Model save
Browse files- README.md +12 -14
- all_results.json +5 -18
- generation_config.json +3 -3
- train_results.json +5 -5
- trainer_state.json +539 -1275
README.md
CHANGED
|
@@ -3,7 +3,6 @@ library_name: transformers
|
|
| 3 |
tags:
|
| 4 |
- trl
|
| 5 |
- dpo
|
| 6 |
-
- alignment-handbook
|
| 7 |
- generated_from_trainer
|
| 8 |
model-index:
|
| 9 |
- name: zephyr-7b-dpo-full
|
|
@@ -17,15 +16,15 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 17 |
|
| 18 |
This model was trained from scratch on an unknown dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
-
-
|
| 21 |
-
-
|
| 22 |
-
-
|
| 23 |
-
-
|
| 24 |
-
-
|
| 25 |
-
-
|
| 26 |
-
-
|
| 27 |
-
-
|
| 28 |
-
-
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
|
@@ -60,10 +59,9 @@ The following hyperparameters were used during training:
|
|
| 60 |
|
| 61 |
### Training results
|
| 62 |
|
| 63 |
-
| Training Loss | Epoch | Step |
|
| 64 |
-
|
| 65 |
-
| 0.
|
| 66 |
-
| 0.3187 | 1.9969 | 954 | -0.5950 | -0.4472 | -426.8990 | -545.3427 | 0.5225 | 0.7734 | -1.5476 | 1.1972 | -2.7448 |
|
| 67 |
|
| 68 |
|
| 69 |
### Framework versions
|
|
|
|
| 3 |
tags:
|
| 4 |
- trl
|
| 5 |
- dpo
|
|
|
|
| 6 |
- generated_from_trainer
|
| 7 |
model-index:
|
| 8 |
- name: zephyr-7b-dpo-full
|
|
|
|
| 16 |
|
| 17 |
This model was trained from scratch on an unknown dataset.
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.5997
|
| 20 |
+
- Rewards/chosen: -0.2062
|
| 21 |
+
- Rewards/rejected: -0.5288
|
| 22 |
+
- Rewards/accuracies: 0.7070
|
| 23 |
+
- Rewards/margins: 0.3226
|
| 24 |
+
- Logps/rejected: -332.8396
|
| 25 |
+
- Logps/chosen: -301.5587
|
| 26 |
+
- Logits/rejected: -1.1529
|
| 27 |
+
- Logits/chosen: -1.2491
|
| 28 |
|
| 29 |
## Model description
|
| 30 |
|
|
|
|
| 59 |
|
| 60 |
### Training results
|
| 61 |
|
| 62 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 63 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 64 |
+
| 0.5873 | 0.9984 | 477 | 0.5997 | -0.2062 | -0.5288 | 0.7070 | 0.3226 | -332.8396 | -301.5587 | -1.1529 | -1.2491 |
|
|
|
|
| 65 |
|
| 66 |
|
| 67 |
### Framework versions
|
all_results.json
CHANGED
|
@@ -1,22 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_logits/chosen": -0.5949550271034241,
|
| 4 |
-
"eval_logits/rejected": -0.44720327854156494,
|
| 5 |
-
"eval_logps/chosen": -426.89898681640625,
|
| 6 |
-
"eval_logps/rejected": -545.3427124023438,
|
| 7 |
-
"eval_loss": 1.1966235637664795,
|
| 8 |
-
"eval_rewards/accuracies": 0.7734375,
|
| 9 |
-
"eval_rewards/chosen": -7.737778663635254,
|
| 10 |
-
"eval_rewards/margins": 5.986159324645996,
|
| 11 |
-
"eval_rewards/rejected": -13.723937034606934,
|
| 12 |
-
"eval_runtime": 182.4829,
|
| 13 |
-
"eval_samples": 2000,
|
| 14 |
-
"eval_samples_per_second": 10.96,
|
| 15 |
-
"eval_steps_per_second": 0.175,
|
| 16 |
"total_flos": 0.0,
|
| 17 |
-
"train_loss": 0.
|
| 18 |
-
"train_runtime":
|
| 19 |
"train_samples": 61134,
|
| 20 |
-
"train_samples_per_second":
|
| 21 |
-
"train_steps_per_second":
|
| 22 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9984301412872841,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6286907725863986,
|
| 5 |
+
"train_runtime": 7498.289,
|
| 6 |
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 8.153,
|
| 8 |
+
"train_steps_per_second": 0.064
|
| 9 |
}
|
generation_config.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
"transformers_version": "4.45.2"
|
| 6 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"eos_token_id": 151643,
|
| 4 |
+
"max_new_tokens": 2048,
|
| 5 |
"transformers_version": "4.45.2"
|
| 6 |
}
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 61134,
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9984301412872841,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6286907725863986,
|
| 5 |
+
"train_runtime": 7498.289,
|
| 6 |
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 8.153,
|
| 8 |
+
"train_steps_per_second": 0.064
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,21 +1,21 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0020931449502878076,
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"logits/chosen": -
|
| 16 |
-
"logits/rejected": -
|
| 17 |
-
"logps/chosen": -
|
| 18 |
-
"logps/rejected": -
|
| 19 |
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
|
@@ -25,1469 +25,733 @@
|
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.020931449502878074,
|
| 28 |
-
"grad_norm":
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"logits/chosen": -
|
| 31 |
-
"logits/rejected": -
|
| 32 |
-
"logps/chosen": -
|
| 33 |
-
"logps/rejected": -
|
| 34 |
-
"loss": 0.
|
| 35 |
-
"rewards/accuracies": 0.
|
| 36 |
-
"rewards/chosen":
|
| 37 |
-
"rewards/margins":
|
| 38 |
-
"rewards/rejected": 0.
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.04186289900575615,
|
| 43 |
-
"grad_norm":
|
| 44 |
-
"learning_rate":
|
| 45 |
-
"logits/chosen": -
|
| 46 |
-
"logits/rejected": -
|
| 47 |
-
"logps/chosen": -
|
| 48 |
-
"logps/rejected": -
|
| 49 |
"loss": 0.6931,
|
| 50 |
-
"rewards/accuracies": 0.
|
| 51 |
-
"rewards/chosen": 0.
|
| 52 |
-
"rewards/margins":
|
| 53 |
-
"rewards/rejected": 0.
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 0.06279434850863422,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate":
|
| 60 |
-
"logits/chosen": -
|
| 61 |
-
"logits/rejected": -
|
| 62 |
-
"logps/chosen": -
|
| 63 |
-
"logps/rejected": -
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"rewards/accuracies": 0.
|
| 66 |
-
"rewards/chosen": 0.
|
| 67 |
-
"rewards/margins": 0.
|
| 68 |
-
"rewards/rejected": 0.
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.0837257980115123,
|
| 73 |
-
"grad_norm":
|
| 74 |
-
"learning_rate":
|
| 75 |
-
"logits/chosen": -
|
| 76 |
-
"logits/rejected": -
|
| 77 |
-
"logps/chosen": -
|
| 78 |
-
"logps/rejected": -266.
|
| 79 |
-
"loss": 0.
|
| 80 |
-
"rewards/accuracies": 0.
|
| 81 |
-
"rewards/chosen": 0.
|
| 82 |
-
"rewards/margins": 0.
|
| 83 |
-
"rewards/rejected": 0.
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
"epoch": 0.10465724751439037,
|
| 88 |
-
"grad_norm":
|
| 89 |
-
"learning_rate":
|
| 90 |
-
"logits/chosen": -
|
| 91 |
-
"logits/rejected": -
|
| 92 |
-
"logps/chosen": -
|
| 93 |
-
"logps/rejected": -
|
| 94 |
-
"loss": 0.
|
| 95 |
-
"rewards/accuracies": 0.
|
| 96 |
-
"rewards/chosen": 0.
|
| 97 |
-
"rewards/margins": 0.
|
| 98 |
-
"rewards/rejected": 0.
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.12558869701726844,
|
| 103 |
-
"grad_norm":
|
| 104 |
-
"learning_rate":
|
| 105 |
-
"logits/chosen": -
|
| 106 |
-
"logits/rejected": -
|
| 107 |
-
"logps/chosen": -
|
| 108 |
-
"logps/rejected": -
|
| 109 |
-
"loss": 0.
|
| 110 |
-
"rewards/accuracies": 0.
|
| 111 |
-
"rewards/chosen": 0.
|
| 112 |
-
"rewards/margins": 0.
|
| 113 |
-
"rewards/rejected": 0.
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.14652014652014653,
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"logits/chosen": -
|
| 121 |
-
"logits/rejected": -
|
| 122 |
-
"logps/chosen": -
|
| 123 |
-
"logps/rejected": -
|
| 124 |
-
"loss": 0.
|
| 125 |
-
"rewards/accuracies": 0.
|
| 126 |
-
"rewards/chosen": 0.
|
| 127 |
-
"rewards/margins": 0.
|
| 128 |
-
"rewards/rejected": 0.
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.1674515960230246,
|
| 133 |
-
"grad_norm":
|
| 134 |
-
"learning_rate": 4.
|
| 135 |
-
"logits/chosen": -
|
| 136 |
-
"logits/rejected": -
|
| 137 |
-
"logps/chosen": -
|
| 138 |
-
"logps/rejected": -
|
| 139 |
-
"loss": 0.
|
| 140 |
-
"rewards/accuracies": 0.
|
| 141 |
-
"rewards/chosen": 0.
|
| 142 |
-
"rewards/margins": 0.
|
| 143 |
-
"rewards/rejected":
|
| 144 |
"step": 80
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.18838304552590268,
|
| 148 |
-
"grad_norm":
|
| 149 |
-
"learning_rate": 4.
|
| 150 |
-
"logits/chosen": -
|
| 151 |
-
"logits/rejected": -
|
| 152 |
-
"logps/chosen": -
|
| 153 |
-
"logps/rejected": -
|
| 154 |
-
"loss": 0.
|
| 155 |
-
"rewards/accuracies": 0.
|
| 156 |
-
"rewards/chosen":
|
| 157 |
-
"rewards/margins": 0.
|
| 158 |
-
"rewards/rejected": -0.
|
| 159 |
"step": 90
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.20931449502878074,
|
| 163 |
-
"grad_norm":
|
| 164 |
-
"learning_rate": 4.
|
| 165 |
-
"logits/chosen": -
|
| 166 |
-
"logits/rejected": -
|
| 167 |
-
"logps/chosen": -
|
| 168 |
-
"logps/rejected": -
|
| 169 |
-
"loss": 0.
|
| 170 |
-
"rewards/accuracies": 0.
|
| 171 |
-
"rewards/chosen":
|
| 172 |
-
"rewards/margins": 0.
|
| 173 |
-
"rewards/rejected": -0.
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.2302459445316588,
|
| 178 |
-
"grad_norm":
|
| 179 |
-
"learning_rate": 4.
|
| 180 |
-
"logits/chosen": -
|
| 181 |
-
"logits/rejected": -
|
| 182 |
-
"logps/chosen": -
|
| 183 |
-
"logps/rejected": -
|
| 184 |
-
"loss": 0.
|
| 185 |
-
"rewards/accuracies": 0.
|
| 186 |
-
"rewards/chosen": -0.
|
| 187 |
-
"rewards/margins": 0.
|
| 188 |
-
"rewards/rejected": -0.
|
| 189 |
"step": 110
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.25117739403453687,
|
| 193 |
-
"grad_norm":
|
| 194 |
-
"learning_rate": 4.
|
| 195 |
-
"logits/chosen": -
|
| 196 |
-
"logits/rejected": -
|
| 197 |
-
"logps/chosen": -
|
| 198 |
-
"logps/rejected": -
|
| 199 |
-
"loss": 0.
|
| 200 |
-
"rewards/accuracies": 0.
|
| 201 |
-
"rewards/chosen": -0.
|
| 202 |
-
"rewards/margins": 0.
|
| 203 |
-
"rewards/rejected": -0.
|
| 204 |
"step": 120
|
| 205 |
},
|
| 206 |
{
|
| 207 |
"epoch": 0.272108843537415,
|
| 208 |
-
"grad_norm":
|
| 209 |
-
"learning_rate": 4.
|
| 210 |
-
"logits/chosen": -
|
| 211 |
-
"logits/rejected": -
|
| 212 |
-
"logps/chosen": -
|
| 213 |
-
"logps/rejected": -
|
| 214 |
-
"loss": 0.
|
| 215 |
-
"rewards/accuracies": 0.
|
| 216 |
-
"rewards/chosen": -0.
|
| 217 |
-
"rewards/margins": 0.
|
| 218 |
-
"rewards/rejected": -0.
|
| 219 |
"step": 130
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.29304029304029305,
|
| 223 |
-
"grad_norm":
|
| 224 |
-
"learning_rate": 4.
|
| 225 |
-
"logits/chosen": -
|
| 226 |
-
"logits/rejected": -
|
| 227 |
-
"logps/chosen": -339.
|
| 228 |
-
"logps/rejected": -
|
| 229 |
-
"loss": 0.
|
| 230 |
-
"rewards/accuracies": 0.
|
| 231 |
-
"rewards/chosen": -0.
|
| 232 |
-
"rewards/margins": 0.
|
| 233 |
-
"rewards/rejected": -0.
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.3139717425431711,
|
| 238 |
-
"grad_norm":
|
| 239 |
-
"learning_rate": 4.
|
| 240 |
-
"logits/chosen": -
|
| 241 |
-
"logits/rejected": -
|
| 242 |
-
"logps/chosen": -
|
| 243 |
-
"logps/rejected": -
|
| 244 |
-
"loss": 0.
|
| 245 |
-
"rewards/accuracies": 0.
|
| 246 |
-
"rewards/chosen": -0.
|
| 247 |
-
"rewards/margins": 0.
|
| 248 |
-
"rewards/rejected": -0.
