewards/chosen": -
|
| 1373 |
-
"rewards/margins": 1.
|
| 1374 |
-
"rewards/rejected": -
|
| 1375 |
"step": 445
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 0.942161737764983,
|
| 1379 |
-
"grad_norm":
|
| 1380 |
"learning_rate": 4.8708793644441086e-09,
|
| 1381 |
-
"logits/chosen": -
|
| 1382 |
-
"logits/rejected": -
|
| 1383 |
-
"logps/chosen": -2.
|
| 1384 |
-
"logps/rejected": -
|
| 1385 |
-
"loss":
|
| 1386 |
-
"rewards/accuracies": 0.
|
| 1387 |
-
"rewards/chosen": -
|
| 1388 |
-
"rewards/margins":
|
| 1389 |
-
"rewards/rejected": -
|
| 1390 |
"step": 450
|
| 1391 |
},
|
| 1392 |
{
|
| 1393 |
"epoch": 0.9526302015179272,
|
| 1394 |
-
"grad_norm":
|
| 1395 |
"learning_rate": 3.2374343405217884e-09,
|
| 1396 |
-
"logits/chosen": -
|
| 1397 |
-
"logits/rejected": -
|
| 1398 |
-
"logps/chosen": -2.
|
| 1399 |
-
"logps/rejected": -
|
| 1400 |
-
"loss":
|
| 1401 |
"rewards/accuracies": 0.706250011920929,
|
| 1402 |
-
"rewards/chosen": -
|
| 1403 |
-
"rewards/margins": 1.
|
| 1404 |
-
"rewards/rejected": -
|
| 1405 |
"step": 455
|
| 1406 |
},
|
| 1407 |
{
|
| 1408 |
"epoch": 0.9630986652708715,
|
| 1409 |
-
"grad_norm":
|
| 1410 |
"learning_rate": 1.9347820230782295e-09,
|
| 1411 |
-
"logits/chosen": -
|
| 1412 |
-
"logits/rejected": -
|
| 1413 |
-
"logps/chosen": -2.
|
| 1414 |
-
"logps/rejected": -3.
|
| 1415 |
-
"loss":
|
| 1416 |
-
"rewards/accuracies": 0.
|
| 1417 |
-
"rewards/chosen": -
|
| 1418 |
-
"rewards/margins":
|
| 1419 |
-
"rewards/rejected": -
|
| 1420 |
"step": 460
|
| 1421 |
},
|
| 1422 |
{
|
| 1423 |
"epoch": 0.9735671290238157,
|
| 1424 |
-
"grad_norm":
|
| 1425 |
"learning_rate": 9.64668657069706e-10,
|
| 1426 |
-
"logits/chosen": -
|
| 1427 |
-
"logits/rejected": -
|
| 1428 |
-
"logps/chosen": -2.
|
| 1429 |
-
"logps/rejected": -3.
|
| 1430 |
-
"loss":
|
| 1431 |
-
"rewards/accuracies": 0.
|
| 1432 |
-
"rewards/chosen": -
|
| 1433 |
-
"rewards/margins":
|
| 1434 |
-
"rewards/rejected": -
|
| 1435 |
"step": 465
|
| 1436 |
},
|
| 1437 |
{
|
| 1438 |
"epoch": 0.98403559277676,
|
| 1439 |
-
"grad_norm":
|
| 1440 |
"learning_rate": 3.2839470889836627e-10,
|
| 1441 |
-
"logits/chosen": -1.
|
| 1442 |
-
"logits/rejected": -1.
|
| 1443 |
-
"logps/chosen": -2.
|
| 1444 |
-
"logps/rejected": -
|
| 1445 |
-
"loss":
|
| 1446 |
-
"rewards/accuracies": 0.
|
| 1447 |
-
"rewards/chosen": -
|
| 1448 |
-
"rewards/margins":
|
| 1449 |
-
"rewards/rejected": -
|
| 1450 |
"step": 470
|
| 1451 |
},
|
| 1452 |
{
|
| 1453 |
"epoch": 0.9945040565297043,
|
| 1454 |
-
"grad_norm":
|
| 1455 |
"learning_rate": 2.6813123097352287e-11,
|
| 1456 |
-
"logits/chosen": -
|
| 1457 |
-
"logits/rejected": -
|
| 1458 |
-
"logps/chosen": -2.
|
| 1459 |
-
"logps/rejected": -
|
| 1460 |
-
"loss":
|
| 1461 |
-
"rewards/accuracies": 0.
|
| 1462 |
-
"rewards/chosen": -
|
| 1463 |
-
"rewards/margins": 1.
|
| 1464 |
-
"rewards/rejected": -
|
| 1465 |
"step": 475
|
| 1466 |
},
|
| 1467 |
{
|
| 1468 |
"epoch": 0.998691442030882,
|
| 1469 |
"step": 477,
|
| 1470 |
"total_flos": 0.0,
|
| 1471 |
-
"train_loss": 1.
|
| 1472 |
-
"train_runtime":
|
| 1473 |
-
"train_samples_per_second": 2.
|
| 1474 |
-
"train_steps_per_second": 0.
|
| 1475 |
}
|
| 1476 |
],
|
| 1477 |
"logging_steps": 5,
|
|
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.002093692750588851,
|
| 13 |
+
"grad_norm": 23.597004065952554,
|
| 14 |
"learning_rate": 1.0416666666666666e-08,
|
| 15 |
"logits/chosen": -2.8280014991760254,
|
| 16 |
"logits/rejected": -2.8466408252716064,
|
| 17 |
"logps/chosen": -1.1081000566482544,
|
| 18 |
"logps/rejected": -1.146370530128479,
|
| 19 |
+
"loss": 1.825,
|
| 20 |
"rewards/accuracies": 0.46875,
|
| 21 |
"rewards/chosen": -2.216200113296509,
|
| 22 |
"rewards/margins": 0.0765407383441925,
|
|
|
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.010468463752944255,
|
| 28 |
+
"grad_norm": 20.151143431686044,
|
| 29 |
"learning_rate": 5.208333333333333e-08,
|
| 30 |
+
"logits/chosen": -2.88156795501709,
|
| 31 |
+
"logits/rejected": -2.84816837310791,
|
| 32 |
+
"logps/chosen": -1.0210057497024536,
|
| 33 |
+
"logps/rejected": -1.0806357860565186,
|
| 34 |
+
"loss": 1.7101,
|
| 35 |
"rewards/accuracies": 0.5390625,
|
| 36 |
+
"rewards/chosen": -2.0420114994049072,
|
| 37 |
+
"rewards/margins": 0.11925993114709854,
|
| 38 |
+
"rewards/rejected": -2.161271572113037,
|
| 39 |
"step": 5
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.02093692750588851,
|
| 43 |
+
"grad_norm": 19.420093362135834,
|
| 44 |
"learning_rate": 1.0416666666666667e-07,
|
| 45 |
+
"logits/chosen": -2.992948055267334,
|
| 46 |
+
"logits/rejected": -2.9420628547668457,
|
| 47 |
+
"logps/chosen": -1.0234142541885376,
|
| 48 |
+
"logps/rejected": -1.165908694267273,
|
| 49 |
+
"loss": 1.7409,
|
| 50 |
"rewards/accuracies": 0.581250011920929,
|
| 51 |
+
"rewards/chosen": -2.046828508377075,
|
| 52 |
+
"rewards/margins": 0.28498905897140503,
|
| 53 |
+
"rewards/rejected": -2.331817388534546,
|
| 54 |
"step": 10
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 0.031405391258832765,
|
| 58 |
+
"grad_norm": 22.360933443209454,
|
| 59 |
"learning_rate": 1.5624999999999999e-07,
|
| 60 |
+
"logits/chosen": -2.9190802574157715,
|
| 61 |
+
"logits/rejected": -2.9058239459991455,
|
| 62 |
+
"logps/chosen": -1.0527833700180054,
|
| 63 |
+
"logps/rejected": -1.1032918691635132,
|
| 64 |
+
"loss": 1.7228,
|
| 65 |
"rewards/accuracies": 0.5375000238418579,
|
| 66 |
+
"rewards/chosen": -2.1055667400360107,
|
| 67 |
+
"rewards/margins": 0.10101678222417831,
|
| 68 |
+
"rewards/rejected": -2.2065837383270264,
|
| 69 |
"step": 15
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.04187385501177702,
|
| 73 |
+
"grad_norm": 18.11495891073632,
|
| 74 |
"learning_rate": 2.0833333333333333e-07,
|
| 75 |
+
"logits/chosen": -2.9407191276550293,
|
| 76 |
+
"logits/rejected": -2.8896334171295166,
|
| 77 |
+
"logps/chosen": -1.0061452388763428,
|
| 78 |
+
"logps/rejected": -1.1030395030975342,
|
| 79 |
+
"loss": 1.6957,
|
| 80 |
+
"rewards/accuracies": 0.550000011920929,
|
| 81 |
+
"rewards/chosen": -2.0122904777526855,
|
| 82 |
+
"rewards/margins": 0.19378867745399475,
|
| 83 |
+
"rewards/rejected": -2.2060790061950684,
|
| 84 |
"step": 20
|
| 85 |
},
|
| 86 |
{
|
| 87 |
"epoch": 0.05234231876472128,
|
| 88 |
+
"grad_norm": 17.73214868239448,
|
| 89 |
"learning_rate": 2.604166666666667e-07,
|
| 90 |
+
"logits/chosen": -2.945413112640381,
|
| 91 |
+
"logits/rejected": -2.8812930583953857,
|
| 92 |
+
"logps/chosen": -0.9288013577461243,
|
| 93 |
+
"logps/rejected": -1.1164944171905518,
|
| 94 |
+
"loss": 1.6642,
|
| 95 |
+
"rewards/accuracies": 0.574999988079071,
|
| 96 |
+
"rewards/chosen": -1.8576027154922485,
|
| 97 |
+
"rewards/margins": 0.3753865361213684,
|
| 98 |
+
"rewards/rejected": -2.2329888343811035,
|
| 99 |
"step": 25
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.06281078251766553,
|
| 103 |
+
"grad_norm": 17.49667039937059,
|
| 104 |
"learning_rate": 3.1249999999999997e-07,
|
| 105 |
+
"logits/chosen": -2.920177459716797,
|
| 106 |
+
"logits/rejected": -2.9008939266204834,
|
| 107 |
+
"logps/chosen": -0.9087249636650085,
|
| 108 |
+
"logps/rejected": -0.9417479634284973,
|
| 109 |
+
"loss": 1.6764,
|
| 110 |
"rewards/accuracies": 0.550000011920929,
|
| 111 |
+
"rewards/chosen": -1.817449927330017,
|
| 112 |
+
"rewards/margins": 0.06604615598917007,
|
| 113 |
+
"rewards/rejected": -1.8834959268569946,
|
| 114 |
"step": 30
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.07327924627060979,
|
| 118 |
+
"grad_norm": 18.317659111307954,
|
| 119 |
"learning_rate": 3.645833333333333e-07,
|
| 120 |
+
"logits/chosen": -2.929032564163208,
|
| 121 |
+
"logits/rejected": -2.8953332901000977,
|
| 122 |
+
"logps/chosen": -0.9193178415298462,
|
| 123 |
+
"logps/rejected": -1.1149028539657593,
|
| 124 |
+
"loss": 1.65,
|
| 125 |
+
"rewards/accuracies": 0.606249988079071,
|
| 126 |
+
"rewards/chosen": -1.8386356830596924,
|
| 127 |
+
"rewards/margins": 0.39116981625556946,
|
| 128 |
+
"rewards/rejected": -2.2298057079315186,
|
| 129 |
