Model save
Browse files- README.md +2 -2
- all_results.json +5 -5
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- runs/Jun21_03-36-42_n136-100-194/events.out.tfevents.1718912225.n136-100-194.566786.0 +3 -0
- train_results.json +5 -5
- trainer_state.json +231 -561
- training_args.bin +2 -2
README.md
CHANGED
|
@@ -14,7 +14,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 14 |
|
| 15 |
# zephyr-7b-dpo-full
|
| 16 |
|
| 17 |
-
This model is a fine-tuned version of [princeton-nlp/Mistral-7B-Base-SFT-DPO](https://huggingface.co/princeton-nlp/Mistral-7B-Base-SFT-DPO) on
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
@@ -55,5 +55,5 @@ The following hyperparameters were used during training:
|
|
| 55 |
|
| 56 |
- Transformers 4.39.3
|
| 57 |
- Pytorch 2.1.2+cu118
|
| 58 |
-
- Datasets 2.
|
| 59 |
- Tokenizers 0.15.2
|
|
|
|
| 14 |
|
| 15 |
# zephyr-7b-dpo-full
|
| 16 |
|
| 17 |
+
This model is a fine-tuned version of [princeton-nlp/Mistral-7B-Base-SFT-DPO](https://huggingface.co/princeton-nlp/Mistral-7B-Base-SFT-DPO) on the None dataset.
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
|
|
| 55 |
|
| 56 |
- Transformers 4.39.3
|
| 57 |
- Pytorch 2.1.2+cu118
|
| 58 |
+
- Datasets 2.16.1
|
| 59 |
- Tokenizers 0.15.2
|
all_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second": 8.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.37812867454580357,
|
| 4 |
+
"train_runtime": 5319.5814,
|
| 5 |
+
"train_samples": 47302,
|
| 6 |
+
"train_samples_per_second": 8.892,
|
| 7 |
+
"train_steps_per_second": 0.035
|
| 8 |
}
|
model-00001-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4943162336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b0d91deaf39d5239312d2ee38ee390acaa4b4fa405cd4b896c533d8687b80b9
|
| 3 |
size 4943162336
|
model-00002-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4999819336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b806d3480fd4992109c5d61a5d48cf88cca223fbec8796f7b25067bfe4956722
|
| 3 |
size 4999819336
|
model-00003-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4540516344
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a766c0fba83e2189c5e7cdab4f60fcf4245ea877bf5d1a20cb88a89e9fbb76e
|
| 3 |
size 4540516344
|
runs/Jun21_03-36-42_n136-100-194/events.out.tfevents.1718912225.n136-100-194.566786.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0c7a82fa25f3fb898720560ce86ca0146accaa60e9f545355c33bd6e7520be9f
|
| 3 |
+
size 18155
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples":
|
| 6 |
-
"train_samples_per_second": 8.
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.37812867454580357,
|
| 4 |
+
"train_runtime": 5319.5814,
|
| 5 |
+
"train_samples": 47302,
|
| 6 |
+
"train_samples_per_second": 8.892,
|
| 7 |
+
"train_steps_per_second": 0.035
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -3,20 +3,20 @@
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 1.0,
|
| 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.
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"logits/chosen": -0.
|
| 16 |
-
"logits/rejected":
|
| 17 |
-
"logps/chosen": -
|
| 18 |
-
"logps/rejected": -
|
| 19 |
-
"loss": 0.
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
@@ -24,617 +24,287 @@
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"epoch": 0.
|
| 28 |
-
"grad_norm":
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"logits/chosen": 0.
|
| 31 |
-
"logits/rejected": 0.
|
| 32 |
-
"logps/chosen": -
|
| 33 |
-
"logps/rejected": -
|
| 34 |
-
"loss": 0.
|
| 35 |
-
"rewards/accuracies": 0.
|
| 36 |
-
"rewards/chosen": 0.
|
| 37 |
-
"rewards/margins": 0.
|
| 38 |
-
"rewards/rejected": -0.
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
-
"epoch": 0.
|
| 43 |
-
"grad_norm":
|
| 44 |
-
"learning_rate":
|
| 45 |
-
"logits/chosen": 0.
|
| 46 |
-
"logits/rejected": 0.
|
| 47 |
-
"logps/chosen": -
|
| 48 |
-
"logps/rejected": -
|
| 49 |
-
"loss": 0.
|
| 50 |
-
"rewards/accuracies": 0.
|
| 51 |
-
"rewards/chosen": 0.
|
| 52 |
-
"rewards/margins":
|
| 53 |
-
"rewards/rejected": -
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
-
"epoch": 0.
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate":
|
| 60 |
-
"logits/chosen": 0.
|
| 61 |
-
"logits/rejected":
|
| 62 |
-
"logps/chosen": -
|
| 63 |
-
"logps/rejected": -
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"rewards/accuracies": 0.
|
| 66 |
-
"rewards/chosen": 0.
|
| 67 |
-
"rewards/margins":
|
| 68 |
-
"rewards/rejected": -
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
-
"epoch": 0.
|
| 73 |
-
"grad_norm":
|
| 74 |
-
"learning_rate":
|
| 75 |
-
"logits/chosen": -0.
|
| 76 |
-
"logits/rejected":
|
| 77 |
-
"logps/chosen": -
|
| 78 |
-
"logps/rejected": -
|
| 79 |
-
"loss": 0.
|
| 80 |
-
"rewards/accuracies": 0.
|
| 81 |
-
"rewards/chosen":
|
| 82 |
-
"rewards/margins":
|
| 83 |
-
"rewards/rejected": -
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
-
"epoch": 0.
|
| 88 |
-
"grad_norm":
|
| 89 |
-
"learning_rate": 4.
|
| 90 |
-
"logits/chosen": 0.
|
| 91 |
-
"logits/rejected": 0.
|
| 92 |
-
"logps/chosen": -
|
| 93 |
-
"logps/rejected": -
|
| 94 |
-
"loss": 0.
|
| 95 |
-
"rewards/accuracies": 0.
|
| 96 |
-
"rewards/chosen":
|
| 97 |
-
"rewards/margins":
|
| 98 |
-
"rewards/rejected": -
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
-
"epoch": 0.
|
| 103 |
-
"grad_norm":
|
| 104 |
-
"learning_rate": 4.
|
| 105 |
-
"logits/chosen": 0.