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.3349031920460492,
|
| 253 |
-
"grad_norm":
|
| 254 |
-
"learning_rate": 4.
|
| 255 |
-
"logits/chosen": -
|
| 256 |
-
"logits/rejected": -
|
| 257 |
-
"logps/chosen": -
|
| 258 |
-
"logps/rejected": -
|
| 259 |
-
"loss": 0.
|
| 260 |
-
"rewards/accuracies": 0.
|
| 261 |
-
"rewards/chosen": -0.
|
| 262 |
-
"rewards/margins": 0.
|
| 263 |
-
"rewards/rejected": -0.
|
| 264 |
"step": 160
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 0.35583464154892724,
|
| 268 |
-
"grad_norm":
|
| 269 |
-
"learning_rate": 4.
|
| 270 |
-
"logits/chosen": -
|
| 271 |
-
"logits/rejected": -
|
| 272 |
-
"logps/chosen": -
|
| 273 |
-
"logps/rejected": -
|
| 274 |
-
"loss": 0.
|
| 275 |
-
"rewards/accuracies": 0.
|
| 276 |
-
"rewards/chosen": -0.
|
| 277 |
-
"rewards/margins": 0.
|
| 278 |
-
"rewards/rejected": -0.
|
| 279 |
"step": 170
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.37676609105180536,
|
| 283 |
-
"grad_norm":
|
| 284 |
-
"learning_rate":
|
| 285 |
-
"logits/chosen": -
|
| 286 |
-
"logits/rejected": -
|
| 287 |
-
"logps/chosen": -
|
| 288 |
-
"logps/rejected": -
|
| 289 |
-
"loss": 0.
|
| 290 |
-
"rewards/accuracies": 0.
|
| 291 |
-
"rewards/chosen": -0.
|
| 292 |
-
"rewards/margins": 0.
|
| 293 |
-
"rewards/rejected": -0.
|
| 294 |
"step": 180
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 0.3976975405546834,
|
| 298 |
-
"grad_norm":
|
| 299 |
-
"learning_rate":
|
| 300 |
-
"logits/chosen": -
|
| 301 |
-
"logits/rejected": -
|
| 302 |
-
"logps/chosen": -
|
| 303 |
-
"logps/rejected": -
|
| 304 |
-
"loss": 0.
|
| 305 |
-
"rewards/accuracies": 0.
|
| 306 |
-
"rewards/chosen": -0.
|
| 307 |
-
"rewards/margins": 0.
|
| 308 |
-
"rewards/rejected": -0.
|
| 309 |
"step": 190
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.4186289900575615,
|
| 313 |
-
"grad_norm":
|
| 314 |
-
"learning_rate":
|
| 315 |
-
"logits/chosen": -
|
| 316 |
-
"logits/rejected": -
|
| 317 |
-
"logps/chosen": -
|
| 318 |
-
"logps/rejected": -
|
| 319 |
-
"loss": 0.
|
| 320 |
-
"rewards/accuracies": 0.
|
| 321 |
-
"rewards/chosen": -0.
|
| 322 |
-
"rewards/margins": 0.
|
| 323 |
-
"rewards/rejected": -0.
|
| 324 |
"step": 200
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 0.43956043956043955,
|
| 328 |
-
"grad_norm":
|
| 329 |
-
"learning_rate":
|
| 330 |
-
"logits/chosen": -
|
| 331 |
-
"logits/rejected": -
|
| 332 |
-
"logps/chosen": -
|
| 333 |
-
"logps/rejected": -
|
| 334 |
-
"loss": 0.
|
| 335 |
-
"rewards/accuracies": 0.
|
| 336 |
-
"rewards/chosen": -0.
|
| 337 |
-
"rewards/margins": 0.
|
| 338 |
-
"rewards/rejected": -0.
|
| 339 |
"step": 210
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.4604918890633176,
|
| 343 |
-
"grad_norm":
|
| 344 |
-
"learning_rate":
|
| 345 |
-
"logits/chosen": -
|
| 346 |
-
"logits/rejected": -
|
| 347 |
-
"logps/chosen": -
|
| 348 |
-
"logps/rejected": -
|
| 349 |
-
"loss": 0.
|
| 350 |
-
"rewards/accuracies": 0.
|
| 351 |
-
"rewards/chosen": -0.
|
| 352 |
-
"rewards/margins": 0.
|
| 353 |
-
"rewards/rejected": -0.
|
| 354 |
"step": 220
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.48142333856619574,
|
| 358 |
-
"grad_norm":
|
| 359 |
-
"learning_rate":
|
| 360 |
-
"logits/chosen": -
|
| 361 |
-
"logits/rejected": -
|
| 362 |
-
"logps/chosen": -
|
| 363 |
-
"logps/rejected": -
|
| 364 |
-
"loss": 0.
|
| 365 |
-
"rewards/accuracies": 0.
|
| 366 |
-
"rewards/chosen": -0.
|
| 367 |
-
"rewards/margins": 0.
|
| 368 |
-
"rewards/rejected": -
|
| 369 |
"step": 230
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.5023547880690737,
|
| 373 |
-
"grad_norm":
|
| 374 |
-
"learning_rate":
|
| 375 |
-
"logits/chosen": -
|
| 376 |
-
"logits/rejected": -
|
| 377 |
-
"logps/chosen": -
|
| 378 |
-
"logps/rejected": -
|
| 379 |
-
"loss": 0.
|
| 380 |
"rewards/accuracies": 0.699999988079071,
|
| 381 |
-
"rewards/chosen": -0.
|
| 382 |
-
"rewards/margins": 0.
|
| 383 |
-
"rewards/rejected": -0.
|
| 384 |
"step": 240
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 0.5232862375719518,
|
| 388 |
-
"grad_norm":
|
| 389 |
-
"learning_rate":
|
| 390 |
-
"logits/chosen": -
|
| 391 |
-
"logits/rejected": -
|
| 392 |
-
"logps/chosen": -
|
| 393 |
-
"logps/rejected": -
|
| 394 |
-
"loss": 0.
|
| 395 |
-
"rewards/accuracies": 0.
|
| 396 |
-
"rewards/chosen": -0.
|
| 397 |
-
"rewards/margins": 0.
|
| 398 |
-
"rewards/rejected": -
|
| 399 |
"step": 250
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.54421768707483,
|
| 403 |
-
"grad_norm":
|
| 404 |
-
"learning_rate":
|
| 405 |
-
"logits/chosen": -
|
| 406 |
-
"logits/rejected": -
|
| 407 |
-
"logps/chosen": -
|
| 408 |
-
"logps/rejected": -
|
| 409 |
-
"loss": 0.
|
| 410 |
-
"rewards/accuracies": 0.
|
| 411 |
-
"rewards/chosen": -0.
|
| 412 |
-
"rewards/margins": 0.
|
| 413 |
-
"rewards/rejected": -
|
| 414 |
"step": 260
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 0.565149136577708,
|
| 418 |
-
"grad_norm":
|
| 419 |
-
"learning_rate":
|
| 420 |
-
"logits/chosen": -
|
| 421 |
-
"logits/rejected": -
|
| 422 |
-
"logps/chosen": -
|
| 423 |
-
"logps/rejected": -
|
| 424 |
-
"loss": 0.
|
| 425 |
-
"rewards/accuracies": 0.
|
| 426 |
-
"rewards/chosen": -0.
|
| 427 |
-
"rewards/margins": 0.
|
| 428 |
-
"rewards/rejected": -
|
| 429 |
"step": 270
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.5860805860805861,
|
| 433 |
-
"grad_norm":
|
| 434 |
-
"learning_rate":
|
| 435 |
-
"logits/chosen": -
|
| 436 |
-
"logits/rejected": -1.
|
| 437 |
-
"logps/chosen": -
|
| 438 |
-
"logps/rejected": -
|
| 439 |
-
"loss": 0.
|
| 440 |
"rewards/accuracies": 0.6625000238418579,
|
| 441 |
-
"rewards/chosen": -0.
|
| 442 |
-
"rewards/margins": 0.
|
| 443 |
-
"rewards/rejected": -
|
| 444 |
"step": 280
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.6070120355834642,
|
| 448 |
-
"grad_norm":
|
| 449 |
-
"learning_rate":
|
| 450 |
-
"logits/chosen": -
|
| 451 |
-
"logits/rejected": -1.
|
| 452 |
-
"logps/chosen": -
|
| 453 |
-
"logps/rejected": -
|
| 454 |
-
"loss": 0.
|
| 455 |
-
"rewards/accuracies": 0.
|
| 456 |
-
"rewards/chosen": -0.
|
| 457 |
-
"rewards/margins": 0.
|
| 458 |
-
"rewards/rejected": -
|
| 459 |
"step": 290
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.6279434850863422,
|
| 463 |
-
"grad_norm":
|
| 464 |
-
"learning_rate":
|
| 465 |
-
"logits/chosen": -
|
| 466 |
-
"logits/rejected": -1.
|
| 467 |
-
"logps/chosen": -
|
| 468 |
-
"logps/rejected": -
|
| 469 |
-
"loss": 0.
|
| 470 |
-
"rewards/accuracies": 0.
|
| 471 |
-
"rewards/chosen": -0.
|
| 472 |
-
"rewards/margins": 0.
|
| 473 |
-
"rewards/rejected": -
|
| 474 |
"step": 300
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 0.6488749345892203,
|
| 478 |
-
"grad_norm":
|
| 479 |
-
"learning_rate":
|
| 480 |
-
"logits/chosen": -
|
| 481 |
-
"logits/rejected": -1.
|
| 482 |
-
"logps/chosen": -
|
| 483 |
-
"logps/rejected": -
|
| 484 |
-
"loss": 0.
|
| 485 |
-
"rewards/accuracies": 0.
|
| 486 |
-
"rewards/chosen": -0.
|
| 487 |
-
"rewards/margins": 0.
|
| 488 |
-
"rewards/rejected": -
|
| 489 |
"step": 310
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.6698063840920984,
|
| 493 |
-
"grad_norm":
|
| 494 |
-
"learning_rate":
|
| 495 |
-
"logits/chosen": -1.
|
| 496 |
-
"logits/rejected": -1.
|
| 497 |
-
"logps/chosen": -
|
| 498 |
-
"logps/rejected": -
|
| 499 |
-
"loss": 0.
|
| 500 |
-
"rewards/accuracies": 0.
|
| 501 |
-
"rewards/chosen": -0.
|
| 502 |
-
"rewards/margins": 0.
|
| 503 |
-
"rewards/rejected": -
|
| 504 |
"step": 320
|
| 505 |
},
|
| 506 |
{
|
| 507 |
"epoch": 0.6907378335949764,
|
| 508 |
-
"grad_norm":
|
| 509 |
-
"learning_rate":
|
| 510 |
-
"logits/chosen": -
|
| 511 |
-
"logits/rejected": -
|
| 512 |
-
"logps/chosen": -
|
| 513 |
-
"logps/rejected": -
|
| 514 |
-
"loss": 0.
|
| 515 |
-
"rewards/accuracies": 0.
|
| 516 |
-
"rewards/chosen": -0.
|
| 517 |
-
"rewards/margins": 0.
|
| 518 |
-
"rewards/rejected": -
|
| 519 |
"step": 330
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.7116692830978545,
|
| 523 |
-
"grad_norm":
|
| 524 |
-
"learning_rate":
|
| 525 |
-
"logits/chosen": -1.
|
| 526 |
-
"logits/rejected": -1.
|
| 527 |
-
"logps/chosen": -
|
| 528 |
-
"logps/rejected": -
|
| 529 |
-
"loss": 0.
|
| 530 |
-
"rewards/accuracies": 0.
|
| 531 |
-
"rewards/chosen": -0.
|
| 532 |
-
"rewards/margins": 0.
|
| 533 |
-
"rewards/rejected": -
|
| 534 |
"step": 340
|
| 535 |
},
|
| 536 |
{
|
| 537 |
"epoch": 0.7326007326007326,
|
| 538 |
-
"grad_norm":
|
| 539 |
-
"learning_rate":
|
| 540 |
-
"logits/chosen": -1.
|
| 541 |
-
"logits/rejected": -1.
|
| 542 |
-
"logps/chosen": -
|
| 543 |
-
"logps/rejected": -
|
| 544 |
-
"loss": 0.
|
| 545 |
-
"rewards/accuracies": 0.
|
| 546 |
-
"rewards/chosen": -0.
|
| 547 |
-
"rewards/margins": 0.
|
| 548 |
-
"rewards/rejected": -
|
| 549 |
"step": 350
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.7535321821036107,
|
| 553 |
-
"grad_norm":
|
| 554 |
-
"learning_rate":
|
| 555 |
-
"logits/chosen": -1.
|
| 556 |
-
"logits/rejected": -1.
|
| 557 |
-
"logps/chosen": -
|
| 558 |
-
"logps/rejected": -
|
| 559 |
-
"loss": 0.
|
| 560 |
-
"rewards/accuracies": 0.
|
| 561 |
-
"rewards/chosen": -0.
|
| 562 |
-
"rewards/margins": 0.
|
| 563 |
-
"rewards/rejected": -
|
| 564 |
"step": 360
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.7744636316064888,
|
| 568 |
-
"grad_norm":
|
| 569 |
-
"learning_rate":
|
| 570 |
-
"logits/chosen": -1.
|
| 571 |
-
"logits/rejected": -1.
|
| 572 |
-
"logps/chosen": -
|
| 573 |
-
"logps/rejected": -
|
| 574 |
-
"loss": 0.
|
| 575 |
-
"rewards/accuracies": 0.
|
| 576 |
-
"rewards/chosen": -0.
|
| 577 |
-
"rewards/margins": 0.
|
| 578 |
-
"rewards/rejected": -
|
| 579 |
"step": 370
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.7953950811093669,
|
| 583 |
-
"grad_norm":
|
| 584 |
-
"learning_rate":
|
| 585 |
-
"logits/chosen": -1.
|
| 586 |
-
"logits/rejected": -1.
|
| 587 |
-
"logps/chosen": -
|
| 588 |
-
"logps/rejected": -
|
| 589 |
-
"loss": 0.
|
| 590 |
-
"rewards/accuracies": 0.
|
| 591 |
-
"rewards/chosen": -0.
|
| 592 |
-
"rewards/margins": 0.
|
| 593 |
-
"rewards/rejected": -
|
| 594 |
"step": 380
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.8163265306122449,
|
| 598 |
-
"grad_norm":
|
| 599 |
-
"learning_rate":
|
| 600 |
-
"logits/chosen": -1.
|
| 601 |
-
"logits/rejected": -1.
|
| 602 |
-
"logps/chosen": -
|
| 603 |
-
"logps/rejected": -
|
| 604 |
-
"loss": 0.
|
| 605 |
"rewards/accuracies": 0.731249988079071,
|
| 606 |
-
"rewards/chosen": -0.
|
| 607 |
-
"rewards/margins": 0.
|
| 608 |
-
"rewards/rejected": -
|
| 609 |
"step": 390
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.837257980115123,
|
| 613 |
-
"grad_norm":
|
| 614 |
-
"learning_rate": 3.
|
| 615 |
-
"logits/chosen": -1.
|
| 616 |
-
"logits/rejected": -1.
|
| 617 |
-
"logps/chosen": -
|
| 618 |
-
"logps/rejected": -
|
| 619 |
-
"loss": 0.
|
| 620 |
-
"rewards/accuracies": 0.
|
| 621 |
-
"rewards/chosen": -
|
| 622 |
-
"rewards/margins": 0.
|
| 623 |
-
"rewards/rejected": -
|
| 624 |
"step": 400
|
| 625 |
},
|
| 626 |
{
|
| 627 |
"epoch": 0.858189429618001,
|
| 628 |
-
"grad_norm":
|
| 629 |
-
"learning_rate":
|
| 630 |
-
"logits/chosen": -
|
| 631 |
-
"logits/rejected": -
|
| 632 |
-
"logps/chosen": -
|
| 633 |
-
"logps/rejected": -
|
| 634 |
-
"loss": 0.
|
| 635 |
-
"rewards/accuracies": 0.
|
| 636 |
-
"rewards/chosen": -0.
|
| 637 |
-
"rewards/margins": 0.
|
| 638 |
-
"rewards/rejected": -
|
| 639 |
"step": 410
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.8791208791208791,
|
| 643 |
-
"grad_norm":
|
| 644 |
-
"learning_rate":
|
| 645 |
-
"logits/chosen": -
|
| 646 |
-
"logits/rejected": -1.