"step": 35
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.08374771002355404,
|
| 133 |
+
"grad_norm": 17.730458935387485,
|
| 134 |
"learning_rate": 4.1666666666666667e-07,
|
| 135 |
+
"logits/chosen": -2.9188663959503174,
|
| 136 |
+
"logits/rejected": -2.877103328704834,
|
| 137 |
+
"logps/chosen": -0.9255539178848267,
|
| 138 |
+
"logps/rejected": -1.0478177070617676,
|
| 139 |
+
"loss": 1.6224,
|
| 140 |
+
"rewards/accuracies": 0.581250011920929,
|
| 141 |
+
"rewards/chosen": -1.8511078357696533,
|
| 142 |
+
"rewards/margins": 0.24452760815620422,
|
| 143 |
+
"rewards/rejected": -2.095635414123535,
|
| 144 |
"step": 40
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.0942161737764983,
|
| 148 |
+
"grad_norm": 15.937073180460068,
|
| 149 |
"learning_rate": 4.6874999999999996e-07,
|
| 150 |
+
"logits/chosen": -2.875232696533203,
|
| 151 |
+
"logits/rejected": -2.8806424140930176,
|
| 152 |
+
"logps/chosen": -0.9038581848144531,
|
| 153 |
+
"logps/rejected": -1.0820366144180298,
|
| 154 |
+
"loss": 1.5724,
|
| 155 |
+
"rewards/accuracies": 0.625,
|
| 156 |
+
"rewards/chosen": -1.8077163696289062,
|
| 157 |
+
"rewards/margins": 0.35635682940483093,
|
| 158 |
+
"rewards/rejected": -2.1640732288360596,
|
| 159 |
"step": 45
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.10468463752944256,
|
| 163 |
+
"grad_norm": 23.986363177533022,
|
| 164 |
"learning_rate": 4.999731868769026e-07,
|
| 165 |
+
"logits/chosen": -2.820544719696045,
|
| 166 |
+
"logits/rejected": -2.832681179046631,
|
| 167 |
+
"logps/chosen": -0.9590412378311157,
|
| 168 |
+
"logps/rejected": -1.2149635553359985,
|
| 169 |
+
"loss": 1.6276,
|
| 170 |
+
"rewards/accuracies": 0.637499988079071,
|
| 171 |
+
"rewards/chosen": -1.9180824756622314,
|
| 172 |
+
"rewards/margins": 0.5118446946144104,
|
| 173 |
+
"rewards/rejected": -2.429927110671997,
|
| 174 |
"step": 50
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.11515310128238682,
|
| 178 |
+
"grad_norm": 24.997319339492417,
|
| 179 |
"learning_rate": 4.996716052911017e-07,
|
| 180 |
+
"logits/chosen": -2.910067081451416,
|
| 181 |
+
"logits/rejected": -2.9242091178894043,
|
| 182 |
+
"logps/chosen": -1.0754766464233398,
|
| 183 |
+
"logps/rejected": -1.1867624521255493,
|
| 184 |
+
"loss": 1.6255,
|
| 185 |
+
"rewards/accuracies": 0.574999988079071,
|
| 186 |
+
"rewards/chosen": -2.1509532928466797,
|
| 187 |
+
"rewards/margins": 0.22257177531719208,
|
| 188 |
+
"rewards/rejected": -2.3735249042510986,
|
| 189 |
"step": 55
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.12562156503533106,
|
| 193 |
+
"grad_norm": 29.673956562215213,
|
| 194 |
"learning_rate": 4.990353313429303e-07,
|
| 195 |
+
"logits/chosen": -2.846463441848755,
|
| 196 |
+
"logits/rejected": -2.8395891189575195,
|
| 197 |
+
"logps/chosen": -1.0373256206512451,
|
| 198 |
+
"logps/rejected": -1.2098807096481323,
|
| 199 |
+
"loss": 1.5824,
|
| 200 |
+
"rewards/accuracies": 0.5625,
|
| 201 |
+
"rewards/chosen": -2.0746512413024902,
|
| 202 |
+
"rewards/margins": 0.34511035680770874,
|
| 203 |
+
"rewards/rejected": -2.4197614192962646,
|
| 204 |
"step": 60
|
| 205 |
},
|
| 206 |
{
|
| 207 |
"epoch": 0.1360900287882753,
|
| 208 |
+
"grad_norm": 25.70155018194367,
|
| 209 |
"learning_rate": 4.980652179769217e-07,
|
| 210 |
+
"logits/chosen": -2.8658077716827393,
|
| 211 |
+
"logits/rejected": -2.889425039291382,
|
| 212 |
+
"logps/chosen": -1.0232688188552856,
|
| 213 |
+
"logps/rejected": -1.1985584497451782,
|
| 214 |
+
"loss": 1.5955,
|
| 215 |
"rewards/accuracies": 0.59375,
|
| 216 |
+
"rewards/chosen": -2.0465376377105713,
|
| 217 |
+
"rewards/margins": 0.3505793511867523,
|
| 218 |
+
"rewards/rejected": -2.3971168994903564,
|
| 219 |
"step": 65
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.14655849254121958,
|
| 223 |
+
"grad_norm": 26.911214626746915,
|
| 224 |
"learning_rate": 4.967625656594781e-07,
|
| 225 |
+
"logits/chosen": -2.8660895824432373,
|
| 226 |
+
"logits/rejected": -2.8656294345855713,
|
| 227 |
+
"logps/chosen": -1.1060866117477417,
|
| 228 |
+
"logps/rejected": -1.3024221658706665,
|
| 229 |
+
"loss": 1.5443,
|
| 230 |
+
"rewards/accuracies": 0.625,
|
| 231 |
+
"rewards/chosen": -2.2121732234954834,
|
| 232 |
+
"rewards/margins": 0.39267101883888245,
|
| 233 |
+
"rewards/rejected": -2.604844331741333,
|
| 234 |
"step": 70
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.15702695629416383,
|
| 238 |
+
"grad_norm": 31.97064281949989,
|
| 239 |
"learning_rate": 4.951291206355559e-07,
|
| 240 |
+
"logits/chosen": -2.813340425491333,
|
| 241 |
+
"logits/rejected": -2.76302170753479,
|
| 242 |
+
"logps/chosen": -1.1282150745391846,
|
| 243 |
+
"logps/rejected": -1.4221516847610474,
|
| 244 |
+
"loss": 1.5693,
|
| 245 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 246 |
+
"rewards/chosen": -2.256430149078369,
|
| 247 |
+
"rewards/margins": 0.5878733992576599,
|
| 248 |
+
"rewards/rejected": -2.8443033695220947,
|
| 249 |
"step": 75
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.16749542004710807,
|
| 253 |
+
"grad_norm": 22.991199981656727,
|
| 254 |
"learning_rate": 4.93167072587771e-07,
|
| 255 |
+
"logits/chosen": -2.7443230152130127,
|
| 256 |
+
"logits/rejected": -2.7397193908691406,
|
| 257 |
+
"logps/chosen": -1.1170539855957031,
|
| 258 |
+
"logps/rejected": -1.485013723373413,
|
| 259 |
+
"loss": 1.5498,
|
| 260 |
"rewards/accuracies": 0.6875,
|
| 261 |
+
"rewards/chosen": -2.2341079711914062,
|
| 262 |
+
"rewards/margins": 0.7359195947647095,
|
| 263 |
+
"rewards/rejected": -2.970027446746826,
|
| 264 |
"step": 80
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 0.17796388380005235,
|
| 268 |
+
"grad_norm": 27.36414724828374,
|
| 269 |
"learning_rate": 4.908790517010636e-07,
|
| 270 |
+
"logits/chosen": -2.798058271408081,
|
| 271 |
+
"logits/rejected": -2.7502901554107666,
|
| 272 |
+
"logps/chosen": -1.1033599376678467,
|
| 273 |
+
"logps/rejected": -1.4456737041473389,
|
| 274 |
+
"loss": 1.5329,
|
| 275 |
+
"rewards/accuracies": 0.637499988079071,
|
| 276 |
+
"rewards/chosen": -2.2067198753356934,
|
| 277 |
+
"rewards/margins": 0.6846276521682739,
|
| 278 |
+
"rewards/rejected": -2.8913474082946777,
|
| 279 |
"step": 85
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.1884323475529966,
|
| 283 |
+
"grad_norm": 21.16101861051188,
|
| 284 |
"learning_rate": 4.882681251368548e-07,
|
| 285 |
+
"logits/chosen": -2.7800021171569824,
|
| 286 |
+
"logits/rejected": -2.7292940616607666,
|
| 287 |
+
"logps/chosen": -1.1692888736724854,
|
| 288 |
+
"logps/rejected": -1.5973972082138062,
|
| 289 |
+
"loss": 1.5052,
|
| 290 |
+
"rewards/accuracies": 0.625,
|
| 291 |
+
"rewards/chosen": -2.3385777473449707,
|
| 292 |
+
"rewards/margins": 0.856216549873352,
|
| 293 |
+
"rewards/rejected": -3.1947944164276123,
|
| 294 |
"step": 90
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 0.19890081130594087,
|
| 298 |
+
"grad_norm": 26.528533058821825,
|
| 299 |
"learning_rate": 4.853377929214243e-07,
|
| 300 |
+
"logits/chosen": -2.704568386077881,
|
| 301 |
+
"logits/rejected": -2.6892647743225098,
|
| 302 |
+
"logps/chosen": -1.1889941692352295,
|
| 303 |
+
"logps/rejected": -1.5050609111785889,
|
| 304 |
+
"loss": 1.5355,
|
| 305 |
+
"rewards/accuracies": 0.581250011920929,
|
| 306 |
+
"rewards/chosen": -2.377988338470459,
|
| 307 |
+
"rewards/margins": 0.6321329474449158,
|
| 308 |
+
"rewards/rejected": -3.0101218223571777,
|
| 309 |
"step": 95
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.2093692750588851,
|
| 313 |
+
"grad_norm": 29.65374977644177,
|
| 314 |
"learning_rate": 4.820919832540181e-07,
|
| 315 |
+
"logits/chosen": -2.7762622833251953,
|
| 316 |
+
"logits/rejected": -2.750223398208618,
|
| 317 |
+
"logps/chosen": -1.1571177244186401,
|
| 318 |
+
"logps/rejected": -1.3776543140411377,
|
| 319 |
+
"loss": 1.5847,
|
| 320 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 321 |
+
"rewards/chosen": -2.3142354488372803,
|
| 322 |
+
"rewards/margins": 0.4410727620124817,
|
| 323 |
+
"rewards/rejected": -2.7553086280822754,