|
| 106 |
-
"logits/rejected": 0.
|
| 107 |
-
"logps/chosen": -
|
| 108 |
-
"logps/rejected": -
|
| 109 |
-
"loss": 0.
|
| 110 |
-
"rewards/accuracies": 0.
|
| 111 |
-
"rewards/chosen": 0.
|
| 112 |
-
"rewards/margins":
|
| 113 |
-
"rewards/rejected": -
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"logits/chosen": 0.
|
| 121 |
-
"logits/rejected": 0.
|
| 122 |
-
"logps/chosen": -
|
| 123 |
-
"logps/rejected": -
|
| 124 |
-
"loss": 0.
|
| 125 |
-
"rewards/accuracies": 0.
|
| 126 |
-
"rewards/chosen": 0.
|
| 127 |
-
"rewards/margins":
|
| 128 |
-
"rewards/rejected": -
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
-
"epoch": 0.
|
| 133 |
-
"grad_norm":
|
| 134 |
-
"learning_rate":
|
| 135 |
-
"logits/chosen": 0.
|
| 136 |
-
"logits/rejected": 0.
|
| 137 |
-
"logps/chosen": -
|
| 138 |
-
"logps/rejected": -
|
| 139 |
-
"loss": 0.
|
| 140 |
-
"rewards/accuracies": 0.
|
| 141 |
-
"rewards/chosen": -1.
|
| 142 |
-
"rewards/margins": 5.
|
| 143 |
-
"rewards/rejected": -6.
|
| 144 |
"step": 80
|
| 145 |
},
|
| 146 |
{
|
| 147 |
-
"epoch": 0.
|
| 148 |
-
"grad_norm":
|
| 149 |
-
"learning_rate":
|
| 150 |
-
"logits/chosen": 0.
|
| 151 |
-
"logits/rejected": 0.
|
| 152 |
-
"logps/chosen": -
|
| 153 |
-
"logps/rejected": -
|
| 154 |
-
"loss": 0.
|
| 155 |
-
"rewards/accuracies": 0.
|
| 156 |
-
"rewards/chosen": 0.
|
| 157 |
-
"rewards/margins": 5.
|
| 158 |
-
"rewards/rejected": -
|
| 159 |
"step": 90
|
| 160 |
},
|
| 161 |
{
|
| 162 |
-
"epoch": 0.
|
| 163 |
-
"grad_norm":
|
| 164 |
-
"learning_rate":
|
| 165 |
-
"logits/chosen": 0.
|
| 166 |
-
"logits/rejected": 0.
|
| 167 |
-
"logps/chosen": -
|
| 168 |
-
"logps/rejected": -
|
| 169 |
-
"loss": 0.
|
| 170 |
-
"rewards/accuracies": 0.
|
| 171 |
-
"rewards/chosen": -
|
| 172 |
-
"rewards/margins":
|
| 173 |
-
"rewards/rejected": -6.
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
-
"epoch": 0.
|
| 178 |
-
"grad_norm":
|
| 179 |
-
"learning_rate":
|
| 180 |
-
"logits/chosen": 0.
|
| 181 |
-
"logits/rejected": 0.
|
| 182 |
-
"logps/chosen": -
|
| 183 |
-
"logps/rejected": -
|
| 184 |
-
"loss": 0.
|
| 185 |
-
"rewards/accuracies": 0.
|
| 186 |
-
"rewards/chosen": -
|
| 187 |
-
"rewards/margins": 5.
|
| 188 |
-
"rewards/rejected": -
|
| 189 |
"step": 110
|
| 190 |
},
|
| 191 |
{
|
| 192 |
-
"epoch": 0.
|
| 193 |
-
"grad_norm":
|
| 194 |
-
"learning_rate":
|
| 195 |
-
"logits/chosen": 0.
|
| 196 |
-
"logits/rejected": 0.
|
| 197 |
-
"logps/chosen": -
|
| 198 |
-
"logps/rejected": -
|
| 199 |
-
"loss": 0.
|
| 200 |
-
"rewards/accuracies": 0.
|
| 201 |
-
"rewards/chosen": -0.
|
| 202 |
-
"rewards/margins": 5.
|
| 203 |
-
"rewards/rejected": -5.
|
| 204 |
"step": 120
|
| 205 |
},
|
| 206 |
{
|
| 207 |
-
"epoch": 0.
|
| 208 |
-
"grad_norm":
|
| 209 |
-
"learning_rate":
|
| 210 |
-
"logits/chosen": 0.
|
| 211 |
-
"logits/rejected": 0.
|
| 212 |
-
"logps/chosen": -
|
| 213 |
-
"logps/rejected": -
|
| 214 |
-
"loss": 0.
|
| 215 |
-
"rewards/accuracies": 0.
|
| 216 |
-
"rewards/chosen": -0.
|
| 217 |
-
"rewards/margins": 5.
|
| 218 |
-
"rewards/rejected": -
|
| 219 |
"step": 130
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"grad_norm":
|
| 224 |
-
"learning_rate":
|
| 225 |
-
"logits/chosen": 0.
|
| 226 |
-
"logits/rejected": 0.
|
| 227 |
-
"logps/chosen": -
|
| 228 |
-
"logps/rejected": -
|
| 229 |
-
"loss": 0.
|
| 230 |
-
"rewards/accuracies": 0.
|
| 231 |
-
"rewards/chosen": -0.
|
| 232 |
-
"rewards/margins": 5.
|
| 233 |
-
"rewards/rejected": -
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"grad_norm":
|
| 239 |
-
"learning_rate":
|
| 240 |
-
"logits/chosen": -0.
|
| 241 |
-
"logits/rejected": 0.
|
| 242 |
-
"logps/chosen": -
|
| 243 |
-
"logps/rejected": -
|
| 244 |
-
"loss": 0.
|
| 245 |
-
"rewards/accuracies": 0.
|
| 246 |
-
"rewards/chosen":
|
| 247 |
-
"rewards/margins":
|
| 248 |
-
"rewards/rejected": -
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
-
"epoch": 0.
|
| 253 |
-
"grad_norm":
|
| 254 |
-
"learning_rate":
|
| 255 |
-
"logits/chosen": 0.
|
| 256 |
-
"logits/rejected":
|
| 257 |
-
"logps/chosen": -
|
| 258 |
-
"logps/rejected": -
|
| 259 |
-
"loss": 0.