|
| 647 |
-
"logps/chosen": -
|
| 648 |
-
"logps/rejected": -
|
| 649 |
-
"loss": 0.
|
| 650 |
-
"rewards/accuracies": 0.
|
| 651 |
-
"rewards/chosen": -0.
|
| 652 |
-
"rewards/margins": 0.
|
| 653 |
-
"rewards/rejected": -
|
| 654 |
"step": 420
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.9000523286237572,
|
| 658 |
-
"grad_norm":
|
| 659 |
-
"learning_rate":
|
| 660 |
-
"logits/chosen": -1.
|
| 661 |
-
"logits/rejected": -1.
|
| 662 |
-
"logps/chosen": -
|
| 663 |
-
"logps/rejected": -
|
| 664 |
-
"loss": 0.
|
| 665 |
-
"rewards/accuracies": 0.
|
| 666 |
-
"rewards/chosen": -0.
|
| 667 |
-
"rewards/margins": 0.
|
| 668 |
-
"rewards/rejected": -
|
| 669 |
"step": 430
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.9209837781266352,
|
| 673 |
-
"grad_norm":
|
| 674 |
-
"learning_rate":
|
| 675 |
-
"logits/chosen": -1.
|
| 676 |
-
"logits/rejected": -1.
|
| 677 |
-
"logps/chosen": -
|
| 678 |
-
"logps/rejected": -
|
| 679 |
-
"loss": 0.
|
| 680 |
-
"rewards/accuracies": 0.
|
| 681 |
-
"rewards/chosen": -
|
| 682 |
-
"rewards/margins": 0.
|
| 683 |
-
"rewards/rejected": -
|
| 684 |
"step": 440
|
| 685 |
},
|
| 686 |
{
|
| 687 |
"epoch": 0.9419152276295133,
|
| 688 |
-
"grad_norm":
|
| 689 |
-
"learning_rate":
|
| 690 |
-
"logits/chosen": -1.
|
| 691 |
-
"logits/rejected": -1.
|
| 692 |
-
"logps/chosen": -
|
| 693 |
-
"logps/rejected": -
|
| 694 |
-
"loss": 0.
|
| 695 |
-
"rewards/accuracies": 0.
|
| 696 |
-
"rewards/chosen": -0.
|
| 697 |
-
"rewards/margins": 0.
|
| 698 |
-
"rewards/rejected": -
|
| 699 |
"step": 450
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.9628466771323915,
|
| 703 |
-
"grad_norm":
|
| 704 |
-
"learning_rate":
|
| 705 |
-
"logits/chosen": -1.
|
| 706 |
-
"logits/rejected": -1.
|
| 707 |
-
"logps/chosen": -
|
| 708 |
-
"logps/rejected": -
|
| 709 |
-
"loss": 0.
|
| 710 |
-
"rewards/accuracies": 0.
|
| 711 |
-
"rewards/chosen": -0.
|
| 712 |
-
"rewards/margins": 0.
|
| 713 |
-
"rewards/rejected": -
|
| 714 |
"step": 460
|
| 715 |
},
|
| 716 |
{
|
| 717 |
"epoch": 0.9837781266352695,
|
| 718 |
-
"grad_norm":
|
| 719 |
-
"learning_rate": 3.
|
| 720 |
-
"logits/chosen": -1.
|
| 721 |
-
"logits/rejected": -1.
|
| 722 |
-
"logps/chosen": -
|
| 723 |
-
"logps/rejected": -
|
| 724 |
-
"loss": 0.
|
| 725 |
-
"rewards/accuracies": 0.
|
| 726 |
-
"rewards/chosen": -0.
|
| 727 |
-
"rewards/margins": 0.
|
| 728 |
-
"rewards/rejected": -
|
| 729 |
"step": 470
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.9984301412872841,
|
| 733 |
-
"eval_logits/chosen": -1.
|
| 734 |
-
"eval_logits/rejected": -1.
|
| 735 |
-
"eval_logps/chosen": -
|
| 736 |
-
"eval_logps/rejected": -
|
| 737 |
-
"eval_loss": 0.
|
| 738 |
-
"eval_rewards/accuracies": 0.
|
| 739 |
-
"eval_rewards/chosen": -0.
|
| 740 |
-
"eval_rewards/margins": 0.
|
| 741 |
-
"eval_rewards/rejected": -
|
| 742 |
-
"eval_runtime":
|
| 743 |
-
"eval_samples_per_second":
|
| 744 |
-
"eval_steps_per_second": 0.
|
| 745 |
"step": 477
|
| 746 |
},
|
| 747 |
{
|
| 748 |
-
"epoch":
|
| 749 |
-
"
|
| 750 |
-
"learning_rate": 2.910060778827554e-07,
|
| 751 |
-
"logits/chosen": -1.3738365173339844,
|
| 752 |
-
"logits/rejected": -1.2765980958938599,
|
| 753 |
-
"logps/chosen": -379.2554931640625,
|
| 754 |
-
"logps/rejected": -449.7948303222656,
|
| 755 |
-
"loss": 0.4798,
|
| 756 |
-
"rewards/accuracies": 0.7875000238418579,
|
| 757 |
-
"rewards/chosen": -0.723404586315155,
|
| 758 |
-
"rewards/margins": 0.8685447573661804,
|
| 759 |
-
"rewards/rejected": -1.591949224472046,
|
| 760 |
-
"step": 480
|
| 761 |
-
},
|
| 762 |
-
{
|
| 763 |
-
"epoch": 1.0256410256410255,
|
| 764 |
-
"grad_norm": 16.51017448411662,
|
| 765 |
-
"learning_rate": 2.8195076242990116e-07,
|
| 766 |
-
"logits/chosen": -1.2624908685684204,
|
| 767 |
-
"logits/rejected": -1.0200715065002441,
|
| 768 |
-
"logps/chosen": -391.1546325683594,
|
| 769 |
-
"logps/rejected": -462.6971130371094,
|
| 770 |
-
"loss": 0.3721,
|
| 771 |
-
"rewards/accuracies": 0.8374999761581421,
|
| 772 |
-
"rewards/chosen": -0.8392894864082336,
|
| 773 |
-
"rewards/margins": 1.1826223134994507,
|
| 774 |
-
"rewards/rejected": -2.021911859512329,
|
| 775 |
-
"step": 490
|
| 776 |
-
},
|
| 777 |
-
{
|
| 778 |
-
"epoch": 1.0465724751439036,
|
| 779 |
-
"grad_norm": 28.692357308634083,
|
| 780 |
-
"learning_rate": 2.7285261601056697e-07,
|
| 781 |
-
"logits/chosen": -1.04556405544281,
|
| 782 |
-
"logits/rejected": -0.8983597755432129,
|
| 783 |
-
"logps/chosen": -385.15594482421875,
|
| 784 |
-
"logps/rejected": -492.3553771972656,
|
| 785 |
-
"loss": 0.3623,
|
| 786 |
-
"rewards/accuracies": 0.862500011920929,
|
| 787 |
-
"rewards/chosen": -1.0549384355545044,
|
| 788 |
-
"rewards/margins": 1.216042399406433,
|
| 789 |
-
"rewards/rejected": -2.2709805965423584,
|
| 790 |
-
"step": 500
|
| 791 |
-
},
|
| 792 |
-
{
|
| 793 |
-
"epoch": 1.0675039246467817,
|
| 794 |
-
"grad_norm": 21.922390152982405,
|
| 795 |
-
"learning_rate": 2.6372383496608186e-07,
|
| 796 |
-
"logits/chosen": -0.9019845128059387,
|
| 797 |
-
"logits/rejected": -0.6276781558990479,
|
| 798 |
-
"logps/chosen": -385.09625244140625,
|
| 799 |
-
"logps/rejected": -487.7792053222656,
|
| 800 |
-
"loss": 0.36,
|
| 801 |
-
"rewards/accuracies": 0.893750011920929,
|
| 802 |
-
"rewards/chosen": -0.9738361239433289,
|
| 803 |
-
"rewards/margins": 1.5190062522888184,
|
| 804 |
-
"rewards/rejected": -2.492842197418213,
|
| 805 |
-
"step": 510
|
| 806 |
-
},
|
| 807 |
-
{
|
| 808 |
-
"epoch": 1.08843537414966,
|
| 809 |
-
"grad_norm": 23.056390516413284,
|
| 810 |
-
"learning_rate": 2.5457665670441937e-07,
|
| 811 |
-
"logits/chosen": -0.8516889810562134,
|
| 812 |
-
"logits/rejected": -0.5428006052970886,
|
| 813 |
-
"logps/chosen": -382.0596923828125,
|
| 814 |
-
"logps/rejected": -487.93701171875,
|
| 815 |
-
"loss": 0.3563,
|
| 816 |
-
"rewards/accuracies": 0.8687499761581421,
|
| 817 |
-
"rewards/chosen": -0.9770641326904297,
|
| 818 |
-
"rewards/margins": 1.3976478576660156,
|
| 819 |
-
"rewards/rejected": -2.3747119903564453,
|
| 820 |
-
"step": 520
|
| 821 |
-
},
|
| 822 |
-
{
|
| 823 |
-
"epoch": 1.109366823652538,
|
| 824 |
-
"grad_norm": 24.839002506991555,
|
| 825 |
-
"learning_rate": 2.454233432955807e-07,
|
| 826 |
-
"logits/chosen": -0.893288254737854,
|
| 827 |
-
"logits/rejected": -0.55360347032547,
|
| 828 |
-
"logps/chosen": -373.3586120605469,
|
| 829 |
-
"logps/rejected": -470.93048095703125,
|
| 830 |
-
"loss": 0.3595,
|
| 831 |
-
"rewards/accuracies": 0.856249988079071,
|
| 832 |
-
"rewards/chosen": -1.079138159751892,
|
| 833 |
-
"rewards/margins": 1.3878281116485596,
|
| 834 |
-
"rewards/rejected": -2.466966152191162,
|
| 835 |
-
"step": 530
|
| 836 |
-
},
|
| 837 |
-
{
|
| 838 |
-
"epoch": 1.130298273155416,
|
| 839 |
-
"grad_norm": 24.9976173018764,
|
| 840 |
-
"learning_rate": 2.3627616503391812e-07,
|
| 841 |
-
"logits/chosen": -0.6603145003318787,
|
| 842 |
-
"logits/rejected": -0.45139390230178833,
|
| 843 |
-
"logps/chosen": -381.8931884765625,
|
| 844 |
-
"logps/rejected": -511.0888671875,
|
| 845 |
-
"loss": 0.3615,
|
| 846 |
-
"rewards/accuracies": 0.862500011920929,
|
| 847 |
-
"rewards/chosen": -1.2078765630722046,
|
| 848 |
-
"rewards/margins": 1.415185809135437,
|
| 849 |
-
"rewards/rejected": -2.6230626106262207,
|
| 850 |
-
"step": 540
|
| 851 |
-
},
|
| 852 |
-
{
|
| 853 |
-
"epoch": 1.1512297226582942,
|
| 854 |
-
"grad_norm": 23.70399953245455,
|
| 855 |
-
"learning_rate": 2.2714738398943308e-07,
|
| 856 |
-
"logits/chosen": -0.7952468395233154,
|
| 857 |
-
"logits/rejected": -0.5057519674301147,
|
| 858 |
-
"logps/chosen": -405.1830139160156,
|
| 859 |
-
"logps/rejected": -512.9744873046875,
|
| 860 |
-
"loss": 0.3432,
|
| 861 |
-
"rewards/accuracies": 0.8374999761581421,
|
| 862 |
-
"rewards/chosen": -1.1684848070144653,
|
| 863 |
-
"rewards/margins": 1.4340193271636963,
|
| 864 |
-
"rewards/rejected": -2.6025044918060303,
|
| 865 |
-
"step": 550
|
| 866 |
-
},
|
| 867 |
-
{
|
| 868 |
-
"epoch": 1.1721611721611722,
|
| 869 |
-
"grad_norm": 26.9881287231271,
|
| 870 |
-
"learning_rate": 2.1804923757009882e-07,
|
| 871 |
-
"logits/chosen": -0.7067749500274658,
|
| 872 |
-
"logits/rejected": -0.4858238697052002,
|
| 873 |
-
"logps/chosen": -358.0526428222656,
|
| 874 |
-
"logps/rejected": -488.3414001464844,
|
| 875 |
-
"loss": 0.3631,
|
| 876 |
-
"rewards/accuracies": 0.925000011920929,
|
| 877 |
-
"rewards/chosen": -0.9131903648376465,
|
| 878 |
-
"rewards/margins": 1.4550573825836182,
|
| 879 |
-
"rewards/rejected": -2.3682477474212646,
|
| 880 |
-
"step": 560
|
| 881 |
-
},
|
| 882 |
-
{
|
| 883 |
-
"epoch": 1.1930926216640503,
|
| 884 |
-
"grad_norm": 26.437360887515176,
|
| 885 |
-
"learning_rate": 2.089939221172446e-07,
|
| 886 |
-
"logits/chosen": -0.6216100454330444,
|
| 887 |
-
"logits/rejected": -0.4619014263153076,
|
| 888 |
-
"logps/chosen": -427.541015625,
|
| 889 |
-
"logps/rejected": -578.9843139648438,
|
| 890 |
-
"loss": 0.3526,
|
| 891 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 892 |
-
"rewards/chosen": -1.259556770324707,
|
| 893 |
-
"rewards/margins": 1.5523223876953125,
|
| 894 |
-
"rewards/rejected": -2.8118791580200195,
|
| 895 |
-
"step": 570
|
| 896 |
-
},
|
| 897 |
-
{
|
| 898 |
-
"epoch": 1.2140240711669283,
|
| 899 |
-
"grad_norm": 24.08017206287918,
|
| 900 |
-
"learning_rate": 1.9999357655598891e-07,
|
| 901 |
-
"logits/chosen": -0.596352219581604,
|
| 902 |
-
"logits/rejected": -0.3721233308315277,
|
| 903 |
-
"logps/chosen": -412.52301025390625,
|
| 904 |
-
"logps/rejected": -562.3777465820312,
|
| 905 |
-
"loss": 0.3344,
|
| 906 |
-
"rewards/accuracies": 0.887499988079071,
|
| 907 |
-
"rewards/chosen": -1.310210943222046,
|
| 908 |
-
"rewards/margins": 1.5689070224761963,
|
| 909 |
-
"rewards/rejected": -2.879117965698242,
|
| 910 |
-
"step": 580
|
| 911 |
-
},
|
| 912 |
-
{
|
| 913 |
-
"epoch": 1.2349555206698064,
|
| 914 |
-
"grad_norm": 19.74419678906988,
|
| 915 |
-
"learning_rate": 1.9106026612264315e-07,
|
| 916 |
-
"logits/chosen": -0.8253545761108398,
|
| 917 |
-
"logits/rejected": -0.6318912506103516,
|
| 918 |
-
"logps/chosen": -399.89910888671875,
|
| 919 |
-
"logps/rejected": -515.3082275390625,
|
| 920 |
-
"loss": 0.3424,
|
| 921 |
-
"rewards/accuracies": 0.893750011920929,
|
| 922 |
-
"rewards/chosen": -1.0740442276000977,
|
| 923 |
-
"rewards/margins": 1.4926326274871826,
|
| 924 |
-
"rewards/rejected": -2.5666770935058594,
|
| 925 |
-
"step": 590
|
| 926 |
-
},
|
| 927 |
-
{
|
| 928 |
-
"epoch": 1.2558869701726845,
|
| 929 |
-
"grad_norm": 30.297500591228765,
|
| 930 |
-
"learning_rate": 1.8220596619089573e-07,
|
| 931 |
-
"logits/chosen": -0.7947873473167419,
|
| 932 |
-
"logits/rejected": -0.5936676859855652,
|
| 933 |
-
"logps/chosen": -452.333740234375,
|
| 934 |
-
"logps/rejected": -589.5086059570312,
|
| 935 |
-
"loss": 0.3349,
|
| 936 |
-
"rewards/accuracies": 0.8374999761581421,
|
| 937 |
-