|
| 324 |
"step": 100
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 0.21983773881182936,
|
| 328 |
+
"grad_norm": 22.838032447560202,
|
| 329 |
"learning_rate": 4.785350472409791e-07,
|
| 330 |
+
"logits/chosen": -2.712097644805908,
|
| 331 |
+
"logits/rejected": -2.684450626373291,
|
| 332 |
+
"logps/chosen": -1.1789191961288452,
|
| 333 |
+
"logps/rejected": -1.5337121486663818,
|
| 334 |
+
"loss": 1.4327,
|
| 335 |
+
"rewards/accuracies": 0.65625,
|
| 336 |
+
"rewards/chosen": -2.3578383922576904,
|
| 337 |
+
"rewards/margins": 0.7095857858657837,
|
| 338 |
+
"rewards/rejected": -3.0674242973327637,
|
| 339 |
"step": 105
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.23030620256477363,
|
| 343 |
+
"grad_norm": 42.989486471273096,
|
| 344 |
"learning_rate": 4.7467175306295647e-07,
|
| 345 |
+
"logits/chosen": -2.6487460136413574,
|
| 346 |
+
"logits/rejected": -2.626556873321533,
|
| 347 |
+
"logps/chosen": -1.1375197172164917,
|
| 348 |
+
"logps/rejected": -1.4804375171661377,
|
| 349 |
+
"loss": 1.4322,
|
| 350 |
+
"rewards/accuracies": 0.625,
|
| 351 |
+
"rewards/chosen": -2.2750394344329834,
|
| 352 |
+
"rewards/margins": 0.6858354210853577,
|
| 353 |
+
"rewards/rejected": -2.9608750343322754,
|
| 354 |
"step": 110
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.24077466631771788,
|
| 358 |
+
"grad_norm": 29.408403128680163,
|
| 359 |
"learning_rate": 4.70507279583015e-07,
|
| 360 |
+
"logits/chosen": -2.6792874336242676,
|
| 361 |
+
"logits/rejected": -2.669962167739868,
|
| 362 |
+
"logps/chosen": -1.179626226425171,
|
| 363 |
+
"logps/rejected": -1.3990623950958252,
|
| 364 |
+
"loss": 1.5301,
|
| 365 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 366 |
+
"rewards/chosen": -2.359252452850342,
|
| 367 |
+
"rewards/margins": 0.43887215852737427,
|
| 368 |
+
"rewards/rejected": -2.7981247901916504,
|
| 369 |
"step": 115
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.2512431300706621,
|
| 373 |
+
"grad_norm": 25.256732287485413,
|
| 374 |
"learning_rate": 4.6604720940421207e-07,
|
| 375 |
+
"logits/chosen": -2.6912872791290283,
|
| 376 |
+
"logits/rejected": -2.6357336044311523,
|
| 377 |
+
"logps/chosen": -1.1251033544540405,
|
| 378 |
+
"logps/rejected": -1.5543628931045532,
|
| 379 |
+
"loss": 1.5099,
|
| 380 |
+
"rewards/accuracies": 0.625,
|
| 381 |
+
"rewards/chosen": -2.250206708908081,
|
| 382 |
+
"rewards/margins": 0.8585190773010254,
|
| 383 |
+
"rewards/rejected": -3.1087257862091064,
|
| 384 |
"step": 120
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 0.26171159382360637,
|
| 388 |
+
"grad_norm": 33.03440558906921,
|
| 389 |
"learning_rate": 4.612975213859487e-07,
|
| 390 |
+
"logits/chosen": -2.584962844848633,
|
| 391 |
+
"logits/rejected": -2.54630970954895,
|
| 392 |
+
"logps/chosen": -1.1838756799697876,
|
| 393 |
+
"logps/rejected": -1.5963369607925415,
|
| 394 |
+
"loss": 1.5107,
|
| 395 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 396 |
+
"rewards/chosen": -2.367751359939575,
|
| 397 |
+
"rewards/margins": 0.8249226808547974,
|
| 398 |
+
"rewards/rejected": -3.192673921585083,
|
| 399 |
"step": 125
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.2721800575765506,
|
| 403 |
+
"grad_norm": 32.822453962687824,
|
| 404 |
"learning_rate": 4.5626458262912735e-07,
|
| 405 |
+
"logits/chosen": -2.6030898094177246,
|
| 406 |
+
"logits/rejected": -2.6097538471221924,
|
| 407 |
+
"logps/chosen": -1.1971023082733154,
|
| 408 |
+
"logps/rejected": -1.6219829320907593,
|
| 409 |
+
"loss": 1.4413,
|
| 410 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 411 |
+
"rewards/chosen": -2.394204616546631,
|
| 412 |
+
"rewards/margins": 0.8497610092163086,
|
| 413 |
+
"rewards/rejected": -3.2439658641815186,
|
| 414 |
"step": 130
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 0.2826485213294949,
|
| 418 |
+
"grad_norm": 28.065819409874635,
|
| 419 |
"learning_rate": 4.5095513994085974e-07,
|
| 420 |
+
"logits/chosen": -2.538353204727173,
|
| 421 |
+
"logits/rejected": -2.5087146759033203,
|
| 422 |
+
"logps/chosen": -1.0895415544509888,
|
| 423 |
+
"logps/rejected": -1.5814229249954224,
|
| 424 |
+
"loss": 1.4933,
|
| 425 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 426 |
+
"rewards/chosen": -2.1790831089019775,
|
| 427 |
+
"rewards/margins": 0.9837629199028015,
|
| 428 |
+
"rewards/rejected": -3.1628458499908447,
|
| 429 |
"step": 135
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.29311698508243916,
|
| 433 |
+
"grad_norm": 32.58412935415965,
|
| 434 |
"learning_rate": 4.453763107901675e-07,
|
| 435 |
+
"logits/chosen": -2.5400352478027344,
|
| 436 |
+
"logits/rejected": -2.5418732166290283,
|
| 437 |
+
"logps/chosen": -1.3108186721801758,
|
| 438 |
+
"logps/rejected": -1.5650501251220703,
|
| 439 |
+
"loss": 1.542,
|
| 440 |
+
"rewards/accuracies": 0.606249988079071,
|
| 441 |
+
"rewards/chosen": -2.6216373443603516,
|
| 442 |
+
"rewards/margins": 0.5084627866744995,
|
| 443 |
+
"rewards/rejected": -3.1301002502441406,
|
| 444 |
"step": 140
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.3035854488353834,
|
| 448 |
+
"grad_norm": 30.400278372842216,
|
| 449 |
"learning_rate": 4.395355737667985e-07,
|
| 450 |
+
"logits/chosen": -2.5238261222839355,
|
| 451 |
+
"logits/rejected": -2.5089011192321777,
|
| 452 |
+
"logps/chosen": -1.1779248714447021,
|
| 453 |
+
"logps/rejected": -1.6300004720687866,
|
| 454 |
+
"loss": 1.4238,
|
| 455 |
"rewards/accuracies": 0.675000011920929,
|
| 456 |
+
"rewards/chosen": -2.3558497428894043,
|
| 457 |
+
"rewards/margins": 0.9041512608528137,
|
| 458 |
+
"rewards/rejected": -3.2600009441375732,
|
| 459 |
"step": 145
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.31405391258832765,
|
| 463 |
+
"grad_norm": 60.958920561437736,
|
| 464 |
"learning_rate": 4.3344075855595097e-07,
|
| 465 |
+
"logits/chosen": -2.566622734069824,
|
| 466 |
+
"logits/rejected": -2.518723964691162,
|
| 467 |
+
"logps/chosen": -1.3376280069351196,
|
| 468 |
+
"logps/rejected": -1.7097371816635132,
|
| 469 |
+
"loss": 1.4812,
|
| 470 |
+
"rewards/accuracies": 0.59375,
|
| 471 |
+
"rewards/chosen": -2.6752560138702393,
|
| 472 |
+
"rewards/margins": 0.7442184686660767,
|
| 473 |
+
"rewards/rejected": -3.4194743633270264,
|
| 474 |
"step": 150
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 0.3245223763412719,
|
| 478 |
+
"grad_norm": 62.56084518909412,
|
| 479 |
"learning_rate": 4.271000354423425e-07,
|
| 480 |
+
"logits/chosen": -2.484179735183716,
|
| 481 |
+
"logits/rejected": -2.459049701690674,
|
| 482 |
+
"logps/chosen": -1.3308216333389282,
|
| 483 |
+
"logps/rejected": -1.9237397909164429,
|
| 484 |
+
"loss": 1.387,
|
| 485 |
"rewards/accuracies": 0.699999988079071,
|
| 486 |
+
"rewards/chosen": -2.6616432666778564,
|
| 487 |
+
"rewards/margins": 1.1858360767364502,
|
| 488 |
+
"rewards/rejected": -3.8474795818328857,
|
| 489 |
"step": 155
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.33499084009421615,
|
| 493 |
+
"grad_norm": 38.23483378023031,
|
| 494 |
"learning_rate": 4.2052190435769554e-07,
|
| 495 |
+
"logits/chosen": -2.454179286956787,
|
| 496 |
+
"logits/rejected": -2.4418118000030518,
|
| 497 |
+
"logps/chosen": -1.250301718711853,
|
| 498 |
+
"logps/rejected": -1.8722400665283203,
|
| 499 |
+
"loss": 1.3988,
|
| 500 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 501 |
+
"rewards/chosen": -2.500603437423706,
|
| 502 |
+
"rewards/margins": 1.2438766956329346,
|
| 503 |
+
"rewards/rejected": -3.7444801330566406,
|
| 504 |
"step": 160
|
| 505 |
},
|
| 506 |
{
|
| 507 |
"epoch": 0.34545930384716045,
|
| 508 |
+
"grad_norm": 36.680206717248936,
|
| 509 |
"learning_rate": 4.137151834863213e-07,
|
| 510 |
+
"logits/chosen": -2.516385316848755,
|
| 511 |
+
"logits/rejected": -2.476135730743408,
|
| 512 |
+
"logps/chosen": -1.2074108123779297,
|
| 513 |
+
"logps/rejected": -1.579379916191101,
|
| 514 |
+
"loss": 1.3981,
|
| 515 |
+
"rewards/accuracies": 0.65625,
|
| 516 |
+
"rewards/chosen": -2.4148216247558594,
|
| 517 |
+
