Model save
Browse files- README.md +1 -17
- 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
- train_results.json +5 -5
- trainer_state.json +454 -616
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -16,16 +16,6 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
# zephyr-7b-dpo-full
|
| 17 |
|
| 18 |
This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on the None dataset.
|
| 19 |
-
It achieves the following results on the evaluation set:
|
| 20 |
-
- Loss: 0.2109
|
| 21 |
-
- Rewards/chosen: -0.8723
|
| 22 |
-
- Rewards/rejected: -1.4751
|
| 23 |
-
- Rewards/accuracies: 0.7109
|
| 24 |
-
- Rewards/margins: 0.6028
|
| 25 |
-
- Logps/rejected: -404.8638
|
| 26 |
-
- Logps/chosen: -344.2667
|
| 27 |
-
- Logits/rejected: -2.3691
|
| 28 |
-
- Logits/chosen: -2.3911
|
| 29 |
|
| 30 |
## Model description
|
| 31 |
|
|
@@ -47,7 +37,7 @@ The following hyperparameters were used during training:
|
|
| 47 |
- learning_rate: 5e-07
|
| 48 |
- train_batch_size: 8
|
| 49 |
- eval_batch_size: 8
|
| 50 |
-
- seed:
|
| 51 |
- distributed_type: multi-GPU
|
| 52 |
- num_devices: 8
|
| 53 |
- gradient_accumulation_steps: 2
|
|
@@ -60,12 +50,6 @@ The following hyperparameters were used during training:
|
|
| 60 |
|
| 61 |
### Training results
|
| 62 |
|
| 63 |
-
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 64 |
-
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 65 |
-
| 0.257 | 0.21 | 100 | 0.2233 | -0.5908 | -0.9685 | 0.6836 | 0.3777 | -354.2039 | -316.1246 | -2.6611 | -2.6767 |
|
| 66 |
-
| 0.2208 | 0.42 | 200 | 0.2272 | -0.7331 | -1.2856 | 0.7109 | 0.5525 | -385.9103 | -330.3494 | -2.4448 | -2.4613 |
|
| 67 |
-
| 0.2288 | 0.63 | 300 | 0.2287 | -0.7577 | -1.2990 | 0.7109 | 0.5414 | -387.2565 | -332.8068 | -2.3991 | -2.4209 |
|
| 68 |
-
| 0.2176 | 0.84 | 400 | 0.2109 | -0.8723 | -1.4751 | 0.7109 | 0.6028 | -404.8638 | -344.2667 | -2.3691 | -2.3911 |
|
| 69 |
|
| 70 |
|
| 71 |
### Framework versions
|
|
|
|
| 16 |
# zephyr-7b-dpo-full
|
| 17 |
|
| 18 |
This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on the None dataset.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
|
|
|
| 37 |
- learning_rate: 5e-07
|
| 38 |
- train_batch_size: 8
|
| 39 |
- eval_batch_size: 8
|
| 40 |
+
- seed: 3
|
| 41 |
- distributed_type: multi-GPU
|
| 42 |
- num_devices: 8
|
| 43 |
- gradient_accumulation_steps: 2
|
|
|
|
| 50 |
|
| 51 |
### Training results
|
| 52 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 53 |
|
| 54 |
|
| 55 |
### Framework versions
|
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":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.1562451661368947,
|
| 4 |
+
"train_runtime": 3209.2418,
|
| 5 |
+
"train_samples": 51894,
|
| 6 |
+
"train_samples_per_second": 16.17,
|
| 7 |
+
"train_steps_per_second": 0.126
|
| 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:e746bdb8da0ed3cb92cfb0eac3514bd900c00b492876034642796dd591398a5b
|
| 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 sha256:54e0fc894931d1546a89c1600598d813b49ed88255568fc1be6110de495a5182
|
| 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:eaef24f314b05e2dbdc3675cb3e5a19536af34dec86696772f465d0801e7198e
|
| 3 |
size 4540516344
|
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":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.1562451661368947,
|
| 4 |
+
"train_runtime": 3209.2418,
|
| 5 |
+
"train_samples": 51894,
|
| 6 |
+
"train_samples_per_second": 16.17,
|
| 7 |
+
"train_steps_per_second": 0.126
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,21 +1,21 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"eval_steps":
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
-
"learning_rate": 1.
|
| 14 |
-
"logits/chosen": -2.
|
| 15 |
-
"logits/rejected": -2.
|
| 16 |
-
"logps/chosen": -
|
| 17 |
-
"logps/rejected": -
|
| 18 |
-
"loss": 0.