"rewards/chosen": -1.3032448291778564,
|
| 938 |
-
"rewards/margins": 1.449398398399353,
|
| 939 |
-
"rewards/rejected": -2.752643346786499,
|
| 940 |
-
"step": 600
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 1.2768184196755625,
|
| 944 |
-
"grad_norm": 25.095514098513064,
|
| 945 |
-
"learning_rate": 1.7344254621846017e-07,
|
| 946 |
-
"logits/chosen": -0.5728591680526733,
|
| 947 |
-
"logits/rejected": -0.2157384604215622,
|
| 948 |
-
"logps/chosen": -446.3045959472656,
|
| 949 |
-
"logps/rejected": -558.5189819335938,
|
| 950 |
-
"loss": 0.3583,
|
| 951 |
-
"rewards/accuracies": 0.856249988079071,
|
| 952 |
-
"rewards/chosen": -1.3666396141052246,
|
| 953 |
-
"rewards/margins": 1.6065607070922852,
|
| 954 |
-
"rewards/rejected": -2.9732003211975098,
|
| 955 |
-
"step": 610
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"epoch": 1.2977498691784406,
|
| 959 |
-
"grad_norm": 26.711706547263283,
|
| 960 |
-
"learning_rate": 1.647817538357072e-07,
|
| 961 |
-
"logits/chosen": -0.6483644247055054,
|
| 962 |
-
"logits/rejected": -0.436624675989151,
|
| 963 |
-
"logps/chosen": -375.31463623046875,
|
| 964 |
-
"logps/rejected": -502.0403747558594,
|
| 965 |
-
"loss": 0.3359,
|
| 966 |
-
"rewards/accuracies": 0.824999988079071,
|
| 967 |
-
"rewards/chosen": -1.1623198986053467,
|
| 968 |
-
"rewards/margins": 1.4659651517868042,
|
| 969 |
-
"rewards/rejected": -2.6282851696014404,
|
| 970 |
-
"step": 620
|
| 971 |
-
},
|
| 972 |
-
{
|
| 973 |
-
"epoch": 1.3186813186813187,
|
| 974 |
-
"grad_norm": 29.34167044151856,
|
| 975 |
-
"learning_rate": 1.562351990976095e-07,
|
| 976 |
-
"logits/chosen": -0.5509764552116394,
|
| 977 |
-
"logits/rejected": -0.2896868884563446,
|
| 978 |
-
"logps/chosen": -404.48773193359375,
|
| 979 |
-
"logps/rejected": -547.499267578125,
|
| 980 |
-
"loss": 0.3416,
|
| 981 |
-
"rewards/accuracies": 0.887499988079071,
|
| 982 |
-
"rewards/chosen": -1.2804601192474365,
|
| 983 |
-
"rewards/margins": 1.6254936456680298,
|
| 984 |
-
"rewards/rejected": -2.905953884124756,
|
| 985 |
-
"step": 630
|
| 986 |
-
},
|
| 987 |
-
{
|
| 988 |
-
"epoch": 1.3396127681841967,
|
| 989 |
-
"grad_norm": 23.153596326850824,
|
| 990 |
-
"learning_rate": 1.478143389201113e-07,
|
| 991 |
-
"logits/chosen": -0.5090593099594116,
|
| 992 |
-
"logits/rejected": -0.1941351443529129,
|
| 993 |
-
"logps/chosen": -416.89141845703125,
|
| 994 |
-
"logps/rejected": -544.4210815429688,
|
| 995 |
-
"loss": 0.3508,
|
| 996 |
-
"rewards/accuracies": 0.8687499761581421,
|
| 997 |
-
"rewards/chosen": -1.4043817520141602,
|
| 998 |
-
"rewards/margins": 1.578391194343567,
|
| 999 |
-
"rewards/rejected": -2.9827733039855957,
|
| 1000 |
-
"step": 640
|
| 1001 |
-
},
|
| 1002 |
-
{
|
| 1003 |
-
"epoch": 1.3605442176870748,
|
| 1004 |
-
"grad_norm": 37.7718208645946,
|
| 1005 |
-
"learning_rate": 1.3953046172178413e-07,
|
| 1006 |
-
"logits/chosen": -0.4676644206047058,
|
| 1007 |
-
"logits/rejected": -0.29953527450561523,
|
| 1008 |
-
"logps/chosen": -402.21417236328125,
|
| 1009 |
-
"logps/rejected": -548.4046020507812,
|
| 1010 |
-
"loss": 0.3546,
|
| 1011 |
-
"rewards/accuracies": 0.925000011920929,
|
| 1012 |
-
"rewards/chosen": -1.1596171855926514,
|
| 1013 |
-
"rewards/margins": 1.624544382095337,
|
| 1014 |
-
"rewards/rejected": -2.7841618061065674,
|
| 1015 |
-
"step": 650
|
| 1016 |
-
},
|
| 1017 |
-
{
|
| 1018 |
-
"epoch": 1.3814756671899528,
|
| 1019 |
-
"grad_norm": 32.318559995180884,
|
| 1020 |
-
"learning_rate": 1.3139467229135998e-07,
|
| 1021 |
-
"logits/chosen": -0.625167191028595,
|
| 1022 |
-
"logits/rejected": -0.40474313497543335,
|
| 1023 |
-
"logps/chosen": -410.2447204589844,
|
| 1024 |
-
"logps/rejected": -531.7794799804688,
|
| 1025 |
-
"loss": 0.3438,
|
| 1026 |
-
"rewards/accuracies": 0.8687499761581421,
|
| 1027 |
-
"rewards/chosen": -1.2046838998794556,
|
| 1028 |
-
"rewards/margins": 1.518923044204712,
|
| 1029 |
-
"rewards/rejected": -2.723607063293457,
|
| 1030 |
-
"step": 660
|
| 1031 |
-
},
|
| 1032 |
-
{
|
| 1033 |
-
"epoch": 1.402407116692831,
|
| 1034 |
-
"grad_norm": 26.63444910378978,
|
| 1035 |
-
"learning_rate": 1.2341787690142435e-07,
|
| 1036 |
-
"logits/chosen": -0.5080076456069946,
|
| 1037 |
-
"logits/rejected": -0.39028996229171753,
|
| 1038 |
-
"logps/chosen": -417.2349548339844,
|
| 1039 |
-
"logps/rejected": -571.6380615234375,
|
| 1040 |
-
"loss": 0.3373,
|
| 1041 |
-
"rewards/accuracies": 0.887499988079071,
|
| 1042 |
-
"rewards/chosen": -1.290380597114563,
|
| 1043 |
-
"rewards/margins": 1.4690136909484863,
|
| 1044 |
-
"rewards/rejected": -2.7593941688537598,
|
| 1045 |
-
"step": 670
|
| 1046 |
-
},
|
| 1047 |
-
{
|
| 1048 |
-
"epoch": 1.423338566195709,
|
| 1049 |
-
"grad_norm": 20.19948420893599,
|
| 1050 |
-
"learning_rate": 1.1561076868822755e-07,
|
| 1051 |
-
"logits/chosen": -0.6434948444366455,
|
| 1052 |
-
"logits/rejected": -0.3340326249599457,
|
| 1053 |
-
"logps/chosen": -425.06585693359375,
|
| 1054 |
-
"logps/rejected": -558.2105712890625,
|
| 1055 |
-
"loss": 0.3408,
|
| 1056 |
-
"rewards/accuracies": 0.8687499761581421,
|
| 1057 |
-
"rewards/chosen": -1.3932050466537476,
|
| 1058 |
-
"rewards/margins": 1.6028707027435303,
|
| 1059 |
-
"rewards/rejected": -2.9960758686065674,
|
| 1060 |
-
"step": 680
|
| 1061 |
-
},
|
| 1062 |
-
{
|
| 1063 |
-
"epoch": 1.4442700156985873,
|
| 1064 |
-
"grad_norm": 23.405895905191862,
|
| 1065 |
-
"learning_rate": 1.0798381331721107e-07,
|
| 1066 |
-
"logits/chosen": -0.45825833082199097,
|
| 1067 |
-
"logits/rejected": -0.26569095253944397,
|
| 1068 |
-
"logps/chosen": -413.29705810546875,
|
| 1069 |
-
"logps/rejected": -516.1429443359375,
|
| 1070 |
-
"loss": 0.3387,
|
| 1071 |
-
"rewards/accuracies": 0.84375,
|
| 1072 |
-
"rewards/chosen": -1.2925164699554443,
|
| 1073 |
-
"rewards/margins": 1.4324119091033936,
|
| 1074 |
-
"rewards/rejected": -2.724928379058838,
|
| 1075 |
-
"step": 690
|
| 1076 |
-
},
|
| 1077 |
-
{
|
| 1078 |
-
"epoch": 1.4652014652014653,
|
| 1079 |
-
"grad_norm": 26.00052056036154,
|
| 1080 |
-
"learning_rate": 1.0054723495346482e-07,
|
| 1081 |
-
"logits/chosen": -0.5345702767372131,
|
| 1082 |
-
"logits/rejected": -0.28709763288497925,
|
| 1083 |
-
"logps/chosen": -439.5155334472656,
|
| 1084 |
-
"logps/rejected": -595.0430908203125,
|
| 1085 |
-
"loss": 0.3385,
|
| 1086 |
-
"rewards/accuracies": 0.893750011920929,
|
| 1087 |
-
"rewards/chosen": -1.335257649421692,
|
| 1088 |
-
"rewards/margins": 1.8328659534454346,
|
| 1089 |
-
"rewards/rejected": -3.168123483657837,
|
| 1090 |
-
"step": 700
|
| 1091 |
-
},
|
| 1092 |
-
{
|
| 1093 |
-
"epoch": 1.4861329147043434,
|
| 1094 |
-
"grad_norm": 29.88922989167434,
|
| 1095 |
-
"learning_rate": 9.331100255592436e-08,
|
| 1096 |
-
"logits/chosen": -0.44970911741256714,
|
| 1097 |
-
"logits/rejected": -0.3438431918621063,
|
| 1098 |
-
"logps/chosen": -440.60009765625,
|
| 1099 |
-
"logps/rejected": -598.9261474609375,
|
| 1100 |
-
"loss": 0.3343,
|
| 1101 |
-
"rewards/accuracies": 0.8687499761581421,
|
| 1102 |
-
"rewards/chosen": -1.5944675207138062,
|
| 1103 |
-
"rewards/margins": 1.555371642112732,
|
| 1104 |
-
"rewards/rejected": -3.149839401245117,
|
| 1105 |
-
"step": 710
|
| 1106 |
-
},
|
| 1107 |
-
{
|
| 1108 |
-
"epoch": 1.5070643642072215,
|
| 1109 |
-
"grad_norm": 35.80679957440786,
|
| 1110 |
-
"learning_rate": 8.628481651367875e-08,
|
| 1111 |
-
"logits/chosen": -0.6270388960838318,
|
| 1112 |
-
"logits/rejected": -0.3709457218647003,
|
| 1113 |
-
"logps/chosen": -430.08734130859375,
|
| 1114 |
-
"logps/rejected": -598.91357421875,
|
| 1115 |
-
"loss": 0.324,
|
| 1116 |
-
"rewards/accuracies": 0.918749988079071,
|
| 1117 |
-
"rewards/chosen": -1.4218626022338867,
|
| 1118 |
-
"rewards/margins": 1.803938865661621,
|
| 1119 |
-
"rewards/rejected": -3.2258009910583496,
|
| 1120 |
-
"step": 720
|
| 1121 |
-
},
|
| 1122 |
-
{
|
| 1123 |
-
"epoch": 1.5279958137100995,
|
| 1124 |
-
"grad_norm": 35.40691778150228,
|
| 1125 |
-
"learning_rate": 7.947809564230445e-08,
|
| 1126 |
-
"logits/chosen": -0.7805054187774658,
|
| 1127 |
-
"logits/rejected": -0.5504633188247681,
|
| 1128 |
-
"logps/chosen": -417.5584411621094,
|
| 1129 |
-
"logps/rejected": -570.6063232421875,
|
| 1130 |
-
"loss": 0.3251,
|
| 1131 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 1132 |
-
"rewards/chosen": -1.2576138973236084,
|
| 1133 |
-
"rewards/margins": 1.5007960796356201,
|
| 1134 |
-
"rewards/rejected": -2.7584102153778076,
|
| 1135 |
-
"step": 730
|
| 1136 |
-
},
|
| 1137 |
-
{
|
| 1138 |
-
"epoch": 1.5489272632129776,
|
| 1139 |
-
"grad_norm": 22.542061727173216,
|
| 1140 |
-
"learning_rate": 7.289996455765748e-08,
|
| 1141 |
-
"logits/chosen": -0.7156350612640381,
|
| 1142 |
-
"logits/rejected": -0.5093780755996704,
|
| 1143 |
-
"logps/chosen": -440.34375,
|
| 1144 |
-
"logps/rejected": -552.8305053710938,
|
| 1145 |
-
"loss": 0.3487,
|
| 1146 |
-
"rewards/accuracies": 0.875,
|
| 1147 |
-
"rewards/chosen": -1.445813536643982,
|
| 1148 |
-
"rewards/margins": 1.539942741394043,
|
| 1149 |
-
"rewards/rejected": -2.9857563972473145,
|
| 1150 |
-
"step": 740
|
| 1151 |
-
},
|
| 1152 |
-
{
|
| 1153 |
-
"epoch": 1.5698587127158556,
|
| 1154 |
-
"grad_norm": 22.391994910590135,
|
| 1155 |
-
"learning_rate": 6.655924144404906e-08,
|
| 1156 |
-
"logits/chosen": -0.5188966393470764,
|
| 1157 |
-
"logits/rejected": -0.31423279643058777,
|
| 1158 |
-
"logps/chosen": -427.5140686035156,
|
| 1159 |
-
"logps/rejected": -574.3653564453125,
|
| 1160 |
-
"loss": 0.3262,
|
| 1161 |
-
"rewards/accuracies": 0.8812500238418579,
|
| 1162 |
-
"rewards/chosen": -1.3562893867492676,
|
| 1163 |
-
"rewards/margins": 1.616431474685669,
|
| 1164 |
-
"rewards/rejected": -2.9727206230163574,
|
| 1165 |
-
"step": 750
|
| 1166 |
-
},
|
| 1167 |
-
{
|
| 1168 |
-
"epoch": 1.5907901622187337,
|
| 1169 |
-
"grad_norm": 27.067458253461155,
|
| 1170 |
-
"learning_rate": 6.046442623320145e-08,
|
| 1171 |
-
"logits/chosen": -0.7770091891288757,
|
| 1172 |
-
"logits/rejected": -0.4757114350795746,
|
| 1173 |
-
"logps/chosen": -492.76416015625,
|
| 1174 |
-
"logps/rejected": -625.2634887695312,
|
| 1175 |
-
"loss": 0.3424,
|
| 1176 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 1177 |
-
"rewards/chosen": -1.4647096395492554,
|
| 1178 |
-
"rewards/margins": 1.647743821144104,
|
| 1179 |
-
"rewards/rejected": -3.1124536991119385,
|
| 1180 |
-
"step": 760
|
| 1181 |
-
},
|
| 1182 |
-
{
|
| 1183 |
-
"epoch": 1.6117216117216118,
|
| 1184 |
-
"grad_norm": 27.917693830137047,
|
| 1185 |
-
"learning_rate": 5.4623689209832484e-08,
|
| 1186 |
-
"logits/chosen": -0.8214691281318665,
|
| 1187 |
-
"logits/rejected": -0.43942517042160034,
|
| 1188 |
-
"logps/chosen": -453.11199951171875,
|
| 1189 |
-
"logps/rejected": -539.7423706054688,
|
| 1190 |
-
"loss": 0.3324,
|
| 1191 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 1192 |
-
"rewards/chosen": -1.3807549476623535,
|
| 1193 |
-
"rewards/margins": 1.52043879032135,