|
| 260 |
-
"rewards/accuracies": 0.
|
| 261 |
-
"rewards/chosen":
|
| 262 |
-
"rewards/margins":
|
| 263 |
-
"rewards/rejected": -5.
|
| 264 |
"step": 160
|
| 265 |
},
|
| 266 |
{
|
| 267 |
-
"epoch": 0.
|
| 268 |
-
"grad_norm":
|
| 269 |
-
"learning_rate":
|
| 270 |
-
"logits/chosen": -0.
|
| 271 |
-
"logits/rejected": 0.
|
| 272 |
-
"logps/chosen": -
|
| 273 |
-
"logps/rejected": -
|
| 274 |
-
"loss": 0.
|
| 275 |
-
"rewards/accuracies": 0.
|
| 276 |
-
"rewards/chosen": -0.
|
| 277 |
-
"rewards/margins": 5.
|
| 278 |
-
"rewards/rejected": -6.
|
| 279 |
"step": 170
|
| 280 |
},
|
| 281 |
-
{
|
| 282 |
-
"epoch": 0.45,
|
| 283 |
-
"grad_norm": 55.57273036697954,
|
| 284 |
-
"learning_rate": 3.355050358314172e-07,
|
| 285 |
-
"logits/chosen": -0.17330403625965118,
|
| 286 |
-
"logits/rejected": -0.37704771757125854,
|
| 287 |
-
"logps/chosen": -335.1918029785156,
|
| 288 |
-
"logps/rejected": -303.3643493652344,
|
| 289 |
-
"loss": 0.1893,
|
| 290 |
-
"rewards/accuracies": 0.940625011920929,
|
| 291 |
-
"rewards/chosen": 0.3236238360404968,
|
| 292 |
-
"rewards/margins": 5.686396598815918,
|
| 293 |
-
"rewards/rejected": -5.3627729415893555,
|
| 294 |
-
"step": 180
|
| 295 |
-
},
|
| 296 |
-
{
|
| 297 |
-
"epoch": 0.47,
|
| 298 |
-
"grad_norm": 34.4619211960604,
|
| 299 |
-
"learning_rate": 3.147047612756302e-07,
|
| 300 |
-
"logits/chosen": 0.051471106708049774,
|
| 301 |
-
"logits/rejected": 0.08085541427135468,
|
| 302 |
-
"logps/chosen": -341.36932373046875,
|
| 303 |
-
"logps/rejected": -306.84906005859375,
|
| 304 |
-
"loss": 0.1714,
|
| 305 |
-
"rewards/accuracies": 0.9312499761581421,
|
| 306 |
-
"rewards/chosen": -0.4273417592048645,
|
| 307 |
-
"rewards/margins": 5.435866355895996,
|
| 308 |
-
"rewards/rejected": -5.863208293914795,
|
| 309 |
-
"step": 190
|
| 310 |
-
},
|
| 311 |
-
{
|
| 312 |
-
"epoch": 0.5,
|
| 313 |
-
"grad_norm": 56.25034316469711,
|
| 314 |
-
"learning_rate": 2.934120444167326e-07,
|
| 315 |
-
"logits/chosen": 0.2834271490573883,
|
| 316 |
-
"logits/rejected": 0.4803285002708435,
|
| 317 |
-
"logps/chosen": -320.8459777832031,
|
| 318 |
-
"logps/rejected": -293.8531799316406,
|
| 319 |
-
"loss": 0.1797,
|
| 320 |
-
"rewards/accuracies": 0.9125000238418579,
|
| 321 |
-
"rewards/chosen": -1.590319275856018,
|
| 322 |
-
"rewards/margins": 5.1105732917785645,
|
| 323 |
-
"rewards/rejected": -6.700892448425293,
|
| 324 |
-
"step": 200
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 0.53,
|
| 328 |
-
"grad_norm": 42.616613118118565,
|
| 329 |
-
"learning_rate": 2.717889356869146e-07,
|
| 330 |
-
"logits/chosen": -0.1653163880109787,
|
| 331 |
-
"logits/rejected": -0.19997279345989227,
|
| 332 |
-
"logps/chosen": -297.6966857910156,
|
| 333 |
-
"logps/rejected": -281.0061950683594,
|
| 334 |
-
"loss": 0.1905,
|
| 335 |
-
"rewards/accuracies": 0.921875,
|
| 336 |
-
"rewards/chosen": 0.32219845056533813,
|
| 337 |
-
"rewards/margins": 5.068012237548828,
|
| 338 |
-
"rewards/rejected": -4.745813846588135,
|
| 339 |
-
"step": 210
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.55,
|
| 343 |
-
"grad_norm": 54.94298294558815,
|
| 344 |
-
"learning_rate": 2.5e-07,
|
| 345 |
-
"logits/chosen": 0.11906982958316803,
|
| 346 |
-
"logits/rejected": 0.16367843747138977,
|
| 347 |
-
"logps/chosen": -322.9801330566406,
|
| 348 |
-
"logps/rejected": -289.7718811035156,
|
| 349 |
-
"loss": 0.2379,
|
| 350 |
-
"rewards/accuracies": 0.9312499761581421,
|
| 351 |
-
"rewards/chosen": -0.8957729339599609,
|
| 352 |
-
"rewards/margins": 4.5600104331970215,
|
| 353 |
-
"rewards/rejected": -5.455783367156982,
|
| 354 |
-
"step": 220
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 0.57,
|
| 358 |
-
"grad_norm": 45.798677109719435,
|
| 359 |
-
"learning_rate": 2.2821106431308543e-07,
|
| 360 |
-
"logits/chosen": 0.26459816098213196,
|
| 361 |
-
"logits/rejected": 0.37918907403945923,
|
| 362 |
-
"logps/chosen": -308.4639892578125,
|
| 363 |
-
"logps/rejected": -284.7757873535156,
|
| 364 |
-
"loss": 0.1974,
|
| 365 |
-
"rewards/accuracies": 0.90625,
|
| 366 |
-
"rewards/chosen": -0.7637487649917603,
|
| 367 |
-
"rewards/margins": 5.488121032714844,
|
| 368 |
-
"rewards/rejected": -6.251870155334473,
|
| 369 |
-
"step": 230
|
| 370 |
-
},
|
| 371 |
-
{
|
| 372 |
-
"epoch": 0.6,
|
| 373 |
-
"grad_norm": 71.171077628863,
|
| 374 |
-
"learning_rate": 2.065879555832674e-07,
|
| 375 |
-
"logits/chosen": 0.11944101750850677,
|
| 376 |
-
"logits/rejected": 0.10174547135829926,
|
| 377 |
-
"logps/chosen": -317.2291564941406,