"rewards/margins": 0.7439382672309875,
|
| 518 |
+
"rewards/rejected": -3.158759832382202,
|
| 519 |
"step": 165
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.3559277676001047,
|
| 523 |
+
"grad_norm": 45.599190265402456,
|
| 524 |
"learning_rate": 4.0668899744407567e-07,
|
| 525 |
+
"logits/chosen": -2.5008692741394043,
|
| 526 |
+
"logits/rejected": -2.456566572189331,
|
| 527 |
+
"logps/chosen": -1.2382426261901855,
|
| 528 |
+
"logps/rejected": -1.7240244150161743,
|
| 529 |
+
"loss": 1.3766,
|
| 530 |
+
"rewards/accuracies": 0.71875,
|
| 531 |
+
"rewards/chosen": -2.476485252380371,
|
| 532 |
+
"rewards/margins": 0.9715633392333984,
|
| 533 |
+
"rewards/rejected": -3.4480488300323486,
|
| 534 |
"step": 170
|
| 535 |
},
|
| 536 |
{
|
| 537 |
"epoch": 0.36639623135304894,
|
| 538 |
+
"grad_norm": 44.372524637477866,
|
| 539 |
"learning_rate": 3.994527650465352e-07,
|
| 540 |
+
"logits/chosen": -2.417241096496582,
|
| 541 |
+
"logits/rejected": -2.406471014022827,
|
| 542 |
+
"logps/chosen": -1.3205429315567017,
|
| 543 |
+
"logps/rejected": -1.8458646535873413,
|
| 544 |
+
"loss": 1.3967,
|
| 545 |
+
"rewards/accuracies": 0.675000011920929,
|
| 546 |
+
"rewards/chosen": -2.6410858631134033,
|
| 547 |
+
"rewards/margins": 1.0506436824798584,
|
| 548 |
+
"rewards/rejected": -3.6917293071746826,
|
| 549 |
"step": 175
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.3768646951059932,
|
| 553 |
+
"grad_norm": 59.067677833224565,
|
| 554 |
"learning_rate": 3.920161866827889e-07,
|
| 555 |
+
"logits/chosen": -2.438202381134033,
|
| 556 |
+
"logits/rejected": -2.4266159534454346,
|
| 557 |
+
"logps/chosen": -1.3483916521072388,
|
| 558 |
+
"logps/rejected": -1.9187591075897217,
|
| 559 |
+
"loss": 1.404,
|
| 560 |
+
"rewards/accuracies": 0.706250011920929,
|
| 561 |
+
"rewards/chosen": -2.6967833042144775,
|
| 562 |
+
"rewards/margins": 1.1407346725463867,
|
| 563 |
+
"rewards/rejected": -3.8375182151794434,
|
| 564 |
"step": 180
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.38733315885893743,
|
| 568 |
+
"grad_norm": 47.3796081172543,
|
| 569 |
"learning_rate": 3.8438923131177237e-07,
|
| 570 |
+
"logits/chosen": -2.385611057281494,
|
| 571 |
+
"logits/rejected": -2.4027440547943115,
|
| 572 |
+
"logps/chosen": -1.4026473760604858,
|
| 573 |
+
"logps/rejected": -1.966017723083496,
|
| 574 |
+
"loss": 1.3488,
|
| 575 |
+
"rewards/accuracies": 0.706250011920929,
|
| 576 |
+
"rewards/chosen": -2.8052947521209717,
|
| 577 |
+
"rewards/margins": 1.1267404556274414,
|
| 578 |
+
"rewards/rejected": -3.932035446166992,
|
| 579 |
"step": 185
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.39780162261188173,
|
| 583 |
+
"grad_norm": 35.917242818850596,
|
| 584 |
"learning_rate": 3.765821230985757e-07,
|
| 585 |
+
"logits/chosen": -2.374563455581665,
|
| 586 |
+
"logits/rejected": -2.382572889328003,
|
| 587 |
+
"logps/chosen": -1.4945417642593384,
|
| 588 |
+
"logps/rejected": -2.0413851737976074,
|
| 589 |
+
"loss": 1.3537,
|
| 590 |
+
"rewards/accuracies": 0.675000011920929,
|
| 591 |
+
"rewards/chosen": -2.9890835285186768,
|
| 592 |
+
"rewards/margins": 1.0936866998672485,
|
| 593 |
+
"rewards/rejected": -4.082770347595215,
|
| 594 |
"step": 190
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.408270086364826,
|
| 598 |
+
"grad_norm": 51.34440878644271,
|
| 599 |
"learning_rate": 3.6860532770864005e-07,
|
| 600 |
+
"logits/chosen": -2.3812708854675293,
|
| 601 |
+
"logits/rejected": -2.3748836517333984,
|
| 602 |
+
"logps/chosen": -1.507237195968628,
|
| 603 |
+
"logps/rejected": -2.213073253631592,
|
| 604 |
+
"loss": 1.4291,
|
| 605 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 606 |
+
"rewards/chosen": -3.014474391937256,
|
| 607 |
+
"rewards/margins": 1.4116714000701904,
|
| 608 |
+
"rewards/rejected": -4.426146507263184,
|
| 609 |
"step": 195
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.4187385501177702,
|
| 613 |
+
"grad_norm": 50.386830292932075,
|
| 614 |
"learning_rate": 3.604695382782159e-07,
|
| 615 |
+
"logits/chosen": -2.2745728492736816,
|
| 616 |
+
"logits/rejected": -2.256195545196533,
|
| 617 |
+
"logps/chosen": -1.647243857383728,
|
| 618 |
+
"logps/rejected": -2.149318218231201,
|
| 619 |
+
"loss": 1.3579,
|
| 620 |
+
"rewards/accuracies": 0.637499988079071,
|
| 621 |
+
"rewards/chosen": -3.294487714767456,
|
| 622 |
+
"rewards/margins": 1.0041488409042358,
|
| 623 |
+
"rewards/rejected": -4.298636436462402,
|
| 624 |
"step": 200
|
| 625 |
},
|
| 626 |
{
|
| 627 |
"epoch": 0.42920701387071447,
|
| 628 |
+
"grad_norm": 65.73777051804583,
|
| 629 |
"learning_rate": 3.5218566107988867e-07,
|
| 630 |
+
"logits/chosen": -2.335175037384033,
|
| 631 |
+
"logits/rejected": -2.326826333999634,
|
| 632 |
+
"logps/chosen": -1.8318570852279663,
|
| 633 |
+
"logps/rejected": -2.546074390411377,
|
| 634 |
+
"loss": 1.3635,
|
| 635 |
+
"rewards/accuracies": 0.675000011920929,
|
| 636 |
+
"rewards/chosen": -3.6637141704559326,
|
| 637 |
+
"rewards/margins": 1.4284346103668213,
|
| 638 |
+
"rewards/rejected": -5.092148780822754,
|
| 639 |
"step": 205
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.4396754776236587,
|
| 643 |
+
"grad_norm": 57.20248301568887,
|
| 644 |
"learning_rate": 3.4376480090239047e-07,
|
| 645 |
+
"logits/chosen": -2.184452533721924,
|
| 646 |
+
"logits/rejected": -2.1850171089172363,
|
| 647 |
+
"logps/chosen": -1.9411817789077759,
|
| 648 |
+
"logps/rejected": -2.38714861869812,
|
| 649 |
+
"loss": 1.468,
|
| 650 |
"rewards/accuracies": 0.699999988079071,
|
| 651 |
+
"rewards/chosen": -3.8823635578155518,
|
| 652 |
+
"rewards/margins": 0.8919339179992676,
|
| 653 |
+
"rewards/rejected": -4.77429723739624,
|
| 654 |
"step": 210
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.45014394137660296,
|
| 658 |
+
"grad_norm": 57.18796146027631,
|
| 659 |
"learning_rate": 3.3521824616429284e-07,
|
| 660 |
+
"logits/chosen": -2.20176100730896,
|
| 661 |
+
"logits/rejected": -2.227836847305298,
|
| 662 |
+
"logps/chosen": -1.8610153198242188,
|
| 663 |
+
"logps/rejected": -2.4027678966522217,
|
| 664 |
+
"loss": 1.3628,
|
| 665 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 666 |
+
"rewards/chosen": -3.7220306396484375,
|
| 667 |
+
"rewards/margins": 1.083505392074585,
|
| 668 |
+
"rewards/rejected": -4.805535793304443,
|
| 669 |
"step": 215
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.46061240512954726,
|
| 673 |
+
"grad_norm": 49.48323591731237,
|
| 674 |
"learning_rate": 3.265574537815398e-07,
|
| 675 |
+
"logits/chosen": -2.202357053756714,
|
| 676 |
+
"logits/rejected": -2.212911367416382,
|
| 677 |
+
"logps/chosen": -1.8679640293121338,
|
| 678 |
+
"logps/rejected": -2.4319264888763428,
|
| 679 |
+
"loss": 1.262,
|
| 680 |
+
"rewards/accuracies": 0.71875,
|
| 681 |
+
"rewards/chosen": -3.7359280586242676,
|
| 682 |
+
"rewards/margins": 1.127925157546997,
|
| 683 |
+
"rewards/rejected": -4.8638529777526855,
|
| 684 |
"step": 220
|
| 685 |
},
|
| 686 |
{
|
| 687 |
"epoch": 0.4710808688824915,
|
| 688 |
+
"grad_norm": 55.810856216454965,
|
| 689 |
"learning_rate": 3.1779403380910425e-07,
|
| 690 |
+
"logits/chosen": -2.1685309410095215,
|
| 691 |
+
"logits/rejected": -2.140174388885498,
|
| 692 |
+
"logps/chosen": -2.047653913497925,
|
| 693 |
+
"logps/rejected": -2.60907244682312,
|
| 694 |
+
"loss": 1.3463,
|
| 695 |
+
"rewards/accuracies": 0.71875,
|
| 696 |
+
"rewards/chosen": -4.09530782699585,
|
| 697 |
+
"rewards/margins": 1.122836709022522,
|
| 698 |
+
"rewards/rejected": -5.21814489364624,
|
| 699 |
"step": 225
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.48154933263543576,
|
| 703 |
+
"grad_norm": 65.54809342203177,
|
| 704 |
"learning_rate": 3.0893973387735683e-07,
|
| 705 |
+
"logits/chosen": -2.1521670818328857,
|
| 706 |
+
"logits/rejected": -2.1663339138031006,
|
| 707 |
+
"logps/chosen": -2.1521265506744385,
|
| 708 |
+
"logps/rejected": -2.8482584953308105,
|
| 709 |
+
"loss": 1.273,
|
| 710 |
+
"rewards/accuracies": 0.71875,
|
| 711 |
+
"rewards/chosen": -4.304253101348877,
|