|
| 1194 |
-
"rewards/rejected": -2.901193618774414,
|
| 1195 |
-
"step": 770
|
| 1196 |
-
},
|
| 1197 |
-
{
|
| 1198 |
-
"epoch": 1.6326530612244898,
|
| 1199 |
-
"grad_norm": 35.7523647379527,
|
| 1200 |
-
"learning_rate": 4.904486005914027e-08,
|
| 1201 |
-
"logits/chosen": -0.7380484342575073,
|
| 1202 |
-
"logits/rejected": -0.5423134565353394,
|
| 1203 |
-
"logps/chosen": -448.4818420410156,
|
| 1204 |
-
"logps/rejected": -601.2604370117188,
|
| 1205 |
-
"loss": 0.3395,
|
| 1206 |
-
"rewards/accuracies": 0.8687499761581421,
|
| 1207 |
-
"rewards/chosen": -1.1892478466033936,
|
| 1208 |
-
"rewards/margins": 1.734458565711975,
|
| 1209 |
-
"rewards/rejected": -2.923706293106079,
|
| 1210 |
-
"step": 780
|
| 1211 |
-
},
|
| 1212 |
-
{
|
| 1213 |
-
"epoch": 1.653584510727368,
|
| 1214 |
-
"grad_norm": 31.011854985332448,
|
| 1215 |
-
"learning_rate": 4.373541737087263e-08,
|
| 1216 |
-
"logits/chosen": -0.7779333591461182,
|
| 1217 |
-
"logits/rejected": -0.49003076553344727,
|
| 1218 |
-
"logps/chosen": -406.6073303222656,
|
| 1219 |
-
"logps/rejected": -552.443359375,
|
| 1220 |
-
"loss": 0.3239,
|
| 1221 |
-
"rewards/accuracies": 0.8687499761581421,
|
| 1222 |
-
"rewards/chosen": -1.2308647632598877,
|
| 1223 |
-
"rewards/margins": 1.5439412593841553,
|
| 1224 |
-
"rewards/rejected": -2.774806261062622,
|
| 1225 |
-
"step": 790
|
| 1226 |
-
},
|
| 1227 |
-
{
|
| 1228 |
-
"epoch": 1.674515960230246,
|
| 1229 |
-
"grad_norm": 25.052300316679634,
|
| 1230 |
-
"learning_rate": 3.8702478614051345e-08,
|
| 1231 |
-
"logits/chosen": -0.7519603371620178,
|
| 1232 |
-
"logits/rejected": -0.4932466447353363,
|
| 1233 |
-
"logps/chosen": -459.8077087402344,
|
| 1234 |
-
"logps/rejected": -562.577392578125,
|
| 1235 |
-
"loss": 0.3121,
|
| 1236 |
-
"rewards/accuracies": 0.893750011920929,
|
| 1237 |
-
"rewards/chosen": -1.291195273399353,
|
| 1238 |
-
"rewards/margins": 1.5607550144195557,
|
| 1239 |
-
"rewards/rejected": -2.851950168609619,
|
| 1240 |
-
"step": 800
|
| 1241 |
-
},
|
| 1242 |
-
{
|
| 1243 |
-
"epoch": 1.695447409733124,
|
| 1244 |
-
"grad_norm": 28.051358555998323,
|
| 1245 |
-
"learning_rate": 3.3952790595787986e-08,
|
| 1246 |
-
"logits/chosen": -0.7064075469970703,
|
| 1247 |
-
"logits/rejected": -0.5052956938743591,
|
| 1248 |
-
"logps/chosen": -427.84832763671875,
|
| 1249 |
-
"logps/rejected": -587.9678955078125,
|
| 1250 |
-
"loss": 0.3363,
|
| 1251 |
-
"rewards/accuracies": 0.887499988079071,
|
| 1252 |
-
"rewards/chosen": -1.3051536083221436,
|
| 1253 |
-
"rewards/margins": 1.7959152460098267,
|
| 1254 |
-
"rewards/rejected": -3.1010687351226807,
|
| 1255 |
-
"step": 810
|
| 1256 |
-
},
|
| 1257 |
-
{
|
| 1258 |
-
"epoch": 1.716378859236002,
|
| 1259 |
-
"grad_norm": 31.67321567378694,
|
| 1260 |
-
"learning_rate": 2.9492720416985e-08,
|
| 1261 |
-
"logits/chosen": -0.8305649757385254,
|
| 1262 |
-
"logits/rejected": -0.7014783620834351,
|
| 1263 |
-
"logps/chosen": -413.65576171875,
|
| 1264 |
-
"logps/rejected": -569.78759765625,
|
| 1265 |
-
"loss": 0.3315,
|
| 1266 |
-
"rewards/accuracies": 0.862500011920929,
|
| 1267 |
-
"rewards/chosen": -1.4015345573425293,
|
| 1268 |
-
"rewards/margins": 1.6014522314071655,
|
| 1269 |
-
"rewards/rejected": -3.002986431121826,
|
| 1270 |
-
"step": 820
|
| 1271 |
-
},
|
| 1272 |
-
{
|
| 1273 |
-
"epoch": 1.7373103087388801,
|
| 1274 |
-
"grad_norm": 21.644645673450903,
|
| 1275 |
-
"learning_rate": 2.5328246937043525e-08,
|
| 1276 |
-
"logits/chosen": -0.8004710078239441,
|
| 1277 |
-
"logits/rejected": -0.6543309092521667,
|
| 1278 |
-
"logps/chosen": -454.3511657714844,
|
| 1279 |
-
"logps/rejected": -579.9097900390625,
|
| 1280 |
-
"loss": 0.3224,
|
| 1281 |
-
"rewards/accuracies": 0.862500011920929,
|
| 1282 |
-
"rewards/chosen": -1.3592346906661987,
|
| 1283 |
-
"rewards/margins": 1.5871307849884033,
|
| 1284 |
-
"rewards/rejected": -2.9463653564453125,
|
| 1285 |
-
"step": 830
|
| 1286 |
-
},
|
| 1287 |
-
{
|
| 1288 |
-
"epoch": 1.7582417582417582,
|
| 1289 |
-
"grad_norm": 29.695508251939298,
|
| 1290 |
-
"learning_rate": 2.1464952759020856e-08,
|
| 1291 |
-
"logits/chosen": -0.5647540092468262,
|
| 1292 |
-
"logits/rejected": -0.4591120779514313,
|
| 1293 |
-
"logps/chosen": -411.2245178222656,
|
| 1294 |
-
"logps/rejected": -579.0653076171875,
|
| 1295 |
-
"loss": 0.3265,
|
| 1296 |
-
"rewards/accuracies": 0.887499988079071,
|
| 1297 |
-
"rewards/chosen": -1.4194848537445068,
|
| 1298 |
-
"rewards/margins": 1.6787960529327393,
|
| 1299 |
-
"rewards/rejected": -3.098280906677246,
|
| 1300 |
-
"step": 840
|
| 1301 |
-
},
|
| 1302 |
-
{
|
| 1303 |
-
"epoch": 1.7791732077446363,
|
| 1304 |
-
"grad_norm": 30.24532074338387,
|
| 1305 |
-
"learning_rate": 1.7908016745981856e-08,
|
| 1306 |
-
"logits/chosen": -0.631384551525116,
|
| 1307 |
-
"logits/rejected": -0.5165312886238098,
|
| 1308 |
-
"logps/chosen": -408.71173095703125,
|
| 1309 |
-
"logps/rejected": -562.2457275390625,
|
| 1310 |
-
"loss": 0.3275,
|
| 1311 |
-
"rewards/accuracies": 0.8500000238418579,
|
| 1312 |
-
"rewards/chosen": -1.459183931350708,
|
| 1313 |
-
"rewards/margins": 1.5402036905288696,
|
| 1314 |
-
"rewards/rejected": -2.999387264251709,
|
| 1315 |
-
"step": 850
|
| 1316 |
-
},
|
| 1317 |
-
{
|
| 1318 |
-
"epoch": 1.8001046572475143,
|
| 1319 |
-
"grad_norm": 28.810585688980133,
|
| 1320 |
-
"learning_rate": 1.4662207078575684e-08,
|
| 1321 |
-
"logits/chosen": -0.8192375302314758,
|
| 1322 |
-
"logits/rejected": -0.6066277623176575,
|
| 1323 |
-
"logps/chosen": -447.3929748535156,
|
| 1324 |
-
"logps/rejected": -533.9404296875,
|
| 1325 |
-
"loss": 0.3479,
|
| 1326 |
-
"rewards/accuracies": 0.84375,
|
| 1327 |
-
"rewards/chosen": -1.381157636642456,
|
| 1328 |
-
"rewards/margins": 1.4070510864257812,
|
| 1329 |
-
"rewards/rejected": -2.788208484649658,
|
| 1330 |
-
"step": 860
|
| 1331 |
-
},
|
| 1332 |
-
{
|
| 1333 |
-
"epoch": 1.8210361067503924,
|
| 1334 |
-
"grad_norm": 24.013363968760547,
|
| 1335 |
-
"learning_rate": 1.1731874863145142e-08,
|
| 1336 |
-
"logits/chosen": -0.7617680430412292,
|
| 1337 |
-
"logits/rejected": -0.47873860597610474,
|
| 1338 |
-
"logps/chosen": -454.9310607910156,
|
| 1339 |
-
"logps/rejected": -573.6890869140625,
|
| 1340 |
-
"loss": 0.331,
|
| 1341 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 1342 |
-
"rewards/chosen": -1.3413336277008057,
|
| 1343 |
-
"rewards/margins": 1.7013572454452515,
|
| 1344 |
-
"rewards/rejected": -3.0426909923553467,
|
| 1345 |
-
"step": 870
|
| 1346 |
-
},
|
| 1347 |
-
{
|
| 1348 |
-
"epoch": 1.8419675562532705,
|
| 1349 |
-
"grad_norm": 32.07599972860766,
|
| 1350 |
-
"learning_rate": 9.12094829893642e-09,
|
| 1351 |
-
"logits/chosen": -0.7640255689620972,
|
| 1352 |
-
"logits/rejected": -0.4442223012447357,
|
| 1353 |
-
"logps/chosen": -438.59490966796875,
|
| 1354 |
-
"logps/rejected": -557.7069091796875,
|
| 1355 |
-
"loss": 0.3402,
|
| 1356 |
-
"rewards/accuracies": 0.8812500238418579,
|
| 1357 |
-
"rewards/chosen": -1.409976840019226,
|
| 1358 |
-
"rewards/margins": 1.640062928199768,
|
| 1359 |
-
"rewards/rejected": -3.050039529800415,
|
| 1360 |
-
"step": 880
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 1.8628990057561485,
|
| 1364 |
-
"grad_norm": 24.432518139204586,
|
| 1365 |
-
"learning_rate": 6.832927412229017e-09,
|
| 1366 |
-
"logits/chosen": -0.8613092303276062,
|
| 1367 |
-
"logits/rejected": -0.6306430101394653,
|
| 1368 |
-
"logps/chosen": -456.99072265625,
|
| 1369 |
-
"logps/rejected": -594.9552001953125,
|
| 1370 |
-
"loss": 0.3223,
|
| 1371 |
-
"rewards/accuracies": 0.9125000238418579,
|
| 1372 |
-
"rewards/chosen": -1.3222033977508545,
|
| 1373 |
-
"rewards/margins": 1.7843767404556274,
|
| 1374 |
-
"rewards/rejected": -3.1065802574157715,
|
| 1375 |
-
"step": 890
|
| 1376 |
-
},
|
| 1377 |
-
{
|
| 1378 |
-
"epoch": 1.8838304552590266,
|
| 1379 |
-
"grad_norm": 27.645379277195858,
|
| 1380 |
-
"learning_rate": 4.8708793644441086e-09,
|
| 1381 |
-
"logits/chosen": -0.7418426275253296,
|
| 1382 |
-
"logits/rejected": -0.4674626886844635,
|
| 1383 |
-
"logps/chosen": -423.3577575683594,
|
| 1384 |
-
"logps/rejected": -554.7745361328125,
|
| 1385 |
-
"loss": 0.3093,
|
| 1386 |
-
"rewards/accuracies": 0.925000011920929,
|
| 1387 |
-
"rewards/chosen": -1.3512873649597168,
|
| 1388 |
-
"rewards/margins": 1.6987825632095337,
|
| 1389 |
-
"rewards/rejected": -3.050069808959961,
|
| 1390 |
-
"step": 900
|
| 1391 |
-
},
|
| 1392 |
-
{
|
| 1393 |
-
"epoch": 1.9047619047619047,
|
| 1394 |
-
"grad_norm": 28.923783797405456,
|
| 1395 |
-
"learning_rate": 3.2374343405217884e-09,
|
| 1396 |
-
"logits/chosen": -0.6597197651863098,
|
| 1397 |
-
"logits/rejected": -0.4865621030330658,
|
| 1398 |
-
"logps/chosen": -408.0905456542969,
|
| 1399 |
-
"logps/rejected": -535.84521484375,
|
| 1400 |
-
"loss": 0.3367,
|
| 1401 |
-
"rewards/accuracies": 0.887499988079071,
|
| 1402 |
-
"rewards/chosen": -1.4176795482635498,
|
| 1403 |
-
"rewards/margins": 1.6123201847076416,
|
| 1404 |
-
"rewards/rejected": -3.0299999713897705,
|
| 1405 |
-
"step": 910
|
| 1406 |
-
},
|
| 1407 |
-
{
|
| 1408 |
-
"epoch": 1.9256933542647827,
|
| 1409 |
-
"grad_norm": 28.331424470243782,
|
| 1410 |
-
"learning_rate": 1.9347820230782295e-09,
|
| 1411 |
-
"logits/chosen": -0.8034309148788452,
|
| 1412 |
-
"logits/rejected": -0.6734964847564697,
|
| 1413 |
-
"logps/chosen": -424.65673828125,
|
| 1414 |
-
"logps/rejected": -592.0272216796875,
|
| 1415 |
-
"loss": 0.3345,
|
| 1416 |
-
"rewards/accuracies": 0.8812500238418579,
|
| 1417 |
-
"rewards/chosen": -1.302234172821045,
|
| 1418 |
-
"rewards/margins": 1.7047960758209229,
|
| 1419 |
-
"rewards/rejected": -3.0070300102233887,
|
| 1420 |
-
"step": 920
|
| 1421 |
-
},
|
| 1422 |
-
{
|
| 1423 |
-
"epoch": 1.9466248037676608,
|
| 1424 |
-
"grad_norm": 31.096436385559773,
|
| 1425 |
-
"learning_rate": 9.64668657069706e-10,
|
| 1426 |
-
"logits/chosen": -0.8711435198783875,
|
| 1427 |
-
"logits/rejected": -0.6492515206336975,
|
| 1428 |
-
"logps/chosen": -412.20159912109375,
|
| 1429 |
-
"logps/rejected": -539.3062744140625,
|
| 1430 |
-
"loss": 0.3326,
|
| 1431 |
-
"rewards/accuracies": 0.856249988079071,
|
| 1432 |
-
"rewards/chosen": -1.3435027599334717,
|
| 1433 |
-
"rewards/margins": 1.5116230249404907,
|
| 1434 |
-
"rewards/rejected": -2.855125904083252,
|
| 1435 |
-
"step": 930
|
| 1436 |
-
},
|
| 1437 |
-
{
|
| 1438 |
-
"epoch": 1.9675562532705388,
|
| 1439 |
-
"grad_norm": 39.27145853425959,
|
| 1440 |
-
"learning_rate": 3.2839470889836627e-10,
|
| 1441 |
-
"logits/chosen": -0.8457353711128235,
|
| 1442 |
-
"logits/rejected": -0.46064504981040955,
|
| 1443 |
-
"logps/chosen": -439.29815673828125,
|
| 1444 |
-
"logps/rejected": -539.5534057617188,
|
| 1445 |
-
"loss": 0.3418,
|
| 1446 |
-
"rewards/accuracies": 0.8812500238418579,
|
| 1447 |
-
"rewards/chosen": -1.3375287055969238,
|
| 1448 |
-
"rewards/margins": 1.5794165134429932,