|
| 378 |
-
"logps/rejected": -301.72357177734375,
|
| 379 |
-
"loss": 0.1761,
|
| 380 |
-
"rewards/accuracies": 0.9281250238418579,
|
| 381 |
-
"rewards/chosen": -0.4725034832954407,
|
| 382 |
-
"rewards/margins": 5.045371055603027,
|
| 383 |
-
"rewards/rejected": -5.517874717712402,
|
| 384 |
-
"step": 240
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 0.62,
|
| 388 |
-
"grad_norm": 58.77603788283333,
|
| 389 |
-
"learning_rate": 1.8529523872436977e-07,
|
| 390 |
-
"logits/chosen": -0.0030237496830523014,
|
| 391 |
-
"logits/rejected": 0.24394559860229492,
|
| 392 |
-
"logps/chosen": -337.0401306152344,
|
| 393 |
-
"logps/rejected": -304.41485595703125,
|
| 394 |
-
"loss": 0.2199,
|
| 395 |
-
"rewards/accuracies": 0.90625,
|
| 396 |
-
"rewards/chosen": -1.1832717657089233,
|
| 397 |
-
"rewards/margins": 4.3705525398254395,
|
| 398 |
-
"rewards/rejected": -5.553823947906494,
|
| 399 |
-
"step": 250
|
| 400 |
-
},
|
| 401 |
-
{
|
| 402 |
-
"epoch": 0.65,
|
| 403 |
-
"grad_norm": 58.549851578658085,
|
| 404 |
-
"learning_rate": 1.6449496416858282e-07,
|
| 405 |
-
"logits/chosen": 0.2534635663032532,
|
| 406 |
-
"logits/rejected": 0.33454760909080505,
|
| 407 |
-
"logps/chosen": -322.7724609375,
|
| 408 |
-
"logps/rejected": -291.37420654296875,
|
| 409 |
-
"loss": 0.1828,
|
| 410 |
-
"rewards/accuracies": 0.9437500238418579,
|
| 411 |
-
"rewards/chosen": -1.3282363414764404,
|
| 412 |
-
"rewards/margins": 5.129273414611816,
|
| 413 |
-
"rewards/rejected": -6.457509517669678,
|
| 414 |
-
"step": 260
|
| 415 |
-
},
|
| 416 |
-
{
|
| 417 |
-
"epoch": 0.68,
|
| 418 |
-
"grad_norm": 61.69689329910415,
|
| 419 |
-
"learning_rate": 1.4434543456482518e-07,
|
| 420 |
-
"logits/chosen": 0.16627629101276398,
|
| 421 |
-
"logits/rejected": 0.17112889885902405,
|
| 422 |
-
"logps/chosen": -328.3280944824219,
|
| 423 |
-
"logps/rejected": -302.532958984375,
|
| 424 |
-
"loss": 0.1701,
|
| 425 |
-
"rewards/accuracies": 0.918749988079071,
|
| 426 |
-
"rewards/chosen": -0.8465593457221985,
|
| 427 |
-
"rewards/margins": 5.068365573883057,
|
| 428 |
-
"rewards/rejected": -5.914924621582031,
|
| 429 |
-
"step": 270
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 0.7,
|
| 433 |
-
"grad_norm": 82.28572950186891,
|
| 434 |
-
"learning_rate": 1.2500000000000005e-07,
|
| 435 |
-
"logits/chosen": 0.07045526802539825,
|
| 436 |
-
"logits/rejected": 0.055424489080905914,
|
| 437 |
-
"logps/chosen": -314.9088134765625,
|
| 438 |
-
"logps/rejected": -283.484130859375,
|
| 439 |
-
"loss": 0.1988,
|
| 440 |
-
"rewards/accuracies": 0.934374988079071,
|
| 441 |
-
"rewards/chosen": -0.05264568328857422,
|
| 442 |
-
"rewards/margins": 5.250518798828125,
|
| 443 |
-
"rewards/rejected": -5.303164958953857,
|
| 444 |
-
"step": 280
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.72,
|
| 448 |
-
"grad_norm": 73.90506249227809,
|
| 449 |
-
"learning_rate": 1.0660589091223854e-07,
|
| 450 |
-
"logits/chosen": 0.08644680678844452,
|
| 451 |
-
"logits/rejected": 0.3770992159843445,
|
| 452 |
-
"logps/chosen": -334.16839599609375,
|
| 453 |
-
"logps/rejected": -295.61138916015625,
|
| 454 |
-
"loss": 0.1816,
|
| 455 |
-
"rewards/accuracies": 0.9312499761581421,
|
| 456 |
-
"rewards/chosen": -0.6267386674880981,
|
| 457 |
-
"rewards/margins": 5.2511420249938965,
|
| 458 |
-
"rewards/rejected": -5.877881050109863,
|
| 459 |
-
"step": 290
|
| 460 |
-
},
|
| 461 |
-
{
|
| 462 |
-
"epoch": 0.75,
|
| 463 |
-
"grad_norm": 55.979754279546064,
|
| 464 |
-
"learning_rate": 8.930309757836516e-08,
|
| 465 |
-
"logits/chosen": 0.0685787945985794,
|
| 466 |
-
"logits/rejected": 0.4188007414340973,
|
| 467 |
-
"logps/chosen": -330.8103332519531,
|
| 468 |
-
"logps/rejected": -297.2096252441406,
|
| 469 |
-
"loss": 0.1831,
|
| 470 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 471 |
-
"rewards/chosen": -0.8838691711425781,
|
| 472 |
-
"rewards/margins": 4.987154960632324,
|
| 473 |
-
"rewards/rejected": -5.871024131774902,
|
| 474 |
-
"step": 300
|
| 475 |
-
},
|
| 476 |
-
{
|
| 477 |
-
"epoch": 0.78,
|
| 478 |
-
"grad_norm": 50.80261213037778,
|
| 479 |
-
"learning_rate": 7.322330470336313e-08,
|
| 480 |
-
"logits/chosen": 0.22002212703227997,
|
| 481 |
-
"logits/rejected": 0.2094310224056244,
|
| 482 |
-
"logps/chosen": -321.89739990234375,
|
| 483 |
-
"logps/rejected": -294.4070739746094,
|
| 484 |
-
"loss": 0.1741,
|
| 485 |
-
"rewards/accuracies": 0.918749988079071,
|
| 486 |
-
"rewards/chosen": -0.5131534337997437,
|
| 487 |
-