| 712 |
+
"rewards/margins": 1.3922632932662964,
|
| 713 |
+
"rewards/rejected": -5.696516990661621,
|
| 714 |
"step": 230
|
| 715 |
},
|
| 716 |
{
|
| 717 |
"epoch": 0.49201779638838,
|
| 718 |
+
"grad_norm": 54.795057799713966,
|
| 719 |
"learning_rate": 3.000064234440111e-07,
|
| 720 |
+
"logits/chosen": -2.0703492164611816,
|
| 721 |
+
"logits/rejected": -2.0829060077667236,
|
| 722 |
+
"logps/chosen": -2.1279358863830566,
|
| 723 |
+
"logps/rejected": -2.6611745357513428,
|
| 724 |
+
"loss": 1.3093,
|
| 725 |
"rewards/accuracies": 0.706250011920929,
|
| 726 |
+
"rewards/chosen": -4.255871772766113,
|
| 727 |
+
"rewards/margins": 1.0664775371551514,
|
| 728 |
+
"rewards/rejected": -5.3223490715026855,
|
| 729 |
"step": 235
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.5024862601413242,
|
| 733 |
+
"grad_norm": 56.25864061883905,
|
| 734 |
"learning_rate": 2.910060778827554e-07,
|
| 735 |
+
"logits/chosen": -2.075435161590576,
|
| 736 |
+
"logits/rejected": -2.08357310295105,
|
| 737 |
+
"logps/chosen": -2.1075491905212402,
|
| 738 |
+
"logps/rejected": -2.84082293510437,
|
| 739 |
+
"loss": 1.3276,
|
| 740 |
"rewards/accuracies": 0.71875,
|
| 741 |
+
"rewards/chosen": -4.2150983810424805,
|
| 742 |
+
"rewards/margins": 1.4665473699569702,
|
| 743 |
+
"rewards/rejected": -5.68164587020874,
|
| 744 |
"step": 240
|
| 745 |
},
|
| 746 |
{
|
| 747 |
"epoch": 0.5129547238942685,
|
| 748 |
+
"grad_norm": 62.1729319764741,
|
| 749 |
"learning_rate": 2.8195076242990116e-07,
|
| 750 |
+
"logits/chosen": -2.0973756313323975,
|
| 751 |
+
"logits/rejected": -2.1240689754486084,
|
| 752 |
+
"logps/chosen": -2.0223472118377686,
|
| 753 |
+
"logps/rejected": -2.807509660720825,
|
| 754 |
+
"loss": 1.2667,
|
| 755 |
+
"rewards/accuracies": 0.737500011920929,
|
| 756 |
+
"rewards/chosen": -4.044694423675537,
|
| 757 |
+
"rewards/margins": 1.570324182510376,
|
| 758 |
+
"rewards/rejected": -5.61501932144165,
|
| 759 |
"step": 245
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 0.5234231876472127,
|
| 763 |
+
"grad_norm": 50.46640637368305,
|
| 764 |
"learning_rate": 2.7285261601056697e-07,
|
| 765 |
+
"logits/chosen": -1.9956550598144531,
|
| 766 |
+
"logits/rejected": -2.027296543121338,
|
| 767 |
+
"logps/chosen": -2.24354887008667,
|
| 768 |
+
"logps/rejected": -3.0298171043395996,
|
| 769 |
+
"loss": 1.3873,
|
| 770 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 771 |
+
"rewards/chosen": -4.48709774017334,
|
| 772 |
+
"rewards/margins": 1.5725353956222534,
|
| 773 |
+
"rewards/rejected": -6.059634208679199,
|
| 774 |
"step": 250
|
| 775 |
},
|
| 776 |
{
|
| 777 |
"epoch": 0.533891651400157,
|
| 778 |
+
"grad_norm": 60.6854410849544,
|
| 779 |
"learning_rate": 2.6372383496608186e-07,
|
| 780 |
+
"logits/chosen": -2.0544848442077637,
|
| 781 |
+
"logits/rejected": -2.0431137084960938,
|
| 782 |
+
"logps/chosen": -2.3652796745300293,
|
| 783 |
+
"logps/rejected": -3.058457136154175,
|
| 784 |
+
"loss": 1.3487,
|
| 785 |
+
"rewards/accuracies": 0.6875,
|
| 786 |
+
"rewards/chosen": -4.730559349060059,
|
| 787 |
+
"rewards/margins": 1.386354923248291,
|
| 788 |
+
"rewards/rejected": -6.11691427230835,
|
| 789 |
"step": 255
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 0.5443601151531012,
|
| 793 |
+
"grad_norm": 73.46613876186566,
|
| 794 |
"learning_rate": 2.5457665670441937e-07,
|
| 795 |
+
"logits/chosen": -1.9965794086456299,
|
| 796 |
+
"logits/rejected": -2.0038514137268066,
|
| 797 |
+
"logps/chosen": -2.4099831581115723,
|
| 798 |
+
"logps/rejected": -3.187678337097168,
|
| 799 |
+
"loss": 1.2961,
|
| 800 |
+
"rewards/accuracies": 0.75,
|
| 801 |
+
"rewards/chosen": -4.8199663162231445,
|
| 802 |
+
"rewards/margins": 1.5553903579711914,
|
| 803 |
+
"rewards/rejected": -6.375356674194336,
|
| 804 |
"step": 260
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 0.5548285789060455,
|
| 808 |
+
"grad_norm": 74.04251596633621,
|
| 809 |
"learning_rate": 2.454233432955807e-07,
|
| 810 |
+
"logits/chosen": -1.9874013662338257,
|
| 811 |
+
"logits/rejected": -1.9983584880828857,
|
| 812 |
+
"logps/chosen": -2.573876142501831,
|
| 813 |
+
"logps/rejected": -3.171705722808838,
|
| 814 |
+
"loss": 1.3148,
|
| 815 |
+
"rewards/accuracies": 0.71875,
|
| 816 |
+
"rewards/chosen": -5.147752285003662,
|
| 817 |
+
"rewards/margins": 1.1956590414047241,
|
| 818 |
+
"rewards/rejected": -6.343411445617676,
|
| 819 |
"step": 265
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 0.5652970426589898,
|
| 823 |
+
"grad_norm": 76.35568912258938,
|
| 824 |
"learning_rate": 2.3627616503391812e-07,
|
| 825 |
+
"logits/chosen": -2.017387628555298,
|
| 826 |
+
"logits/rejected": -2.024291753768921,
|
| 827 |
+
"logps/chosen": -2.639585018157959,
|
| 828 |
+
"logps/rejected": -3.2247214317321777,
|
| 829 |
+
"loss": 1.2604,
|
| 830 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 831 |
+
"rewards/chosen": -5.279170036315918,
|
| 832 |
+
"rewards/margins": 1.1702723503112793,
|
| 833 |
+
"rewards/rejected": -6.4494428634643555,
|
| 834 |
"step": 270
|
| 835 |
},
|
| 836 |
{
|
| 837 |
"epoch": 0.575765506411934,
|
| 838 |
+
"grad_norm": 47.964272961306655,
|
| 839 |
"learning_rate": 2.2714738398943308e-07,
|
| 840 |
+
"logits/chosen": -1.9820992946624756,
|
| 841 |
+
"logits/rejected": -1.986830472946167,
|
| 842 |
+
"logps/chosen": -2.4876163005828857,
|
| 843 |
+
"logps/rejected": -3.4850916862487793,
|
| 844 |
+
"loss": 1.2026,
|
| 845 |
+
"rewards/accuracies": 0.768750011920929,
|
| 846 |
+
"rewards/chosen": -4.9752326011657715,
|
| 847 |
+
"rewards/margins": 1.9949508905410767,
|
| 848 |
+
"rewards/rejected": -6.970183372497559,
|
| 849 |
"step": 275
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 0.5862339701648783,
|
| 853 |
+
"grad_norm": 66.6426445717076,
|
| 854 |
"learning_rate": 2.1804923757009882e-07,
|
| 855 |
+
"logits/chosen": -1.9717257022857666,
|
| 856 |
+
"logits/rejected": -1.9761455059051514,
|
| 857 |
+
"logps/chosen": -2.604233980178833,
|
| 858 |
+
"logps/rejected": -3.449998140335083,
|
| 859 |
+
"loss": 1.3207,
|
| 860 |
+
"rewards/accuracies": 0.737500011920929,
|
| 861 |
+
"rewards/chosen": -5.208467960357666,
|
| 862 |
+
"rewards/margins": 1.691528081893921,
|
| 863 |
+
"rewards/rejected": -6.899996280670166,
|
| 864 |
"step": 280
|
| 865 |
},
|
| 866 |
{
|
| 867 |
"epoch": 0.5967024339178225,
|
| 868 |
+
"grad_norm": 78.85704380261409,
|
| 869 |
"learning_rate": 2.089939221172446e-07,
|
| 870 |
+
"logits/chosen": -1.9817512035369873,
|
| 871 |
+
"logits/rejected": -1.976833701133728,
|
| 872 |
+
"logps/chosen": -2.4398930072784424,
|
| 873 |
+
"logps/rejected": -3.1973671913146973,
|
| 874 |
+
"loss": 1.275,
|
| 875 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 876 |
+
"rewards/chosen": -4.879786014556885,
|
| 877 |
+
"rewards/margins": 1.5149486064910889,
|
| 878 |
+
"rewards/rejected": -6.3947343826293945,
|
| 879 |
"step": 285
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 0.6071708976707668,
|
| 883 |
+
"grad_norm": 70.57880724344702,
|
| 884 |
"learning_rate": 1.9999357655598891e-07,
|
| 885 |
+
"logits/chosen": -1.9725377559661865,
|
| 886 |
+
"logits/rejected": -1.9514259099960327,
|
| 887 |
+
"logps/chosen": -2.342991828918457,
|
| 888 |
+
"logps/rejected": -2.9306890964508057,
|
| 889 |
+
"loss": 1.3026,
|
| 890 |
+
"rewards/accuracies": 0.71875,
|
| 891 |
+
"rewards/chosen": -4.685983657836914,
|
| 892 |
+
"rewards/margins": 1.1753947734832764,
|
| 893 |
+
"rewards/rejected": -5.861378192901611,
|
| 894 |
"step": 290
|
| 895 |
},
|
| 896 |
{
|
| 897 |
"epoch": 0.6176393614237111,
|
| 898 |
+
"grad_norm": 64.72926534110535,
|
| 899 |
"learning_rate": 1.9106026612264315e-07,
|
| 900 |
+
"logits/chosen": -2.001755714416504,
|
| 901 |
+
"logits/rejected": -1.9883407354354858,
|
| 902 |
+
"logps/chosen": -2.397254705429077,
|
| 903 |
+
"logps/rejected": -2.957282781600952,
|
| 904 |
+
"loss": 1.249,
|
| 905 |
+
"rewards/accuracies": 0.71875,
|
| 906 |
+
"rewards/chosen": -4.794509410858154,
|
| 907 |
+
"rewards/margins": 1.1200562715530396,