|
| 1449 |
-
"rewards/rejected": -2.916945457458496,
|
| 1450 |
-
"step": 940
|
| 1451 |
-
},
|
| 1452 |
-
{
|
| 1453 |
-
"epoch": 1.988487702773417,
|
| 1454 |
-
"grad_norm": 30.98924464228554,
|
| 1455 |
-
"learning_rate": 2.6813123097352287e-11,
|
| 1456 |
-
"logits/chosen": -0.7241233587265015,
|
| 1457 |
-
"logits/rejected": -0.4993128180503845,
|
| 1458 |
-
"logps/chosen": -443.7752380371094,
|
| 1459 |
-
"logps/rejected": -581.8499755859375,
|
| 1460 |
-
"loss": 0.3187,
|
| 1461 |
-
"rewards/accuracies": 0.893750011920929,
|
| 1462 |
-
"rewards/chosen": -1.385292649269104,
|
| 1463 |
-
"rewards/margins": 1.7712970972061157,
|
| 1464 |
-
"rewards/rejected": -3.1565897464752197,
|
| 1465 |
-
"step": 950
|
| 1466 |
-
},
|
| 1467 |
-
{
|
| 1468 |
-
"epoch": 1.9968602825745683,
|
| 1469 |
-
"eval_logits/chosen": -0.5949550271034241,
|
| 1470 |
-
"eval_logits/rejected": -0.44720327854156494,
|
| 1471 |
-
"eval_logps/chosen": -426.89898681640625,
|
| 1472 |
-
"eval_logps/rejected": -545.3427124023438,
|
| 1473 |
-
"eval_loss": 0.5225037932395935,
|
| 1474 |
-
"eval_rewards/accuracies": 0.7734375,
|
| 1475 |
-
"eval_rewards/chosen": -1.547555685043335,
|
| 1476 |
-
"eval_rewards/margins": 1.1972318887710571,
|
| 1477 |
-
"eval_rewards/rejected": -2.7447874546051025,
|
| 1478 |
-
"eval_runtime": 170.0564,
|
| 1479 |
-
"eval_samples_per_second": 11.761,
|
| 1480 |
-
"eval_steps_per_second": 0.188,
|
| 1481 |
-
"step": 954
|
| 1482 |
-
},
|
| 1483 |
-
{
|
| 1484 |
-
"epoch": 1.9968602825745683,
|
| 1485 |
-
"step": 954,
|
| 1486 |
"total_flos": 0.0,
|
| 1487 |
-
"train_loss": 0.
|
| 1488 |
-
"train_runtime":
|
| 1489 |
-
"train_samples_per_second":
|
| 1490 |
-
"train_steps_per_second":
|
| 1491 |
}
|
| 1492 |
],
|
| 1493 |
"logging_steps": 10,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9984301412872841,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 477,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0020931449502878076,
|
| 13 |
+
"grad_norm": 2.2216637005002386,
|
| 14 |
+
"learning_rate": 1.0416666666666666e-08,
|
| 15 |
+
"logits/chosen": -1.6314841508865356,
|
| 16 |
+
"logits/rejected": -1.353194236755371,
|
| 17 |
+
"logps/chosen": -368.29144287109375,
|
| 18 |
+
"logps/rejected": -370.7230224609375,
|
| 19 |
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
|
|
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.020931449502878074,
|
| 28 |
+
"grad_norm": 2.1524494849038778,
|
| 29 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 30 |
+
"logits/chosen": -1.2345138788223267,
|
| 31 |
+
"logits/rejected": -1.203288197517395,
|
| 32 |
+
"logps/chosen": -314.1005554199219,
|
| 33 |
+
"logps/rejected": -286.3752136230469,
|
| 34 |
+
"loss": 0.693,
|
| 35 |
+
"rewards/accuracies": 0.4930555522441864,
|
| 36 |
+
"rewards/chosen": -2.686006700969301e-05,
|
| 37 |
+
"rewards/margins": 0.0007661848794668913,
|
| 38 |
+
"rewards/rejected": -0.0007930449792183936,
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.04186289900575615,
|
| 43 |
+
"grad_norm": 2.0655596511164083,
|
| 44 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 45 |
+
"logits/chosen": -1.304213523864746,
|
| 46 |
+
"logits/rejected": -1.1859759092330933,
|
| 47 |
+
"logps/chosen": -320.55169677734375,
|
| 48 |
+
"logps/rejected": -283.34149169921875,
|
| 49 |
"loss": 0.6931,
|
| 50 |
+
"rewards/accuracies": 0.518750011920929,
|
| 51 |
+
"rewards/chosen": -0.00017411960288882256,
|
| 52 |
+
"rewards/margins": 5.0039543566526845e-05,
|
| 53 |
+
"rewards/rejected": -0.0002241591428173706,
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 0.06279434850863422,
|
| 58 |
+
"grad_norm": 2.12111028976462,
|
| 59 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 60 |
+
"logits/chosen": -1.3263781070709229,
|
| 61 |
+
"logits/rejected": -1.2141128778457642,
|
| 62 |
+
"logps/chosen": -294.7076110839844,
|
| 63 |
+
"logps/rejected": -251.52481079101562,
|
| 64 |
+
"loss": 0.6931,
|
| 65 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 66 |
+
"rewards/chosen": 0.0012247291160747409,
|
| 67 |
+
"rewards/margins": -0.0003233866882510483,
|
| 68 |
+
"rewards/rejected": 0.0015481156297028065,
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.0837257980115123,
|
| 73 |
+
"grad_norm": 1.9880970924710342,
|
| 74 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 75 |
+
"logits/chosen": -1.2525466680526733,
|
| 76 |
+
"logits/rejected": -1.1410564184188843,
|
| 77 |
+
"logps/chosen": -270.8800354003906,
|
| 78 |
+
"logps/rejected": -266.47222900390625,
|
| 79 |
+
"loss": 0.6922,
|
| 80 |
+
"rewards/accuracies": 0.5625,
|
| 81 |
+
"rewards/chosen": 0.005721802823245525,
|
| 82 |
+
"rewards/margins": 0.0018067390192300081,
|
| 83 |
+
"rewards/rejected": 0.003915064036846161,
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
"epoch": 0.10465724751439037,
|
| 88 |
+
"grad_norm": 2.1837707951870677,
|
| 89 |
+
"learning_rate": 4.999731868769026e-07,
|
| 90 |
+
"logits/chosen": -1.2781437635421753,
|
| 91 |
+
"logits/rejected": -1.1634522676467896,
|
| 92 |
+
"logps/chosen": -299.9478454589844,
|
| 93 |
+
"logps/rejected": -279.74188232421875,
|
| 94 |
+
"loss": 0.6902,
|
| 95 |
+
"rewards/accuracies": 0.59375,
|
| 96 |
+
"rewards/chosen": 0.014337332919239998,
|
| 97 |
+
"rewards/margins": 0.005479713901877403,
|
| 98 |
+
"rewards/rejected": 0.00885761994868517,
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.12558869701726844,
|
| 103 |
+
"grad_norm": 2.3039683597630227,
|
| 104 |
+
"learning_rate": 4.990353313429303e-07,
|
| 105 |
+
"logits/chosen": -1.3196525573730469,
|
| 106 |
+
"logits/rejected": -1.2014460563659668,
|
| 107 |
+
"logps/chosen": -259.46942138671875,
|
| 108 |
+
"logps/rejected": -252.1083984375,
|
| 109 |
+
"loss": 0.6885,
|
| 110 |
+
"rewards/accuracies": 0.731249988079071,
|
| 111 |
+
"rewards/chosen": 0.027976389974355698,
|
| 112 |
+
"rewards/margins": 0.010958237573504448,
|
| 113 |
+
"rewards/rejected": 0.0170181542634964,
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.14652014652014653,
|
| 118 |
+
"grad_norm": 2.025942457662755,
|
| 119 |
+
"learning_rate": 4.967625656594781e-07,
|
| 120 |
+
"logits/chosen": -1.246511697769165,
|
| 121 |
+
"logits/rejected": -1.1918096542358398,
|
| 122 |
+
"logps/chosen": -304.3538513183594,
|
| 123 |
+
"logps/rejected": -286.70849609375,
|
| 124 |
+
"loss": 0.6857,
|
| 125 |
+
"rewards/accuracies": 0.65625,
|
| 126 |
+
"rewards/chosen": 0.034880757331848145,
|
| 127 |
+
"rewards/margins": 0.013793361373245716,
|
| 128 |
+
"rewards/rejected": 0.021087396889925003,
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.1674515960230246,
|
| 133 |
+
"grad_norm": 2.139816541586556,
|
| 134 |
+
"learning_rate": 4.93167072587771e-07,
|
| 135 |
+
"logits/chosen": -1.3581794500350952,
|
| 136 |
+
"logits/rejected": -1.1709249019622803,
|
| 137 |
+
"logps/chosen": -337.73638916015625,
|
| 138 |
+
"logps/rejected": -256.8132629394531,
|
| 139 |
+
"loss": 0.6816,
|
| 140 |
+
"rewards/accuracies": 0.737500011920929,
|
| 141 |
+
"rewards/chosen": 0.04202472046017647,
|
| 142 |
+
"rewards/margins": 0.029817480593919754,
|
| 143 |
+
"rewards/rejected": 0.01220723893493414,
|
| 144 |
"step": 80
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.18838304552590268,
|
| 148 |
+
"grad_norm": 2.184463703029387,
|
| 149 |
+
"learning_rate": 4.882681251368548e-07,
|
| 150 |
+
"logits/chosen": -1.2384573221206665,
|
| 151 |
+
"logits/rejected": -1.1635406017303467,
|
| 152 |
+
"logps/chosen": -260.00115966796875,
|
| 153 |
+
"logps/rejected": -258.2013244628906,
|
| 154 |
+
"loss": 0.6781,
|
| 155 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 156 |
+
"rewards/chosen": 0.025884713977575302,
|
| 157 |
+
"rewards/margins": 0.02733892761170864,
|
| 158 |
+
"rewards/rejected": -0.0014542154967784882,
|
| 159 |
"step": 90
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.20931449502878074,
|
| 163 |
+
"grad_norm": 2.302037128469088,
|
| 164 |
+
"learning_rate": 4.820919832540181e-07,
|
| 165 |
+
"logits/chosen": -1.304317831993103,
|
| 166 |
+
"logits/rejected": -1.1416256427764893,
|
| 167 |
+
"logps/chosen": -312.20867919921875,
|
| 168 |
+
"logps/rejected": -292.9761962890625,
|
| 169 |
+
"loss": 0.6731,
|
| 170 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 171 |
+
"rewards/chosen": 0.016404878348112106,
|
| 172 |
+
"rewards/margins": 0.0562920868396759,
|
| 173 |
+
"rewards/rejected": -0.0398872084915638,
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
"epoch": 0.2302459445316588,
|
| 178 |
+
"grad_norm": 2.476640494327824,
|
| 179 |
+
"learning_rate": 4.7467175306295647e-07,
|
| 180 |
+
"logits/chosen": -1.3065650463104248,
|
| 181 |
+
"logits/rejected": -1.2025437355041504,
|
| 182 |
+
"logps/chosen": -299.5341796875,
|
| 183 |
+
"logps/rejected": -286.65081787109375,
|
| 184 |
+
"loss": 0.6719,
|
| 185 |
+
"rewards/accuracies": 0.637499988079071,
|
| 186 |
+
"rewards/chosen": -0.019125858321785927,
|
| 187 |
+
"rewards/margins": 0.05537647008895874,
|
| 188 |
+
"rewards/rejected": -0.07450231909751892,
|
| 189 |
"step": 110
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.25117739403453687,
|
| 193 |
+
"grad_norm": 2.42503756687085,
|
| 194 |
+
"learning_rate": 4.6604720940421207e-07,
|
| 195 |
+
"logits/chosen": -1.3212846517562866,
|
| 196 |
+
"logits/rejected": -1.3139406442642212,
|
| 197 |
+
"logps/chosen": -273.289306640625,
|
| 198 |
+
"logps/rejected": -277.60711669921875,
|
| 199 |
+
"loss": 0.6601,
|
| 200 |
+
"rewards/accuracies": 0.65625,
|
| 201 |
+
"rewards/chosen": -0.03048776648938656,
|
| 202 |
+
"rewards/margins": 0.06633703410625458,
|
| 203 |
+
"rewards/rejected": -0.09682480990886688,
|
| 204 |
"step": 120
|
| 205 |
},
|
| 206 |
{
|
| 207 |
"epoch": 0.272108843537415,
|
| 208 |
+
"grad_norm": 2.7217854249308493,
|
| 209 |
+
"learning_rate": 4.5626458262912735e-07,
|
| 210 |
+
"logits/chosen": -1.3054877519607544,
|
| 211 |
+
"logits/rejected": -1.209935188293457,
|
| 212 |
+
"logps/chosen": -295.87884521484375,
|
| 213 |
+
"logps/rejected": -290.7591247558594,
|
| 214 |
+
"loss": 0.652,
|
| 215 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 216 |
+
"rewards/chosen": -0.06190527230501175,
|
| 217 |
+
"rewards/margins": 0.07534609735012054,
|
| 218 |
+
"rewards/rejected": -0.1372513771057129,
|
| 219 |
"step": 130
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.29304029304029305,
|
| 223 |
+
"grad_norm": 2.884195278457084,
|
| 224 |
+
"learning_rate": 4.453763107901675e-07,
|
| 225 |
+
"logits/chosen": -1.3432964086532593,
|
| 226 |
+
"logits/rejected": -1.202289342880249,
|
| 227 |
+
"logps/chosen": -339.2923278808594,
|
| 228 |
+
"logps/rejected": -292.0403137207031,
|
| 229 |
+
"loss": 0.6479,
|
| 230 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 231 |
+
"rewards/chosen": -0.05155748128890991,
|
| 232 |
+
"rewards/margins": 0.12817394733428955,
|
| 233 |
+