"rewards/margins": 5.067374229431152,
|
| 488 |
-
"rewards/rejected": -5.580528259277344,
|
| 489 |
-
"step": 310
|
| 490 |
-
},
|
| 491 |
-
{
|
| 492 |
-
"epoch": 0.8,
|
| 493 |
-
"grad_norm": 62.71796095386463,
|
| 494 |
-
"learning_rate": 5.848888922025552e-08,
|
| 495 |
-
"logits/chosen": 0.0758807510137558,
|
| 496 |
-
"logits/rejected": 0.19501206278800964,
|
| 497 |
-
"logps/chosen": -337.13995361328125,
|
| 498 |
-
"logps/rejected": -296.71954345703125,
|
| 499 |
-
"loss": 0.1841,
|
| 500 |
-
"rewards/accuracies": 0.953125,
|
| 501 |
-
"rewards/chosen": -0.4793934226036072,
|
| 502 |
-
"rewards/margins": 5.182127952575684,
|
| 503 |
-
"rewards/rejected": -5.661521911621094,
|
| 504 |
-
"step": 320
|
| 505 |
-
},
|
| 506 |
-
{
|
| 507 |
-
"epoch": 0.82,
|
| 508 |
-
"grad_norm": 55.862146356475115,
|
| 509 |
-
"learning_rate": 4.521198892775202e-08,
|
| 510 |
-
"logits/chosen": 0.08274734020233154,
|
| 511 |
-
"logits/rejected": -0.0012732266914099455,
|
| 512 |
-
"logps/chosen": -333.7369079589844,
|
| 513 |
-
"logps/rejected": -296.9526062011719,
|
| 514 |
-
"loss": 0.1768,
|
| 515 |
-
"rewards/accuracies": 0.925000011920929,
|
| 516 |
-
"rewards/chosen": -0.4573189616203308,
|
| 517 |
-
"rewards/margins": 5.402385711669922,
|
| 518 |
-
"rewards/rejected": -5.859705448150635,
|
| 519 |
-
"step": 330
|
| 520 |
-
},
|
| 521 |
-
{
|
| 522 |
-
"epoch": 0.85,
|
| 523 |
-
"grad_norm": 78.6552872883194,
|
| 524 |
-
"learning_rate": 3.349364905389032e-08,
|
| 525 |
-
"logits/chosen": -0.16344629228115082,
|
| 526 |
-
"logits/rejected": 0.0033723146189004183,
|
| 527 |
-
"logps/chosen": -322.8833312988281,
|
| 528 |
-
"logps/rejected": -298.59588623046875,
|
| 529 |
-
"loss": 0.1682,
|
| 530 |
-
"rewards/accuracies": 0.8999999761581421,
|
| 531 |
-
"rewards/chosen": -0.6776003837585449,
|
| 532 |
-
"rewards/margins": 5.152725696563721,
|
| 533 |
-
"rewards/rejected": -5.830325126647949,
|
| 534 |
-
"step": 340
|
| 535 |
-
},
|
| 536 |
-
{
|
| 537 |
-
"epoch": 0.88,
|
| 538 |
-
"grad_norm": 57.94825784387087,
|
| 539 |
-
"learning_rate": 2.3423053240837514e-08,
|
| 540 |
-
"logits/chosen": 0.06374481320381165,
|
| 541 |
-
"logits/rejected": 0.13793043792247772,
|
| 542 |
-
"logps/chosen": -319.5912780761719,
|
| 543 |
-
"logps/rejected": -297.6089172363281,
|
| 544 |
-
"loss": 0.1912,
|
| 545 |
-
"rewards/accuracies": 0.903124988079071,
|
| 546 |
-
"rewards/chosen": -0.5590634346008301,
|
| 547 |
-
"rewards/margins": 5.25864315032959,
|
| 548 |
-
"rewards/rejected": -5.81770658493042,
|
| 549 |
-
"step": 350
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.9,
|
| 553 |
-
"grad_norm": 48.20871737390797,
|
| 554 |
-
"learning_rate": 1.507684480352292e-08,
|
| 555 |
-
"logits/chosen": 0.1408473700284958,
|
| 556 |
-
"logits/rejected": 0.24622826278209686,
|
| 557 |
-
"logps/chosen": -323.8467712402344,
|
| 558 |
-
"logps/rejected": -301.0394592285156,
|
| 559 |
-
"loss": 0.1714,
|
| 560 |
-
"rewards/accuracies": 0.9312499761581421,
|
| 561 |
-
"rewards/chosen": -0.5114877820014954,
|
| 562 |
-
"rewards/margins": 5.479299068450928,
|
| 563 |
-
"rewards/rejected": -5.990786552429199,
|
| 564 |
-
"step": 360
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 0.93,
|
| 568 |
-
"grad_norm": 66.71686397462808,
|
| 569 |
-
"learning_rate": 8.518543427732949e-09,
|
| 570 |
-
"logits/chosen": -0.14612798392772675,
|
| 571 |
-
"logits/rejected": -0.002347910311073065,
|
| 572 |
-
"logps/chosen": -305.1243896484375,
|
| 573 |
-
"logps/rejected": -282.12835693359375,
|
| 574 |
-
"loss": 0.1673,
|
| 575 |
-
"rewards/accuracies": 0.921875,
|
| 576 |
-
"rewards/chosen": -0.6524958610534668,
|
| 577 |
-
"rewards/margins": 5.196988105773926,
|
| 578 |
-
"rewards/rejected": -5.849484443664551,
|
| 579 |
-
"step": 370
|
| 580 |
-
},
|
| 581 |
-
{
|
| 582 |
-
"epoch": 0.95,
|
| 583 |
-
"grad_norm": 61.291842343672215,
|
| 584 |
-
"learning_rate": 3.798061746947995e-09,
|
| 585 |
-
"logits/chosen": 0.044129520654678345,
|
| 586 |
-
"logits/rejected": 0.14980553090572357,
|
| 587 |
-
"logps/chosen": -319.0976257324219,
|
| 588 |
-
"logps/rejected": -283.139892578125,
|
| 589 |
-
"loss": 0.1714,
|
| 590 |
-
"rewards/accuracies": 0.940625011920929,
|
| 591 |
-
"rewards/chosen": -0.5317240953445435,
|
| 592 |
-
"rewards/margins": 5.177973747253418,
|
| 593 |
-
"rewards/rejected": -5.70969820022583,
|
| 594 |
-
"step": 380
|
| 595 |
-
},
|
| 596 |
{
|
| 597 |
"epoch": 0.97,
|
| 598 |
-
"grad_norm":
|
| 599 |
-
"learning_rate":
|
| 600 |
-
"logits/chosen": -0.