|
| 908 |
+
"rewards/rejected": -5.914565563201904,
|
| 909 |
"step": 295
|
| 910 |
},
|
| 911 |
{
|
| 912 |
"epoch": 0.6281078251766553,
|
| 913 |
+
"grad_norm": 91.3728151363829,
|
| 914 |
"learning_rate": 1.8220596619089573e-07,
|
| 915 |
+
"logits/chosen": -1.8486881256103516,
|
| 916 |
+
"logits/rejected": -1.8306758403778076,
|
| 917 |
+
"logps/chosen": -2.625481128692627,
|
| 918 |
+
"logps/rejected": -3.437105178833008,
|
| 919 |
+
"loss": 1.2989,
|
| 920 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 921 |
+
"rewards/chosen": -5.250962257385254,
|
| 922 |
+
"rewards/margins": 1.6232483386993408,
|
| 923 |
+
"rewards/rejected": -6.874210357666016,
|
| 924 |
"step": 300
|
| 925 |
},
|
| 926 |
{
|
| 927 |
"epoch": 0.6385762889295996,
|
| 928 |
+
"grad_norm": 72.03502684523782,
|
| 929 |
"learning_rate": 1.7344254621846017e-07,
|
| 930 |
+
"logits/chosen": -1.9142115116119385,
|
| 931 |
+
"logits/rejected": -1.9298557043075562,
|
| 932 |
+
"logps/chosen": -2.655311107635498,
|
| 933 |
+
"logps/rejected": -3.314296007156372,
|
| 934 |
+
"loss": 1.2273,
|
| 935 |
+
"rewards/accuracies": 0.731249988079071,
|
| 936 |
+
"rewards/chosen": -5.310622215270996,
|
| 937 |
+
"rewards/margins": 1.3179702758789062,
|
| 938 |
+
"rewards/rejected": -6.628592014312744,
|
| 939 |
"step": 305
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 0.6490447526825438,
|
| 943 |
+
"grad_norm": 69.39610632162989,
|
| 944 |
"learning_rate": 1.647817538357072e-07,
|
| 945 |
+
"logits/chosen": -1.8438358306884766,
|
| 946 |
+
"logits/rejected": -1.8745906352996826,
|
| 947 |
+
"logps/chosen": -2.8499069213867188,
|
| 948 |
+
"logps/rejected": -3.5957629680633545,
|
| 949 |
+
"loss": 1.2436,
|
| 950 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 951 |
+
"rewards/chosen": -5.6998138427734375,
|
| 952 |
+
"rewards/margins": 1.491711974143982,
|
| 953 |
+
"rewards/rejected": -7.191525936126709,
|
| 954 |
"step": 310
|
| 955 |
},
|
| 956 |
{
|
| 957 |
"epoch": 0.6595132164354881,
|
| 958 |
+
"grad_norm": 60.61281750085851,
|
| 959 |
"learning_rate": 1.562351990976095e-07,
|
| 960 |
+
"logits/chosen": -1.9220491647720337,
|
| 961 |
+
"logits/rejected": -1.9316775798797607,
|
| 962 |
+
"logps/chosen": -2.5944266319274902,
|
| 963 |
+
"logps/rejected": -3.3770358562469482,
|
| 964 |
+
"loss": 1.2926,
|
| 965 |
"rewards/accuracies": 0.6937500238418579,
|
| 966 |
+
"rewards/chosen": -5.1888532638549805,
|
| 967 |
+
"rewards/margins": 1.5652191638946533,
|
| 968 |
+
"rewards/rejected": -6.7540717124938965,
|
| 969 |
"step": 315
|
| 970 |
},
|
| 971 |
{
|
| 972 |
"epoch": 0.6699816801884323,
|
| 973 |
+
"grad_norm": 64.49517967587298,
|
| 974 |
"learning_rate": 1.478143389201113e-07,
|
| 975 |
+
"logits/chosen": -1.954124093055725,
|
| 976 |
+
"logits/rejected": -1.9501478672027588,
|
| 977 |
+
"logps/chosen": -2.416090250015259,
|
| 978 |
+
"logps/rejected": -3.2859389781951904,
|
| 979 |
+
"loss": 1.1985,
|
| 980 |
+
"rewards/accuracies": 0.706250011920929,
|
| 981 |
+
"rewards/chosen": -4.832180500030518,
|
| 982 |
+
"rewards/margins": 1.7396974563598633,
|
| 983 |
+
"rewards/rejected": -6.571877956390381,
|
| 984 |
"step": 320
|
| 985 |
},
|
| 986 |
{
|
| 987 |
"epoch": 0.6804501439413766,
|
| 988 |
+
"grad_norm": 77.10656027012405,
|
| 989 |
"learning_rate": 1.3953046172178413e-07,
|
| 990 |
+
"logits/chosen": -1.9018971920013428,
|
| 991 |
+
"logits/rejected": -1.9205232858657837,
|
| 992 |
+
"logps/chosen": -2.819211483001709,
|
| 993 |
+
"logps/rejected": -3.631371021270752,
|
| 994 |
+
"loss": 1.3417,
|
| 995 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 996 |
+
"rewards/chosen": -5.638422966003418,
|
| 997 |
+
"rewards/margins": 1.6243181228637695,
|
| 998 |
+
"rewards/rejected": -7.262742042541504,
|
| 999 |
"step": 325
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
"epoch": 0.6909186076943209,
|
| 1003 |
+
"grad_norm": 64.1696366253842,
|
| 1004 |
"learning_rate": 1.3139467229135998e-07,
|
| 1005 |
+
"logits/chosen": -1.8825531005859375,
|
| 1006 |
+
"logits/rejected": -1.8870893716812134,
|
| 1007 |
+
"logps/chosen": -2.785536289215088,
|
| 1008 |
+
"logps/rejected": -3.764813184738159,
|
| 1009 |
+
"loss": 1.2374,
|
| 1010 |
"rewards/accuracies": 0.7437499761581421,
|
| 1011 |
+
"rewards/chosen": -5.571072578430176,
|
| 1012 |
+
"rewards/margins": 1.9585535526275635,
|
| 1013 |
+
"rewards/rejected": -7.529626369476318,
|
| 1014 |
"step": 330
|
| 1015 |
},
|
| 1016 |
{
|
| 1017 |
"epoch": 0.7013870714472651,
|
| 1018 |
+
"grad_norm": 80.55442176578629,
|
| 1019 |
"learning_rate": 1.2341787690142435e-07,
|
| 1020 |
+
"logits/chosen": -1.8318296670913696,
|
| 1021 |
+
"logits/rejected": -1.8602631092071533,
|
| 1022 |
+
"logps/chosen": -3.0103182792663574,
|
| 1023 |
+
"logps/rejected": -3.835423707962036,
|
| 1024 |
+
"loss": 1.2754,
|
| 1025 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1026 |
+
"rewards/chosen": -6.020636558532715,
|
| 1027 |
+
"rewards/margins": 1.6502106189727783,
|
| 1028 |
+
"rewards/rejected": -7.670847415924072,
|
| 1029 |
"step": 335
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
"epoch": 0.7118555352002094,
|
| 1033 |
+
"grad_norm": 87.70174157824101,
|
| 1034 |
"learning_rate": 1.1561076868822755e-07,
|
| 1035 |
+
"logits/chosen": -1.8182014226913452,
|
| 1036 |
+
"logits/rejected": -1.893776535987854,
|
| 1037 |
+
"logps/chosen": -2.9932782649993896,
|
| 1038 |
+
"logps/rejected": -4.012537002563477,
|
| 1039 |
+
"loss": 1.2641,
|
| 1040 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1041 |
+
"rewards/chosen": -5.986556529998779,
|
| 1042 |
+
"rewards/margins": 2.0385169982910156,
|
| 1043 |
+
"rewards/rejected": -8.025074005126953,
|
| 1044 |
"step": 340
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
"epoch": 0.7223239989531536,
|
| 1048 |
+
"grad_norm": 58.44045230920631,
|
| 1049 |
"learning_rate": 1.0798381331721107e-07,
|
| 1050 |
+
"logits/chosen": -1.874028205871582,
|
| 1051 |
+
"logits/rejected": -1.8978573083877563,
|
| 1052 |
+
"logps/chosen": -2.8497276306152344,
|
| 1053 |
+
"logps/rejected": -3.7342123985290527,
|
| 1054 |
+
"loss": 1.2615,
|
| 1055 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1056 |
+
"rewards/chosen": -5.699455261230469,
|
| 1057 |
+
"rewards/margins": 1.7689688205718994,
|
| 1058 |
+
"rewards/rejected": -7.4684247970581055,
|
| 1059 |
"step": 345
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
"epoch": 0.7327924627060979,
|
| 1063 |
+
"grad_norm": 72.00876135010044,
|
| 1064 |
"learning_rate": 1.0054723495346482e-07,
|
| 1065 |
+
"logits/chosen": -1.9043935537338257,
|
| 1066 |
+
"logits/rejected": -1.9324207305908203,
|
| 1067 |
+
"logps/chosen": -2.6209471225738525,
|
| 1068 |
+
"logps/rejected": -3.339960813522339,
|
| 1069 |
+
"loss": 1.2008,
|
| 1070 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1071 |
+
"rewards/chosen": -5.241894245147705,
|
| 1072 |
+
"rewards/margins": 1.4380273818969727,
|
| 1073 |
+
"rewards/rejected": -6.679921627044678,
|
| 1074 |
"step": 350
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
"epoch": 0.7432609264590422,
|
| 1078 |
+
"grad_norm": 55.618657855003114,
|
| 1079 |
"learning_rate": 9.331100255592436e-08,
|
| 1080 |
+
"logits/chosen": -1.8889986276626587,
|
| 1081 |
+
"logits/rejected": -1.9200356006622314,
|
| 1082 |
+
"logps/chosen": -2.522282838821411,
|
| 1083 |
+
"logps/rejected": -3.240691661834717,
|
| 1084 |
+
"loss": 1.3353,
|
| 1085 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1086 |
+
"rewards/chosen": -5.044565677642822,
|
| 1087 |
+
"rewards/margins": 1.43681800365448,
|
| 1088 |
+
"rewards/rejected": -6.481383323669434,
|
| 1089 |
"step": 355
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
"epoch": 0.7537293902119864,
|
| 1093 |
+
"grad_norm": 67.73876893903605,
|
| 1094 |
"learning_rate": 8.628481651367875e-08,
|
| 1095 |
+
"logits/chosen": -1.8986294269561768,
|
| 1096 |
+
"logits/rejected": -1.9174484014511108,
|
| 1097 |
+
"logps/chosen": -2.437084674835205,
|
| 1098 |
+