"rewards/rejected": -0.17973145842552185,
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
"epoch": 0.3139717425431711,
|
| 238 |
+
"grad_norm": 3.293629728295891,
|
| 239 |
+
"learning_rate": 4.3344075855595097e-07,
|
| 240 |
+
"logits/chosen": -1.3512450456619263,
|
| 241 |
+
"logits/rejected": -1.267796516418457,
|
| 242 |
+
"logps/chosen": -301.23553466796875,
|
| 243 |
+
"logps/rejected": -276.87689208984375,
|
| 244 |
+
"loss": 0.6454,
|
| 245 |
+
"rewards/accuracies": 0.731249988079071,
|
| 246 |
+
"rewards/chosen": -0.09927480667829514,
|
| 247 |
+
"rewards/margins": 0.12617069482803345,
|
| 248 |
+
"rewards/rejected": -0.22544550895690918,
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.3349031920460492,
|
| 253 |
+
"grad_norm": 5.329039698823607,
|
| 254 |
+
"learning_rate": 4.2052190435769554e-07,
|
| 255 |
+
"logits/chosen": -1.3824660778045654,
|
| 256 |
+
"logits/rejected": -1.1998783349990845,
|
| 257 |
+
"logps/chosen": -293.38714599609375,
|
| 258 |
+
"logps/rejected": -284.198486328125,
|
| 259 |
+
"loss": 0.6335,
|
| 260 |
+
"rewards/accuracies": 0.668749988079071,
|
| 261 |
+
"rewards/chosen": -0.07924506813287735,
|
| 262 |
+
"rewards/margins": 0.18151573836803436,
|
| 263 |
+
"rewards/rejected": -0.2607608139514923,
|
| 264 |
"step": 160
|
| 265 |
},
|
| 266 |
{
|
| 267 |
"epoch": 0.35583464154892724,
|
| 268 |
+
"grad_norm": 4.73498829155891,
|
| 269 |
+
"learning_rate": 4.0668899744407567e-07,
|
| 270 |
+
"logits/chosen": -1.3166488409042358,
|
| 271 |
+
"logits/rejected": -1.225121021270752,
|
| 272 |
+
"logps/chosen": -283.9803161621094,
|
| 273 |
+
"logps/rejected": -269.71551513671875,
|
| 274 |
+
"loss": 0.6341,
|
| 275 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 276 |
+
"rewards/chosen": -0.157345250248909,
|
| 277 |
+
"rewards/margins": 0.14062240719795227,
|
| 278 |
+
"rewards/rejected": -0.2979676425457001,
|
| 279 |
"step": 170
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.37676609105180536,
|
| 283 |
+
"grad_norm": 4.148844444439802,
|
| 284 |
+
"learning_rate": 3.920161866827889e-07,
|
| 285 |
+
"logits/chosen": -1.2676044702529907,
|
| 286 |
+
"logits/rejected": -1.2061588764190674,
|
| 287 |
+
"logps/chosen": -293.41998291015625,
|
| 288 |
+
"logps/rejected": -282.3377380371094,
|
| 289 |
+
"loss": 0.6255,
|
| 290 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 291 |
+
"rewards/chosen": -0.14930489659309387,
|
| 292 |
+
"rewards/margins": 0.13889411091804504,
|
| 293 |
+
"rewards/rejected": -0.2881990075111389,
|
| 294 |
"step": 180
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 0.3976975405546834,
|
| 298 |
+
"grad_norm": 4.210268144972537,
|
| 299 |
+
"learning_rate": 3.765821230985757e-07,
|
| 300 |
+
"logits/chosen": -1.270572543144226,
|
| 301 |
+
"logits/rejected": -1.255990982055664,
|
| 302 |
+
"logps/chosen": -278.5187683105469,
|
| 303 |
+
"logps/rejected": -289.1994934082031,
|
| 304 |
+
"loss": 0.6236,
|
| 305 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 306 |
+
"rewards/chosen": -0.05700838565826416,
|
| 307 |
+
"rewards/margins": 0.1534106433391571,
|
| 308 |
+
"rewards/rejected": -0.21041902899742126,
|
| 309 |
"step": 190
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.4186289900575615,
|
| 313 |
+
"grad_norm": 4.513968949295029,
|
| 314 |
+
"learning_rate": 3.604695382782159e-07,
|
| 315 |
+
"logits/chosen": -1.2794849872589111,
|
| 316 |
+
"logits/rejected": -1.2373203039169312,
|
| 317 |
+
"logps/chosen": -275.9900817871094,
|
| 318 |
+
"logps/rejected": -305.5406188964844,
|
| 319 |
+
"loss": 0.6285,
|
| 320 |
+
"rewards/accuracies": 0.643750011920929,
|
| 321 |
+
"rewards/chosen": -0.1595189869403839,
|
| 322 |
+
"rewards/margins": 0.13092303276062012,
|
| 323 |
+
"rewards/rejected": -0.29044201970100403,
|
| 324 |
"step": 200
|
| 325 |
},
|
| 326 |
{
|
| 327 |
"epoch": 0.43956043956043955,
|
| 328 |
+
"grad_norm": 4.130260296784633,
|
| 329 |
+
"learning_rate": 3.4376480090239047e-07,
|
| 330 |
+
"logits/chosen": -1.3395261764526367,
|
| 331 |
+
"logits/rejected": -1.160781741142273,
|
| 332 |
+
"logps/chosen": -336.36309814453125,
|
| 333 |
+
"logps/rejected": -299.7681884765625,
|
| 334 |
+
"loss": 0.6274,
|
| 335 |
+
"rewards/accuracies": 0.706250011920929,
|
| 336 |
+
"rewards/chosen": -0.1335216909646988,
|
| 337 |
+
"rewards/margins": 0.22881793975830078,
|
| 338 |
+
"rewards/rejected": -0.36233964562416077,
|
| 339 |
"step": 210
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.4604918890633176,
|
| 343 |
+
"grad_norm": 4.2020468567803135,
|
| 344 |
+
"learning_rate": 3.265574537815398e-07,
|
| 345 |
+
"logits/chosen": -1.2917451858520508,
|
| 346 |
+
"logits/rejected": -1.2045356035232544,
|
| 347 |
+
"logps/chosen": -264.0050354003906,
|
| 348 |
+
"logps/rejected": -286.49566650390625,
|
| 349 |
+
"loss": 0.6234,
|
| 350 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 351 |
+
"rewards/chosen": -0.21042123436927795,
|
| 352 |
+
"rewards/margins": 0.1836375892162323,
|
| 353 |
+
"rewards/rejected": -0.39405879378318787,
|
| 354 |
"step": 220
|
| 355 |
},
|
| 356 |
{
|
| 357 |
"epoch": 0.48142333856619574,
|
| 358 |
+
"grad_norm": 5.462895375975161,
|
| 359 |
+
"learning_rate": 3.0893973387735683e-07,
|
| 360 |
+
"logits/chosen": -1.33823561668396,
|
| 361 |
+
"logits/rejected": -1.204892873764038,
|
| 362 |
+
"logps/chosen": -285.23309326171875,
|
| 363 |
+
"logps/rejected": -287.7167663574219,
|
| 364 |
+
"loss": 0.6225,
|
| 365 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 366 |
+
"rewards/chosen": -0.1863180696964264,
|
| 367 |
+
"rewards/margins": 0.2673104703426361,
|
| 368 |
+
"rewards/rejected": -0.4536285400390625,
|
| 369 |
"step": 230
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.5023547880690737,
|
| 373 |
+
"grad_norm": 4.249954784852311,
|
| 374 |
+
"learning_rate": 2.910060778827554e-07,
|
| 375 |
+
"logits/chosen": -1.311199426651001,
|
| 376 |
+
"logits/rejected": -1.22099769115448,
|
| 377 |
+
"logps/chosen": -313.4962463378906,
|
| 378 |
+
"logps/rejected": -319.8878479003906,
|
| 379 |
+
"loss": 0.606,
|
| 380 |
"rewards/accuracies": 0.699999988079071,
|
| 381 |
+
"rewards/chosen": -0.2013070285320282,
|
| 382 |
+
"rewards/margins": 0.27799052000045776,
|
| 383 |
+
"rewards/rejected": -0.47929757833480835,
|
| 384 |
"step": 240
|
| 385 |
},
|
| 386 |
{
|
| 387 |
"epoch": 0.5232862375719518,
|
| 388 |
+
"grad_norm": 6.044167355639739,
|
| 389 |
+
"learning_rate": 2.7285261601056697e-07,
|
| 390 |
+
"logits/chosen": -1.3467825651168823,
|
| 391 |
+
"logits/rejected": -1.194777011871338,
|
| 392 |
+
"logps/chosen": -313.8829040527344,
|
| 393 |
+
"logps/rejected": -296.2748107910156,
|
| 394 |
+
"loss": 0.612,
|
| 395 |
+
"rewards/accuracies": 0.706250011920929,
|
| 396 |
+
"rewards/chosen": -0.2433101385831833,
|
| 397 |
+
"rewards/margins": 0.253059446811676,
|
| 398 |
+
"rewards/rejected": -0.49636954069137573,
|
| 399 |
"step": 250
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.54421768707483,
|
| 403 |
+
"grad_norm": 4.939384686255436,
|
| 404 |
+
"learning_rate": 2.5457665670441937e-07,
|
| 405 |
+
"logits/chosen": -1.2902387380599976,
|
| 406 |
+
"logits/rejected": -1.2375991344451904,
|
| 407 |
+
"logps/chosen": -289.4366455078125,
|
| 408 |
+
"logps/rejected": -296.7834167480469,
|
| 409 |
+
"loss": 0.608,
|
| 410 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 411 |
+
"rewards/chosen": -0.23678669333457947,
|
| 412 |
+
"rewards/margins": 0.2222580909729004,
|
| 413 |
+
"rewards/rejected": -0.45904478430747986,
|
| 414 |
"step": 260
|
| 415 |
},
|
| 416 |
{
|
| 417 |
"epoch": 0.565149136577708,
|
| 418 |
+
"grad_norm": 3.639946353207058,
|
| 419 |
+
"learning_rate": 2.3627616503391812e-07,
|
| 420 |
+
"logits/chosen": -1.321746826171875,
|
| 421 |
+
"logits/rejected": -1.22855806350708,
|
| 422 |
+
"logps/chosen": -322.65325927734375,
|
| 423 |
+
"logps/rejected": -311.2021179199219,
|
| 424 |
+
"loss": 0.6071,
|
| 425 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 426 |
+
"rewards/chosen": -0.13227801024913788,
|
| 427 |
+
"rewards/margins": 0.22432568669319153,
|
| 428 |
+
"rewards/rejected": -0.3566036820411682,
|
| 429 |
"step": 270
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.5860805860805861,
|
| 433 |
+
"grad_norm": 4.275254606059205,
|
| 434 |
+
"learning_rate": 2.1804923757009882e-07,
|
| 435 |
+
"logits/chosen": -1.275221347808838,
|
| 436 |
+
"logits/rejected": -1.1522352695465088,
|
| 437 |
+
"logps/chosen": -280.47003173828125,
|
| 438 |
+
"logps/rejected": -272.01348876953125,
|
| 439 |
+
"loss": 0.6143,
|
| 440 |
"rewards/accuracies": 0.6625000238418579,
|
| 441 |
+
"rewards/chosen": -0.12610167264938354,
|
| 442 |
+
"rewards/margins": 0.22788269817829132,
|
| 443 |
+
"rewards/rejected": -0.35398435592651367,
|
| 444 |
"step": 280
|
| 445 |
},
|
| 446 |
{
|
| 447 |
"epoch": 0.6070120355834642,
|
| 448 |
+
"grad_norm": 5.423999562212257,
|
| 449 |
+
"learning_rate": 1.9999357655598891e-07,
|
| 450 |
+
"logits/chosen": -1.2647807598114014,
|
| 451 |
+
"logits/rejected": -1.2297732830047607,
|
| 452 |
+
"logps/chosen": -280.0606384277344,
|
| 453 |
+
"logps/rejected": -303.05706787109375,
|
| 454 |
+
"loss": 0.606,
|
| 455 |
+
"rewards/accuracies": 0.706250011920929,
|
| 456 |
+
"rewards/chosen": -0.24480751156806946,
|
| 457 |
+
"rewards/margins": 0.22092871367931366,
|
| 458 |
+
"rewards/rejected": -0.4657362103462219,
|
| 459 |
"step": 290
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.6279434850863422,
|
| 463 |
+
"grad_norm": 4.555810704174,
|
| 464 |
+
"learning_rate": 1.8220596619089573e-07,
|
| 465 |
+
"logits/chosen": -1.3479591608047485,
|
| 466 |
+
"logits/rejected": -1.2260494232177734,
|
| 467 |
+
"logps/chosen": -363.6427307128906,
|
| 468 |
+
"logps/rejected": -332.61077880859375,
|
| 469 |
+
"loss": 0.6015,
|
| 470 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 471 |
+
"rewards/chosen": -0.21298007667064667,
|
| 472 |
+
"rewards/margins": 0.23558974266052246,
|
| 473 |
+
"rewards/rejected": -0.4485698342323303,
|
| 474 |
"step": 300
|
| 475 |
},
|
| 476 |
{
|
| 477 |
"epoch": 0.6488749345892203,
|
| 478 |
+
"grad_norm": 4.921819098251592,
|
| 479 |
+
"learning_rate": 1.647817538357072e-07,
|
| 480 |
+
"logits/chosen": -1.3201913833618164,
|
| 481 |
+
"logits/rejected": -1.2364635467529297,
|
| 482 |
+
"logps/chosen": -336.68865966796875,
|
| 483 |
+
"logps/rejected": -303.8711242675781,
|
| 484 |
+
"loss": 0.5976,
|
| 485 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 486 |
+
"rewards/chosen": -0.18723243474960327,
|
| 487 |
+
"rewards/margins": 0.2934957444667816,
|
| 488 |
+
"rewards/rejected": -0.4807282090187073,
|
| 489 |
"step": 310
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.6698063840920984,
|
| 493 |
+
"grad_norm": 5.4800197473848025,
|
| 494 |
+
"learning_rate": 1.478143389201113e-07,
|
| 495 |
+
"logits/chosen": -1.2567358016967773,