|
| 601 |
-
"logits/rejected": 0.
|
| 602 |
-
"logps/chosen": -
|
| 603 |
-
"logps/rejected": -
|
| 604 |
-
"loss": 0.
|
| 605 |
-
"rewards/accuracies": 0.
|
| 606 |
-
"rewards/chosen": -0.
|
| 607 |
-
"rewards/margins": 5.
|
| 608 |
-
"rewards/rejected": -
|
| 609 |
-
"step":
|
| 610 |
-
},
|
| 611 |
-
{
|
| 612 |
-
"epoch": 1.0,
|
| 613 |
-
"grad_norm": 40.78979544232789,
|
| 614 |
-
"learning_rate": 0.0,
|
| 615 |
-
"logits/chosen": 0.030362462624907494,
|
| 616 |
-
"logits/rejected": 0.237002894282341,
|
| 617 |
-
"logps/chosen": -320.5105895996094,
|
| 618 |
-
"logps/rejected": -285.7564697265625,
|
| 619 |
-
"loss": 0.1703,
|
| 620 |
-
"rewards/accuracies": 0.9156249761581421,
|
| 621 |
-
"rewards/chosen": -0.8242006301879883,
|
| 622 |
-
"rewards/margins": 5.20082950592041,
|
| 623 |
-
"rewards/rejected": -6.025030136108398,
|
| 624 |
-
"step": 400
|
| 625 |
},
|
| 626 |
{
|
| 627 |
"epoch": 1.0,
|
| 628 |
-
"step":
|
| 629 |
"total_flos": 0.0,
|
| 630 |
-
"train_loss": 0.
|
| 631 |
-
"train_runtime":
|
| 632 |
-
"train_samples_per_second": 8.
|
| 633 |
-
"train_steps_per_second": 0.
|
| 634 |
}
|
| 635 |
],
|
| 636 |
"logging_steps": 10,
|
| 637 |
-
"max_steps":
|
| 638 |
"num_input_tokens_seen": 0,
|
| 639 |
"num_train_epochs": 1,
|
| 640 |
"save_steps": 100,
|
|
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 1.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 185,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.01,
|
| 13 |
+
"grad_norm": 518.0837836920463,
|
| 14 |
+
"learning_rate": 2.6315789473684208e-08,
|
| 15 |
+
"logits/chosen": -0.1266070306301117,
|
| 16 |
+
"logits/rejected": 0.7204304933547974,
|
| 17 |
+
"logps/chosen": -319.01666259765625,
|
| 18 |
+
"logps/rejected": -252.47039794921875,
|
| 19 |
+
"loss": 0.6957,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.05,
|
| 28 |
+
"grad_norm": 455.8868449558538,
|
| 29 |
+
"learning_rate": 2.631578947368421e-07,
|
| 30 |
+
"logits/chosen": -0.38653168082237244,
|
| 31 |
+
"logits/rejected": 0.3361072242259979,
|
| 32 |
+
"logps/chosen": -266.4560546875,
|
| 33 |
+
"logps/rejected": -224.02757263183594,
|
| 34 |
+
"loss": 0.6557,
|
| 35 |
+
"rewards/accuracies": 0.53125,
|
| 36 |
+
"rewards/chosen": -0.08313964307308197,
|
| 37 |
+
"rewards/margins": 0.08996326476335526,
|
| 38 |
+
"rewards/rejected": -0.17310291528701782,
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
+
"epoch": 0.11,
|
| 43 |
+
"grad_norm": 261.73889640243516,
|
| 44 |
+
"learning_rate": 4.999552306674344e-07,
|
| 45 |
+
"logits/chosen": -0.3945041298866272,
|
| 46 |
+
"logits/rejected": 0.5968992114067078,
|
| 47 |
+
"logps/chosen": -283.4828796386719,
|
| 48 |
+
"logps/rejected": -241.41641235351562,
|
| 49 |
+
"loss": 0.4425,
|
| 50 |
+
"rewards/accuracies": 0.8218749761581421,
|
| 51 |
+
"rewards/chosen": 0.35899922251701355,
|
| 52 |
+
"rewards/margins": 2.6830153465270996,
|
| 53 |
+
"rewards/rejected": -2.3240160942077637,
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
+
"epoch": 0.16,
|
| 58 |
+
"grad_norm": 261.2394624512226,
|
| 59 |
+
"learning_rate": 4.946022852363932e-07,
|
| 60 |
+
"logits/chosen": -0.47767549753189087,
|
| 61 |
+
"logits/rejected": 0.3392488360404968,
|
| 62 |
+
"logps/chosen": -265.13568115234375,
|
| 63 |
+
"logps/rejected": -243.9243927001953,
|
| 64 |
+
"loss": 0.462,
|
| 65 |
+
"rewards/accuracies": 0.815625011920929,
|
| 66 |
+
"rewards/chosen": -0.15065696835517883,
|
| 67 |
+
"rewards/margins": 5.174683094024658,
|
| 68 |
+
"rewards/rejected": -5.3253397941589355,
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
+
"epoch": 0.22,
|
| 73 |
+
"grad_norm": 310.52802941610554,
|
| 74 |
+
"learning_rate": 4.805146507594034e-07,
|
| 75 |
+
"logits/chosen": -0.5098148584365845,
|
| 76 |
+
"logits/rejected": 0.3751198649406433,
|
| 77 |
+
"logps/chosen": -261.0537109375,
|
| 78 |
+
"logps/rejected": -233.42532348632812,
|
| 79 |
+
"loss": 0.4377,
|
| 80 |
+
"rewards/accuracies": 0.862500011920929,
|
| 81 |
+
"rewards/chosen": -0.6695741415023804,
|
| 82 |
+
"rewards/margins": 6.071677207946777,
|
| 83 |
+
"rewards/rejected": -6.7412519454956055,
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
+
"epoch": 0.27,
|
| 88 |
+
"grad_norm": 313.59929050730216,
|
| 89 |
+
"learning_rate": 4.581953932909403e-07,
|
| 90 |
+
"logits/chosen": -0.5670371055603027,
|
| 91 |
+
"logits/rejected": 0.24361911416053772,
|
| 92 |
+
"logps/chosen": -281.9254150390625,