"logps/rejected": -3.152095317840576,
|
| 1099 |
+
"loss": 1.2784,
|
| 1100 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1101 |
+
"rewards/chosen": -4.87416934967041,
|
| 1102 |
+
"rewards/margins": 1.43002188205719,
|
| 1103 |
+
"rewards/rejected": -6.304190635681152,
|
| 1104 |
"step": 360
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
"epoch": 0.7641978539649307,
|
| 1108 |
+
"grad_norm": 60.67240818670899,
|
| 1109 |
"learning_rate": 7.947809564230445e-08,
|
| 1110 |
+
"logits/chosen": -1.921534538269043,
|
| 1111 |
+
"logits/rejected": -1.9346189498901367,
|
| 1112 |
+
"logps/chosen": -2.425230026245117,
|
| 1113 |
+
"logps/rejected": -3.2995903491973877,
|
| 1114 |
+
"loss": 1.2435,
|
| 1115 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1116 |
+
"rewards/chosen": -4.850460052490234,
|
| 1117 |
+
"rewards/margins": 1.7487205266952515,
|
| 1118 |
+
"rewards/rejected": -6.599180698394775,
|
| 1119 |
"step": 365
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
"epoch": 0.7746663177178749,
|
| 1123 |
+
"grad_norm": 59.56252419947236,
|
| 1124 |
"learning_rate": 7.289996455765748e-08,
|
| 1125 |
+
"logits/chosen": -1.898923635482788,
|
| 1126 |
+
"logits/rejected": -1.9162523746490479,
|
| 1127 |
+
"logps/chosen": -2.415956497192383,
|
| 1128 |
+
"logps/rejected": -3.294524669647217,
|
| 1129 |
+
"loss": 1.1245,
|
| 1130 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1131 |
+
"rewards/chosen": -4.831912994384766,
|
| 1132 |
+
"rewards/margins": 1.757136583328247,
|
| 1133 |
+
"rewards/rejected": -6.589049339294434,
|
| 1134 |
"step": 370
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
"epoch": 0.7851347814708192,
|
| 1138 |
+
"grad_norm": 71.07320248012725,
|
| 1139 |
"learning_rate": 6.655924144404906e-08,
|
| 1140 |
+
"logits/chosen": -1.866040587425232,
|
| 1141 |
+
"logits/rejected": -1.8857864141464233,
|
| 1142 |
+
"logps/chosen": -2.5201144218444824,
|
| 1143 |
+
"logps/rejected": -3.3979249000549316,
|
| 1144 |
+
"loss": 1.2524,
|
| 1145 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1146 |
+
"rewards/chosen": -5.040228843688965,
|
| 1147 |
+
"rewards/margins": 1.7556209564208984,
|
| 1148 |
+
"rewards/rejected": -6.795849800109863,
|
| 1149 |
"step": 375
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 0.7956032452237635,
|
| 1153 |
+
"grad_norm": 66.56871290886237,
|
| 1154 |
"learning_rate": 6.046442623320145e-08,
|
| 1155 |
+
"logits/chosen": -1.910027265548706,
|
| 1156 |
+
"logits/rejected": -1.9534069299697876,
|
| 1157 |
+
"logps/chosen": -2.5121512413024902,
|
| 1158 |
+
"logps/rejected": -3.314141035079956,
|
| 1159 |
+
"loss": 1.2226,
|
| 1160 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1161 |
+
"rewards/chosen": -5.0243024826049805,
|
| 1162 |
+
"rewards/margins": 1.6039783954620361,
|
| 1163 |
+
"rewards/rejected": -6.628282070159912,
|
| 1164 |
"step": 380
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
"epoch": 0.8060717089767077,
|
| 1168 |
+
"grad_norm": 62.398640149446855,
|
| 1169 |
"learning_rate": 5.4623689209832484e-08,
|
| 1170 |
+
"logits/chosen": -1.8576492071151733,
|
| 1171 |
+
"logits/rejected": -1.8747966289520264,
|
| 1172 |
+
"logps/chosen": -2.6333398818969727,
|
| 1173 |
+
"logps/rejected": -3.4325785636901855,
|
| 1174 |
+
"loss": 1.2061,
|
| 1175 |
"rewards/accuracies": 0.7562500238418579,
|
| 1176 |
+
"rewards/chosen": -5.266679763793945,
|
| 1177 |
+
"rewards/margins": 1.5984779596328735,
|
| 1178 |
+
"rewards/rejected": -6.865157127380371,
|
| 1179 |
"step": 385
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 0.816540172729652,
|
| 1183 |
+
"grad_norm": 70.95289047647472,
|
| 1184 |
"learning_rate": 4.904486005914027e-08,
|
| 1185 |
+
"logits/chosen": -1.8401978015899658,
|
| 1186 |
+
"logits/rejected": -1.8537371158599854,
|
| 1187 |
+
"logps/chosen": -2.873885154724121,
|
| 1188 |
+
"logps/rejected": -3.697465419769287,
|
| 1189 |
+
"loss": 1.1959,
|
| 1190 |
"rewards/accuracies": 0.731249988079071,
|
| 1191 |
+
"rewards/chosen": -5.747770309448242,
|
| 1192 |
+
"rewards/margins": 1.6471607685089111,
|
| 1193 |
+
"rewards/rejected": -7.394930839538574,
|
| 1194 |
"step": 390
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
"epoch": 0.8270086364825961,
|
| 1198 |
+
"grad_norm": 67.80264650886268,
|
| 1199 |
"learning_rate": 4.373541737087263e-08,
|
| 1200 |
+
"logits/chosen": -1.7799686193466187,
|
| 1201 |
+
"logits/rejected": -1.8199493885040283,
|
| 1202 |
+
"logps/chosen": -2.8214058876037598,
|
| 1203 |
+
"logps/rejected": -3.7097229957580566,
|
| 1204 |
+
"loss": 1.2552,
|
| 1205 |
+
"rewards/accuracies": 0.71875,
|
| 1206 |
+
"rewards/chosen": -5.6428117752075195,
|
| 1207 |
+
"rewards/margins": 1.7766335010528564,
|
| 1208 |
+
"rewards/rejected": -7.419445991516113,
|
| 1209 |
"step": 395
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
"epoch": 0.8374771002355405,
|
| 1213 |
+
"grad_norm": 65.87128834814716,
|
| 1214 |
"learning_rate": 3.8702478614051345e-08,
|
| 1215 |
+
"logits/chosen": -1.8025257587432861,
|
| 1216 |
+
"logits/rejected": -1.8182452917099,
|
| 1217 |
+
"logps/chosen": -2.924140214920044,
|
| 1218 |
+
"logps/rejected": -3.7234292030334473,
|
| 1219 |
+
"loss": 1.267,
|
| 1220 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1221 |
+
"rewards/chosen": -5.848280429840088,
|
| 1222 |
+
"rewards/margins": 1.5985779762268066,
|
| 1223 |
+
"rewards/rejected": -7.4468584060668945,
|
| 1224 |
"step": 400
|
| 1225 |
},
|
| 1226 |
{
|
| 1227 |
"epoch": 0.8374771002355405,
|
| 1228 |
+
"eval_logits/chosen": -1.8795839548110962,
|
| 1229 |
+
"eval_logits/rejected": -1.8947079181671143,
|
| 1230 |
+
"eval_logps/chosen": -2.906954526901245,
|
| 1231 |
+
"eval_logps/rejected": -3.8314788341522217,
|
| 1232 |
+
"eval_loss": 1.209315299987793,
|
| 1233 |
"eval_rewards/accuracies": 0.7319999933242798,
|
| 1234 |
+
"eval_rewards/chosen": -5.81390905380249,
|
| 1235 |
+
"eval_rewards/margins": 1.8490480184555054,
|
| 1236 |
+
"eval_rewards/rejected": -7.662957668304443,
|
| 1237 |
+
"eval_runtime": 171.5509,
|
| 1238 |
+
"eval_samples_per_second": 11.658,
|
| 1239 |
"eval_steps_per_second": 0.729,
|
| 1240 |
"step": 400
|
| 1241 |
},
|
| 1242 |
{
|
| 1243 |
"epoch": 0.8479455639884846,
|
| 1244 |
+
"grad_norm": 60.339051405492995,
|
| 1245 |
"learning_rate": 3.3952790595787986e-08,
|
| 1246 |
+
"logits/chosen": -1.8179550170898438,
|
| 1247 |
+
"logits/rejected": -1.843348741531372,
|
| 1248 |
+
"logps/chosen": -3.005042552947998,
|
| 1249 |
+
"logps/rejected": -3.817539930343628,
|
| 1250 |
+
"loss": 1.1835,
|
| 1251 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1252 |
+
"rewards/chosen": -6.010085105895996,
|
| 1253 |
+
"rewards/margins": 1.6249958276748657,
|
| 1254 |
+
"rewards/rejected": -7.635079860687256,
|
| 1255 |
"step": 405
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
"epoch": 0.8584140277414289,
|
| 1259 |
+
"grad_norm": 59.27380738977581,
|
| 1260 |
"learning_rate": 2.9492720416985e-08,
|
| 1261 |
+
"logits/chosen": -1.7845852375030518,
|
| 1262 |
+
"logits/rejected": -1.819650650024414,
|
| 1263 |
+
"logps/chosen": -2.9665215015411377,
|
| 1264 |
+
"logps/rejected": -3.7718894481658936,
|
| 1265 |
+
"loss": 1.2356,
|
| 1266 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1267 |
+
"rewards/chosen": -5.933043003082275,
|
| 1268 |
+
"rewards/margins": 1.610735297203064,
|
| 1269 |
+
"rewards/rejected": -7.543778896331787,
|
| 1270 |
"step": 410
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
"epoch": 0.8688824914943732,
|
| 1274 |
+
"grad_norm": 91.6278901379261,
|
| 1275 |
"learning_rate": 2.5328246937043525e-08,
|
| 1276 |
+
"logits/chosen": -1.8167459964752197,
|
| 1277 |
+
"logits/rejected": -1.827742576599121,
|
| 1278 |
+
"logps/chosen": -3.0838284492492676,
|
| 1279 |
+
"logps/rejected": -3.7930634021759033,
|
| 1280 |
+
"loss": 1.2442,
|
| 1281 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1282 |
+
"rewards/chosen": -6.167656898498535,
|
| 1283 |
+
"rewards/margins": 1.4184691905975342,
|
| 1284 |
+