|
| 496 |
+
"logits/rejected": -1.1461724042892456,
|
| 497 |
+
"logps/chosen": -284.4916076660156,
|
| 498 |
+
"logps/rejected": -278.87237548828125,
|
| 499 |
+
"loss": 0.5938,
|
| 500 |
+
"rewards/accuracies": 0.75,
|
| 501 |
+
"rewards/chosen": -0.20529761910438538,
|
| 502 |
+
"rewards/margins": 0.2834058701992035,
|
| 503 |
+
"rewards/rejected": -0.48870348930358887,
|
| 504 |
"step": 320
|
| 505 |
},
|
| 506 |
{
|
| 507 |
"epoch": 0.6907378335949764,
|
| 508 |
+
"grad_norm": 5.489304392569405,
|
| 509 |
+
"learning_rate": 1.3139467229135998e-07,
|
| 510 |
+
"logits/chosen": -1.3123420476913452,
|
| 511 |
+
"logits/rejected": -1.279762625694275,
|
| 512 |
+
"logps/chosen": -300.2234802246094,
|
| 513 |
+
"logps/rejected": -313.7867126464844,
|
| 514 |
+
"loss": 0.6092,
|
| 515 |
+
"rewards/accuracies": 0.706250011920929,
|
| 516 |
+
"rewards/chosen": -0.14779067039489746,
|
| 517 |
+
"rewards/margins": 0.2438393384218216,
|
| 518 |
+
"rewards/rejected": -0.39162999391555786,
|
| 519 |
"step": 330
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.7116692830978545,
|
| 523 |
+
"grad_norm": 7.6710695980682,
|
| 524 |
+
"learning_rate": 1.1561076868822755e-07,
|
| 525 |
+
"logits/chosen": -1.3397032022476196,
|
| 526 |
+
"logits/rejected": -1.2899221181869507,
|
| 527 |
+
"logps/chosen": -317.85455322265625,
|
| 528 |
+
"logps/rejected": -308.87054443359375,
|
| 529 |
+
"loss": 0.5991,
|
| 530 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 531 |
+
"rewards/chosen": -0.12762676179409027,
|
| 532 |
+
"rewards/margins": 0.22320690751075745,
|
| 533 |
+
"rewards/rejected": -0.3508336842060089,
|
| 534 |
"step": 340
|
| 535 |
},
|
| 536 |
{
|
| 537 |
"epoch": 0.7326007326007326,
|
| 538 |
+
"grad_norm": 4.866645044943069,
|
| 539 |
+
"learning_rate": 1.0054723495346482e-07,
|
| 540 |
+
"logits/chosen": -1.2954719066619873,
|
| 541 |
+
"logits/rejected": -1.1627123355865479,
|
| 542 |
+
"logps/chosen": -287.058837890625,
|
| 543 |
+
"logps/rejected": -284.2725524902344,
|
| 544 |
+
"loss": 0.5907,
|
| 545 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 546 |
+
"rewards/chosen": -0.17578962445259094,
|
| 547 |
+
"rewards/margins": 0.2943018078804016,
|
| 548 |
+
"rewards/rejected": -0.47009140253067017,
|
| 549 |
"step": 350
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.7535321821036107,
|
| 553 |
+
"grad_norm": 4.349425498711494,
|
| 554 |
+
"learning_rate": 8.628481651367875e-08,
|
| 555 |
+
"logits/chosen": -1.2945648431777954,
|
| 556 |
+
"logits/rejected": -1.2246530055999756,
|
| 557 |
+
"logps/chosen": -341.4446716308594,
|
| 558 |
+
"logps/rejected": -321.95196533203125,
|
| 559 |
+
"loss": 0.6128,
|
| 560 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 561 |
+
"rewards/chosen": -0.1850513368844986,
|
| 562 |
+
"rewards/margins": 0.2748066782951355,
|
| 563 |
+
"rewards/rejected": -0.4598580300807953,
|
| 564 |
"step": 360
|
| 565 |
},
|
| 566 |
{
|
| 567 |
"epoch": 0.7744636316064888,
|
| 568 |
+
"grad_norm": 4.798940209140212,
|
| 569 |
+
"learning_rate": 7.289996455765748e-08,
|
| 570 |
+
"logits/chosen": -1.2601999044418335,
|
| 571 |
+
"logits/rejected": -1.1612764596939087,
|
| 572 |
+
"logps/chosen": -292.4561767578125,
|
| 573 |
+
"logps/rejected": -281.00152587890625,
|
| 574 |
+
"loss": 0.5979,
|
| 575 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 576 |
+
"rewards/chosen": -0.17593416571617126,
|
| 577 |
+
"rewards/margins": 0.2297048270702362,
|
| 578 |
+
"rewards/rejected": -0.40563899278640747,
|
| 579 |
"step": 370
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.7953950811093669,
|
| 583 |
+
"grad_norm": 6.704056490224638,
|
| 584 |
+
"learning_rate": 6.046442623320145e-08,
|
| 585 |
+
"logits/chosen": -1.3166625499725342,
|
| 586 |
+
"logits/rejected": -1.2426071166992188,
|
| 587 |
+
"logps/chosen": -280.49658203125,
|
| 588 |
+
"logps/rejected": -327.62945556640625,
|
| 589 |
+
"loss": 0.5965,
|
| 590 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 591 |
+
"rewards/chosen": -0.27102166414260864,
|
| 592 |
+
"rewards/margins": 0.3087245225906372,
|
| 593 |
+
"rewards/rejected": -0.5797461271286011,
|
| 594 |
"step": 380
|
| 595 |
},
|
| 596 |
{
|
| 597 |
"epoch": 0.8163265306122449,
|
| 598 |
+
"grad_norm": 4.8192929971991605,
|
| 599 |
+
"learning_rate": 4.904486005914027e-08,
|
| 600 |
+
"logits/chosen": -1.4123605489730835,
|
| 601 |
+
"logits/rejected": -1.334108591079712,
|
| 602 |
+
"logps/chosen": -365.5756530761719,
|
| 603 |
+
"logps/rejected": -360.2679443359375,
|
| 604 |
+
"loss": 0.5969,
|
| 605 |
"rewards/accuracies": 0.731249988079071,
|
| 606 |
+
"rewards/chosen": -0.1903412640094757,
|
| 607 |
+
"rewards/margins": 0.3084810674190521,
|
| 608 |
+
"rewards/rejected": -0.4988223612308502,
|
| 609 |
"step": 390
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.837257980115123,
|
| 613 |
+
"grad_norm": 5.625190550048412,
|
| 614 |
+
"learning_rate": 3.8702478614051345e-08,
|
| 615 |
+
"logits/chosen": -1.3394266366958618,
|
| 616 |
+
"logits/rejected": -1.2082809209823608,
|
| 617 |
+
"logps/chosen": -290.7579040527344,
|
| 618 |
+
"logps/rejected": -297.6111145019531,
|
| 619 |
+
"loss": 0.6015,
|
| 620 |
+
"rewards/accuracies": 0.731249988079071,
|
| 621 |
+
"rewards/chosen": -0.20424337685108185,
|
| 622 |
+
"rewards/margins": 0.30908313393592834,
|
| 623 |
+
"rewards/rejected": -0.5133264660835266,
|
| 624 |
"step": 400
|
| 625 |
},
|
| 626 |
{
|
| 627 |
"epoch": 0.858189429618001,
|
| 628 |
+
"grad_norm": 6.001911832374207,
|
| 629 |
+
"learning_rate": 2.9492720416985e-08,
|
| 630 |
+
"logits/chosen": -1.3413333892822266,
|
| 631 |
+
"logits/rejected": -1.1819798946380615,
|
| 632 |
+
"logps/chosen": -345.768798828125,
|
| 633 |
+
"logps/rejected": -324.0213317871094,
|
| 634 |
+
"loss": 0.6113,
|
| 635 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 636 |
+
"rewards/chosen": -0.2449842393398285,
|
| 637 |
+
"rewards/margins": 0.26237279176712036,
|
| 638 |
+
"rewards/rejected": -0.5073570013046265,
|
| 639 |
"step": 410
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.8791208791208791,
|
| 643 |
+
"grad_norm": 7.098479038891308,
|
| 644 |
+
"learning_rate": 2.1464952759020856e-08,
|
| 645 |
+
"logits/chosen": -1.3054345846176147,
|
| 646 |
+
"logits/rejected": -1.212334394454956,
|
| 647 |
+
"logps/chosen": -300.0458984375,
|
| 648 |
+
"logps/rejected": -334.25799560546875,
|
| 649 |
+
"loss": 0.5909,
|
| 650 |
+
"rewards/accuracies": 0.668749988079071,
|
| 651 |
+
"rewards/chosen": -0.3056102991104126,
|
| 652 |
+
"rewards/margins": 0.30079561471939087,
|
| 653 |
+
"rewards/rejected": -0.6064059138298035,
|
| 654 |
"step": 420
|
| 655 |
},
|
| 656 |
{
|
| 657 |
"epoch": 0.9000523286237572,
|
| 658 |
+
"grad_norm": 5.986649846414336,
|
| 659 |
+
"learning_rate": 1.4662207078575684e-08,
|
| 660 |
+
"logits/chosen": -1.2815589904785156,
|
| 661 |
+
"logits/rejected": -1.230791449546814,
|
| 662 |
+
"logps/chosen": -327.59326171875,
|
| 663 |
+
"logps/rejected": -328.6330261230469,
|
| 664 |
+
"loss": 0.5866,
|
| 665 |
+
"rewards/accuracies": 0.699999988079071,
|
| 666 |
+
"rewards/chosen": -0.2670283317565918,
|
| 667 |
+
"rewards/margins": 0.29276418685913086,
|
| 668 |
+
"rewards/rejected": -0.5597925186157227,
|
| 669 |
"step": 430
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.9209837781266352,
|
| 673 |
+
"grad_norm": 6.667255378058642,
|
| 674 |
+
"learning_rate": 9.12094829893642e-09,
|
| 675 |
+
"logits/chosen": -1.3328527212142944,
|
| 676 |
+
"logits/rejected": -1.2644946575164795,
|
| 677 |
+
"logps/chosen": -298.36688232421875,
|
| 678 |
+
"logps/rejected": -278.701416015625,
|
| 679 |
+
"loss": 0.5928,
|
| 680 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 681 |
+
"rewards/chosen": -0.23760783672332764,
|
| 682 |
+
"rewards/margins": 0.25196802616119385,
|
| 683 |
+
"rewards/rejected": -0.4895758628845215,
|
| 684 |
"step": 440
|
| 685 |
},
|
| 686 |
{
|
| 687 |
"epoch": 0.9419152276295133,
|
| 688 |
+
"grad_norm": 6.876184304010807,
|
| 689 |
+
"learning_rate": 4.8708793644441086e-09,
|
| 690 |
+
"logits/chosen": -1.387279748916626,
|
| 691 |
+
"logits/rejected": -1.2943140268325806,
|
| 692 |
+
"logps/chosen": -305.9947509765625,
|
| 693 |
+
"logps/rejected": -321.4436340332031,
|
| 694 |
+
"loss": 0.5988,
|
| 695 |
+
"rewards/accuracies": 0.699999988079071,
|
| 696 |
+
"rewards/chosen": -0.2561241090297699,
|
| 697 |
+
"rewards/margins": 0.28780630230903625,
|
| 698 |
+
"rewards/rejected": -0.5439304113388062,
|
| 699 |
"step": 450
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 0.9628466771323915,
|
| 703 |
+
"grad_norm": 5.120319319432807,
|
| 704 |
+
"learning_rate": 1.9347820230782295e-09,
|
| 705 |
+
"logits/chosen": -1.3500282764434814,
|
| 706 |
+
"logits/rejected": -1.2091432809829712,
|
| 707 |
+
"logps/chosen": -310.1429443359375,
|
| 708 |
+
"logps/rejected": -289.270263671875,
|
| 709 |
+
"loss": 0.6001,
|
| 710 |
+
"rewards/accuracies": 0.699999988079071,
|
| 711 |
+
"rewards/chosen": -0.25589779019355774,
|
| 712 |
+
"rewards/margins": 0.28265419602394104,
|
| 713 |
+
"rewards/rejected": -0.5385519862174988,
|
| 714 |
"step": 460
|
| 715 |
},
|
| 716 |
{
|
| 717 |
"epoch": 0.9837781266352695,
|
| 718 |
+
"grad_norm": 4.959973504548126,
|
| 719 |
+
"learning_rate": 3.2839470889836627e-10,
|
| 720 |
+
"logits/chosen": -1.307715892791748,
|
| 721 |
+
"logits/rejected": -1.2506450414657593,
|
| 722 |
+
"logps/chosen": -334.43817138671875,
|
| 723 |
+
"logps/rejected": -338.33966064453125,
|
| 724 |
+
"loss": 0.5873,
|
| 725 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 726 |
+
"rewards/chosen": -0.23161163926124573,
|
| 727 |
+
"rewards/margins": 0.2939731180667877,
|
| 728 |
+
"rewards/rejected": -0.5255848169326782,
|
| 729 |
"step": 470
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 0.9984301412872841,
|
| 733 |
+
"eval_logits/chosen": -1.2491042613983154,
|
| 734 |
+
"eval_logits/rejected": -1.1529250144958496,
|
| 735 |
+
"eval_logps/chosen": -301.5587463378906,
|
| 736 |
+
"eval_logps/rejected": -332.839599609375,
|
| 737 |
+
"eval_loss": 0.5996649861335754,
|
| 738 |
+
"eval_rewards/accuracies": 0.70703125,
|
| 739 |
+
"eval_rewards/chosen": -0.20619867742061615,
|
| 740 |
+
"eval_rewards/margins": 0.3225747346878052,
|
| 741 |
+
"eval_rewards/rejected": -0.5287734866142273,
|
| 742 |
+
"eval_runtime": 144.7492,
|
| 743 |
+
"eval_samples_per_second": 13.817,
|
| 744 |
+
"eval_steps_per_second": 0.221,
|
| 745 |
"step": 477
|
| 746 |
},
|
| 747 |
{
|
| 748 |
+
"epoch": 0.9984301412872841,
|
| 749 |
+
"step": 477,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 750 |
"total_flos": 0.0,
|
| 751 |
+
"train_loss": 0.6286907725863986,
|
| 752 |
+
"train_runtime": 7498.289,
|
| 753 |
+
"train_samples_per_second": 8.153,
|
| 754 |
+
"train_steps_per_second": 0.064
|
| 755 |
}
|
| 756 |
],
|
| 757 |
"logging_steps": 10,
|