|
| 93 |
+
"logps/rejected": -255.1023712158203,
|
| 94 |
+
"loss": 0.3616,
|
| 95 |
+
"rewards/accuracies": 0.871874988079071,
|
| 96 |
+
"rewards/chosen": 0.5183612108230591,
|
| 97 |
+
"rewards/margins": 6.197042465209961,
|
| 98 |
+
"rewards/rejected": -5.678681373596191,
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
+
"epoch": 0.32,
|
| 103 |
+
"grad_norm": 266.5544579622121,
|
| 104 |
+
"learning_rate": 4.284415281717847e-07,
|
| 105 |
+
"logits/chosen": -0.47226476669311523,
|
| 106 |
+
"logits/rejected": 0.4281126856803894,
|
| 107 |
+
"logps/chosen": -278.83251953125,
|
| 108 |
+
"logps/rejected": -258.22943115234375,
|
| 109 |
+
"loss": 0.3576,
|
| 110 |
+
"rewards/accuracies": 0.878125011920929,
|
| 111 |
+
"rewards/chosen": -0.19791364669799805,
|
| 112 |
+
"rewards/margins": 6.015913486480713,
|
| 113 |
+
"rewards/rejected": -6.213827133178711,
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.38,
|
| 118 |
+
"grad_norm": 302.58453941602346,
|
| 119 |
+
"learning_rate": 3.923155588020165e-07,
|
| 120 |
+
"logits/chosen": -0.5101458430290222,
|
| 121 |
+
"logits/rejected": 0.430727481842041,
|
| 122 |
+
"logps/chosen": -249.14736938476562,
|
| 123 |
+
"logps/rejected": -227.1038055419922,
|
| 124 |
+
"loss": 0.3598,
|
| 125 |
+
"rewards/accuracies": 0.871874988079071,
|
| 126 |
+
"rewards/chosen": 0.11757852882146835,
|
| 127 |
+
"rewards/margins": 5.637847900390625,
|
| 128 |
+
"rewards/rejected": -5.520269870758057,
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
+
"epoch": 0.43,
|
| 133 |
+
"grad_norm": 307.7393549252261,
|
| 134 |
+
"learning_rate": 3.511075348989692e-07,
|
| 135 |
+
"logits/chosen": -0.28342846035957336,
|
| 136 |
+
"logits/rejected": 0.524147629737854,
|
| 137 |
+
"logps/chosen": -275.883544921875,
|
| 138 |
+
"logps/rejected": -249.2301788330078,
|
| 139 |
+
"loss": 0.4166,
|
| 140 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 141 |
+
"rewards/chosen": -1.1018508672714233,
|
| 142 |
+
"rewards/margins": 5.499217510223389,
|
| 143 |
+
"rewards/rejected": -6.601068019866943,
|
| 144 |
"step": 80
|
| 145 |
},
|
| 146 |
{
|
| 147 |
+
"epoch": 0.49,
|
| 148 |
+
"grad_norm": 307.7617928649449,
|
| 149 |
+
"learning_rate": 3.062889851306735e-07,
|
| 150 |
+
"logits/chosen": -0.23101525008678436,
|
| 151 |
+
"logits/rejected": 0.594648003578186,
|
| 152 |
+
"logps/chosen": -263.6214294433594,
|
| 153 |
+
"logps/rejected": -237.5282440185547,
|
| 154 |
+
"loss": 0.3926,
|
| 155 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 156 |
+
"rewards/chosen": -0.46620288491249084,
|
| 157 |
+
"rewards/margins": 5.611274719238281,
|
| 158 |
+
"rewards/rejected": -6.077476978302002,
|
| 159 |
"step": 90
|
| 160 |
},
|
| 161 |
{
|
| 162 |
+
"epoch": 0.54,
|
| 163 |
+
"grad_norm": 263.0128253752098,
|
| 164 |
+
"learning_rate": 2.594603691794176e-07,
|
| 165 |
+
"logits/chosen": -0.35276657342910767,
|
| 166 |
+
"logits/rejected": 0.5542042851448059,
|
| 167 |
+
"logps/chosen": -268.01763916015625,
|
| 168 |
+
"logps/rejected": -240.2915496826172,
|
| 169 |
+
"loss": 0.3427,
|
| 170 |
+
"rewards/accuracies": 0.831250011920929,
|
| 171 |
+
"rewards/chosen": -0.9488876461982727,
|
| 172 |
+
"rewards/margins": 5.2973761558532715,
|
| 173 |
+
"rewards/rejected": -6.246264457702637,
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
+
"epoch": 0.59,
|
| 178 |
+
"grad_norm": 321.2698886355005,
|
| 179 |
+
"learning_rate": 2.1229392570965654e-07,
|
| 180 |
+
"logits/chosen": 0.03777565434575081,
|
| 181 |
+
"logits/rejected": 0.7461130023002625,
|
| 182 |
+
"logps/chosen": -262.41558837890625,
|
| 183 |
+
"logps/rejected": -246.7896270751953,
|
| 184 |
+
"loss": 0.3086,
|
| 185 |
+
"rewards/accuracies": 0.871874988079071,
|
| 186 |
+
"rewards/chosen": -1.0481889247894287,
|
| 187 |
+
"rewards/margins": 5.13235330581665,
|
| 188 |
+
"rewards/rejected": -6.180542945861816,
|
| 189 |
"step": 110
|
| 190 |
},
|
| 191 |
{
|
| 192 |
+
"epoch": 0.65,
|
| 193 |
+
"grad_norm": 257.6145945098329,
|
| 194 |
+
"learning_rate": 1.6647395712565254e-07,
|
| 195 |
+
"logits/chosen": -0.36657968163490295,
|
| 196 |
+
"logits/rejected": 0.6054766774177551,
|
| 197 |
+
"logps/chosen": -276.4320373535156,
|
| 198 |
+
"logps/rejected": -247.306640625,
|
| 199 |
+
"loss": 0.3606,
|
| 200 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 201 |
+
"rewards/chosen": -0.45442095398902893,
|
| 202 |
+