"rewards/rejected": -7.586126804351807,
|
| 1285 |
"step": 415
|
| 1286 |
},
|
| 1287 |
{
|
| 1288 |
"epoch": 0.8793509552473174,
|
| 1289 |
+
"grad_norm": 76.88765480732745,
|
| 1290 |
"learning_rate": 2.1464952759020856e-08,
|
| 1291 |
+
"logits/chosen": -1.767154335975647,
|
| 1292 |
+
"logits/rejected": -1.7657482624053955,
|
| 1293 |
+
"logps/chosen": -2.938920497894287,
|
| 1294 |
+
"logps/rejected": -3.925022840499878,
|
| 1295 |
+
"loss": 1.1843,
|
| 1296 |
"rewards/accuracies": 0.699999988079071,
|
| 1297 |
+
"rewards/chosen": -5.877840995788574,
|
| 1298 |
+
"rewards/margins": 1.972204566001892,
|
| 1299 |
+
"rewards/rejected": -7.850045680999756,
|
| 1300 |
"step": 420
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
"epoch": 0.8898194190002617,
|
| 1304 |
+
"grad_norm": 71.47276585551965,
|
| 1305 |
"learning_rate": 1.7908016745981856e-08,
|
| 1306 |
+
"logits/chosen": -1.7983713150024414,
|
| 1307 |
+
"logits/rejected": -1.8315776586532593,
|
| 1308 |
+
"logps/chosen": -2.782409429550171,
|
| 1309 |
+
"logps/rejected": -3.797370433807373,
|
| 1310 |
+
"loss": 1.1376,
|
| 1311 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1312 |
+
"rewards/chosen": -5.564818859100342,
|
| 1313 |
+
"rewards/margins": 2.0299227237701416,
|
| 1314 |
+
"rewards/rejected": -7.594740867614746,
|
| 1315 |
"step": 425
|
| 1316 |
},
|
| 1317 |
{
|
| 1318 |
"epoch": 0.9002878827532059,
|
| 1319 |
+
"grad_norm": 73.92788729675112,
|
| 1320 |
"learning_rate": 1.4662207078575684e-08,
|
| 1321 |
+
"logits/chosen": -1.7802276611328125,
|
| 1322 |
+
"logits/rejected": -1.8125686645507812,
|
| 1323 |
+
"logps/chosen": -2.976722240447998,
|
| 1324 |
+
"logps/rejected": -3.855865955352783,
|
| 1325 |
+
"loss": 1.138,
|
| 1326 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1327 |
+
"rewards/chosen": -5.953444480895996,
|
| 1328 |
+
"rewards/margins": 1.7582868337631226,
|
| 1329 |
+
"rewards/rejected": -7.711731910705566,
|
| 1330 |
"step": 430
|
| 1331 |
},
|
| 1332 |
{
|
| 1333 |
"epoch": 0.9107563465061502,
|
| 1334 |
+
"grad_norm": 67.44618318869006,
|
| 1335 |
"learning_rate": 1.1731874863145142e-08,
|
| 1336 |
+
"logits/chosen": -1.7870285511016846,
|
| 1337 |
+
"logits/rejected": -1.7898231744766235,
|
| 1338 |
+
"logps/chosen": -2.865501880645752,
|
| 1339 |
+
"logps/rejected": -3.5326790809631348,
|
| 1340 |
+
"loss": 1.1911,
|
| 1341 |
"rewards/accuracies": 0.668749988079071,
|
| 1342 |
+
"rewards/chosen": -5.731003761291504,
|
| 1343 |
+
"rewards/margins": 1.3343536853790283,
|
| 1344 |
+
"rewards/rejected": -7.0653581619262695,
|
| 1345 |
"step": 435
|
| 1346 |
},
|
| 1347 |
{
|
| 1348 |
"epoch": 0.9212248102590945,
|
| 1349 |
+
"grad_norm": 80.91405517261822,
|
| 1350 |
"learning_rate": 9.12094829893642e-09,
|
| 1351 |
+
"logits/chosen": -1.8377736806869507,
|
| 1352 |
+
"logits/rejected": -1.8392757177352905,
|
| 1353 |
+
"logps/chosen": -2.909419536590576,
|
| 1354 |
+
"logps/rejected": -3.649876356124878,
|
| 1355 |
+
"loss": 1.2198,
|
| 1356 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1357 |
+
"rewards/chosen": -5.818839073181152,
|
| 1358 |
+
"rewards/margins": 1.4809141159057617,
|
| 1359 |
+
"rewards/rejected": -7.299752712249756,
|
| 1360 |
"step": 440
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
"epoch": 0.9316932740120387,
|
| 1364 |
+
"grad_norm": 71.48155448577984,
|
| 1365 |
"learning_rate": 6.832927412229017e-09,
|
| 1366 |
+
"logits/chosen": -1.7438195943832397,
|
| 1367 |
+
"logits/rejected": -1.7705223560333252,
|
| 1368 |
+
"logps/chosen": -2.9339725971221924,
|
| 1369 |
+
"logps/rejected": -3.8835110664367676,
|
| 1370 |
+
"loss": 1.2304,
|
| 1371 |
"rewards/accuracies": 0.737500011920929,
|
| 1372 |
+
"rewards/chosen": -5.867945194244385,
|
| 1373 |
+
"rewards/margins": 1.899076223373413,
|
| 1374 |
+
"rewards/rejected": -7.767022132873535,
|
| 1375 |
"step": 445
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
"epoch": 0.942161737764983,
|
| 1379 |
+
"grad_norm": 75.54710086328386,
|
| 1380 |
"learning_rate": 4.8708793644441086e-09,
|
| 1381 |
+
"logits/chosen": -1.8001352548599243,
|
| 1382 |
+
"logits/rejected": -1.8404371738433838,
|
| 1383 |
+
"logps/chosen": -2.698850154876709,
|
| 1384 |
+
"logps/rejected": -3.7303435802459717,
|
| 1385 |
+
"loss": 1.1496,
|
| 1386 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1387 |
+
"rewards/chosen": -5.397700309753418,
|
| 1388 |
+
"rewards/margins": 2.0629868507385254,
|
| 1389 |
+
"rewards/rejected": -7.460687160491943,
|
| 1390 |
"step": 450
|
| 1391 |
},
|
| 1392 |
{
|
| 1393 |
"epoch": 0.9526302015179272,
|
| 1394 |
+
"grad_norm": 67.21839705145132,
|
| 1395 |
"learning_rate": 3.2374343405217884e-09,
|
| 1396 |
+
"logits/chosen": -1.7952674627304077,
|
| 1397 |
+
"logits/rejected": -1.7791658639907837,
|
| 1398 |
+
"logps/chosen": -2.888683795928955,
|
| 1399 |
+
"logps/rejected": -3.7380385398864746,
|
| 1400 |
+
"loss": 1.2744,
|
| 1401 |
"rewards/accuracies": 0.706250011920929,
|
| 1402 |
+
"rewards/chosen": -5.77736759185791,
|
| 1403 |
+
"rewards/margins": 1.6987104415893555,
|
| 1404 |
+
"rewards/rejected": -7.476077079772949,
|
| 1405 |
"step": 455
|
| 1406 |
},
|
| 1407 |
{
|
| 1408 |
"epoch": 0.9630986652708715,
|
| 1409 |
+
"grad_norm": 83.08799813263991,
|
| 1410 |
"learning_rate": 1.9347820230782295e-09,
|
| 1411 |
+
"logits/chosen": -1.7771613597869873,
|
| 1412 |
+
"logits/rejected": -1.8283106088638306,
|
| 1413 |
+
"logps/chosen": -2.847031831741333,
|
| 1414 |
+
"logps/rejected": -3.8970909118652344,
|
| 1415 |
+
"loss": 1.2244,
|
| 1416 |
+
"rewards/accuracies": 0.8125,
|
| 1417 |
+
"rewards/chosen": -5.694063663482666,
|
| 1418 |
+
"rewards/margins": 2.1001174449920654,
|
| 1419 |
+
"rewards/rejected": -7.794181823730469,
|
| 1420 |
"step": 460
|
| 1421 |
},
|
| 1422 |
{
|
| 1423 |
"epoch": 0.9735671290238157,
|
| 1424 |
+
"grad_norm": 55.2538190833758,
|
| 1425 |
"learning_rate": 9.64668657069706e-10,
|
| 1426 |
+
"logits/chosen": -1.8154325485229492,
|
| 1427 |
+
"logits/rejected": -1.8061832189559937,
|
| 1428 |
+
"logps/chosen": -2.8277297019958496,
|
| 1429 |
+
"logps/rejected": -3.931652069091797,
|
| 1430 |
+
"loss": 1.1408,
|
| 1431 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1432 |
+
"rewards/chosen": -5.655459403991699,
|
| 1433 |
+
"rewards/margins": 2.2078452110290527,
|
| 1434 |
+
"rewards/rejected": -7.863304138183594,
|
| 1435 |
"step": 465
|
| 1436 |
},
|
| 1437 |
{
|
| 1438 |
"epoch": 0.98403559277676,
|
| 1439 |
+
"grad_norm": 63.55697491973192,
|
| 1440 |
"learning_rate": 3.2839470889836627e-10,
|
| 1441 |
+
"logits/chosen": -1.762578010559082,
|
| 1442 |
+
"logits/rejected": -1.7515432834625244,
|
| 1443 |
+
"logps/chosen": -2.639904022216797,
|
| 1444 |
+
"logps/rejected": -3.7671687602996826,
|
| 1445 |
+
"loss": 1.1961,
|
| 1446 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1447 |
+
"rewards/chosen": -5.279808044433594,
|
| 1448 |
+
"rewards/margins": 2.2545294761657715,
|
| 1449 |
+
"rewards/rejected": -7.534337520599365,
|
| 1450 |
"step": 470
|
| 1451 |
},
|
| 1452 |
{
|
| 1453 |
"epoch": 0.9945040565297043,
|
| 1454 |
+
"grad_norm": 96.99566903301525,
|
| 1455 |
"learning_rate": 2.6813123097352287e-11,
|
| 1456 |
+
"logits/chosen": -1.8280794620513916,
|
| 1457 |
+
"logits/rejected": -1.8603473901748657,
|
| 1458 |
+
"logps/chosen": -2.8405585289001465,
|
| 1459 |
+
"logps/rejected": -3.6063880920410156,
|
| 1460 |
+
"loss": 1.1728,
|
| 1461 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1462 |
+
"rewards/chosen": -5.681117057800293,
|
| 1463 |
+
"rewards/margins": 1.5316593647003174,
|
| 1464 |
+
"rewards/rejected": -7.212776184082031,
|
| 1465 |
"step": 475
|
| 1466 |
},
|
| 1467 |
{
|
| 1468 |
"epoch": 0.998691442030882,
|
| 1469 |
"step": 477,
|
| 1470 |
"total_flos": 0.0,
|
| 1471 |
+
"train_loss": 1.3686470238167785,
|
| 1472 |
+
"train_runtime": 22217.0339,
|
| 1473 |
+
"train_samples_per_second": 2.752,
|
| 1474 |
+
"train_steps_per_second": 0.021
|
| 1475 |
}
|
| 1476 |
],
|
| 1477 |
"logging_steps": 5,
|