"rewards/margins": 5.263146877288818,
|
| 203 |
+
"rewards/rejected": -5.717567443847656,
|
| 204 |
"step": 120
|
| 205 |
},
|
| 206 |
{
|
| 207 |
+
"epoch": 0.7,
|
| 208 |
+
"grad_norm": 224.99155866870328,
|
| 209 |
+
"learning_rate": 1.2363668353585485e-07,
|
| 210 |
+
"logits/chosen": -0.39301735162734985,
|
| 211 |
+
"logits/rejected": 0.5232549905776978,
|
| 212 |
+
"logps/chosen": -264.6087951660156,
|
| 213 |
+
"logps/rejected": -236.9209442138672,
|
| 214 |
+
"loss": 0.3158,
|
| 215 |
+
"rewards/accuracies": 0.903124988079071,
|
| 216 |
+
"rewards/chosen": -0.47178196907043457,
|
| 217 |
+
"rewards/margins": 5.685537815093994,
|
| 218 |
+
"rewards/rejected": -6.157320499420166,
|
| 219 |
"step": 130
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.76,
|
| 223 |
+
"grad_norm": 234.77001103810687,
|
| 224 |
+
"learning_rate": 8.53118137245516e-08,
|
| 225 |
+
"logits/chosen": -0.21741196513175964,
|
| 226 |
+
"logits/rejected": 0.6465774178504944,
|
| 227 |
+
"logps/chosen": -270.5892028808594,
|
| 228 |
+
"logps/rejected": -251.4233856201172,
|
| 229 |
+
"loss": 0.2942,
|
| 230 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 231 |
+
"rewards/chosen": -0.42987537384033203,
|
| 232 |
+
"rewards/margins": 5.9034929275512695,
|
| 233 |
+
"rewards/rejected": -6.33336877822876,
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.81,
|
| 238 |
+
"grad_norm": 289.2759977765359,
|
| 239 |
+
"learning_rate": 5.2867919617408553e-08,
|
| 240 |
+
"logits/chosen": -0.30728083848953247,
|
| 241 |
+
"logits/rejected": 0.5963491201400757,
|
| 242 |
+
"logps/chosen": -267.1575927734375,
|
| 243 |
+
"logps/rejected": -238.6244659423828,
|
| 244 |
+
"loss": 0.3253,
|
| 245 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 246 |
+
"rewards/chosen": 0.4330620765686035,
|
| 247 |
+
"rewards/margins": 5.231642723083496,
|
| 248 |
+
"rewards/rejected": -4.798580646514893,
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
+
"epoch": 0.86,
|
| 253 |
+
"grad_norm": 230.9320948308424,
|
| 254 |
+
"learning_rate": 2.7463564905650853e-08,
|
| 255 |
+
"logits/chosen": -0.37829893827438354,
|
| 256 |
+
"logits/rejected": 0.43174609541893005,
|
| 257 |
+
"logps/chosen": -270.28094482421875,
|
| 258 |
+
"logps/rejected": -246.8420867919922,
|
| 259 |
+
"loss": 0.3039,
|
| 260 |
+
"rewards/accuracies": 0.893750011920929,
|
| 261 |
+
"rewards/chosen": 0.014125394634902477,
|
| 262 |
+
"rewards/margins": 5.368089199066162,
|
| 263 |
+
"rewards/rejected": -5.353963375091553,
|
| 264 |
"step": 160
|
| 265 |
},
|
| 266 |
{
|
| 267 |
+
"epoch": 0.92,
|
| 268 |
+
"grad_norm": 253.43499472230928,
|
| 269 |
+
"learning_rate": 1.0005933014019307e-08,
|
| 270 |
+
"logits/chosen": -0.3070162534713745,
|
| 271 |
+
"logits/rejected": 0.6451749205589294,
|
| 272 |
+
"logps/chosen": -269.88385009765625,
|
| 273 |
+
"logps/rejected": -247.8417205810547,
|
| 274 |
+
"loss": 0.314,
|
| 275 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 276 |
+
"rewards/chosen": -0.7552144527435303,
|
| 277 |
+
"rewards/margins": 5.324892044067383,
|
| 278 |
+
"rewards/rejected": -6.080106258392334,
|
| 279 |
"step": 170
|
| 280 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 281 |
{
|
| 282 |
"epoch": 0.97,
|
| 283 |
+
"grad_norm": 264.72644139971993,
|
| 284 |
+
"learning_rate": 1.1184317978602808e-09,
|
| 285 |
+
"logits/chosen": -0.4602000117301941,
|
| 286 |
+
"logits/rejected": 0.40653783082962036,
|
| 287 |
+
"logps/chosen": -261.251220703125,
|
| 288 |
+
"logps/rejected": -241.342529296875,
|
| 289 |
+
"loss": 0.404,
|
| 290 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 291 |
+
"rewards/chosen": -0.2922298312187195,
|
| 292 |
+
"rewards/margins": 5.988170623779297,
|
| 293 |
+
"rewards/rejected": -6.28040075302124,
|
| 294 |
+
"step": 180
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 295 |
},
|
| 296 |
{
|
| 297 |
"epoch": 1.0,
|
| 298 |
+
"step": 185,
|
| 299 |
"total_flos": 0.0,
|
| 300 |
+
"train_loss": 0.37812867454580357,
|
| 301 |
+
"train_runtime": 5319.5814,
|
| 302 |
+
"train_samples_per_second": 8.892,
|
| 303 |
+
"train_steps_per_second": 0.035
|
| 304 |
}
|
| 305 |
],
|
| 306 |
"logging_steps": 10,
|
| 307 |
+
"max_steps": 185,
|
| 308 |
"num_input_tokens_seen": 0,
|
| 309 |
"num_train_epochs": 1,
|
| 310 |
"save_steps": 100,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbdc9c5363b2f9445f9cb4d5b25a0e5791e43f82c6521e4905a43fd8fc82e524
|
| 3 |
+
size 6264
|