|
| 19 |
"rewards/accuracies": 0.0,
|
| 20 |
"rewards/chosen": 0.0,
|
| 21 |
"rewards/margins": 0.0,
|
|
@@ -24,738 +24,576 @@
|
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.02,
|
| 27 |
-
"learning_rate": 1.
|
| 28 |
-
"logits/chosen": -2.
|
| 29 |
-
"logits/rejected": -2.
|
| 30 |
-
"logps/chosen": -
|
| 31 |
-
"logps/rejected": -
|
| 32 |
-
"loss": 0.
|
| 33 |
-
"rewards/accuracies": 0.
|
| 34 |
-
"rewards/chosen": 0.
|
| 35 |
-
"rewards/margins": 0.
|
| 36 |
-
"rewards/rejected":
|
| 37 |
"step": 10
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"learning_rate": 2.
|
| 42 |
-
"logits/chosen": -2.
|
| 43 |
-
"logits/rejected": -2.
|
| 44 |
-
"logps/chosen": -
|
| 45 |
-
"logps/rejected": -
|
| 46 |
-
"loss": 0.
|
| 47 |
-
"rewards/accuracies": 0.
|
| 48 |
-
"rewards/chosen": 0.
|
| 49 |
-
"rewards/margins": 0.
|
| 50 |
-
"rewards/rejected": -0.
|
| 51 |
"step": 20
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"learning_rate": 3.
|
| 56 |
-
"logits/chosen": -2.
|
| 57 |
-
"logits/rejected": -2.
|
| 58 |
-
"logps/chosen": -
|
| 59 |
-
"logps/rejected": -
|
| 60 |
-
"loss": 0.
|
| 61 |
-
"rewards/accuracies": 0.
|
| 62 |
-
"rewards/chosen":
|
| 63 |
-
"rewards/margins": 0.
|
| 64 |
-
"rewards/rejected": -0.
|
| 65 |
"step": 30
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"learning_rate": 4.
|
| 70 |
-
"logits/chosen": -2.
|
| 71 |
-
"logits/rejected": -2.
|
| 72 |
-
"logps/chosen": -
|
| 73 |
-
"logps/rejected": -
|
| 74 |
-
"loss": 0.
|
| 75 |
-
"rewards/accuracies": 0.
|
| 76 |
-
"rewards/chosen":
|
| 77 |
-
"rewards/margins": 0.
|
| 78 |
-
"rewards/rejected": -0.
|
| 79 |
"step": 40
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"learning_rate": 4.
|
| 84 |
-
"logits/chosen": -2.
|
| 85 |
-
"logits/rejected": -2.
|
| 86 |
-
"logps/chosen": -
|
| 87 |
-
"logps/rejected": -
|
| 88 |
-
"loss": 0.
|
| 89 |
-
"rewards/accuracies": 0.
|
| 90 |
-
"rewards/chosen": -0.
|
| 91 |
-
"rewards/margins": 0.
|
| 92 |
-
"rewards/rejected": -0.
|
| 93 |
"step": 50
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"learning_rate": 4.
|
| 98 |
-
"logits/chosen": -2.
|
| 99 |
-
"logits/rejected": -2.
|
| 100 |
-
"logps/chosen": -
|
| 101 |
-
"logps/rejected": -
|
| 102 |
-
"loss": 0.
|
| 103 |
-
"rewards/accuracies": 0.
|
| 104 |
-
"rewards/chosen": -0.
|
| 105 |
-
"rewards/margins": 0.
|
| 106 |
-
"rewards/rejected": -
|
| 107 |
"step": 60
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"learning_rate": 4.
|
| 112 |
-
"logits/chosen": -2.
|
| 113 |
-
"logits/rejected": -2.
|
| 114 |
-
"logps/chosen": -
|
| 115 |
-
"logps/rejected": -
|
| 116 |
-
"loss": 0.
|
| 117 |
-
"rewards/accuracies": 0.
|
| 118 |
-
"rewards/chosen": -0.
|
| 119 |
-
"rewards/margins":
|
| 120 |
-
"rewards/rejected": -
|
| 121 |
"step": 70
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"learning_rate": 4.
|
| 126 |
-
"logits/chosen": -2.
|
| 127 |
-
"logits/rejected": -2.
|
| 128 |
-
"logps/chosen": -
|
| 129 |
-
"logps/rejected": -
|
| 130 |
-
"loss": 0.
|
| 131 |
-
"rewards/accuracies": 0.
|
| 132 |
-
"rewards/chosen": -0.
|
| 133 |
-
"rewards/margins":
|
| 134 |
-
"rewards/rejected": -
|
| 135 |
"step": 80
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"learning_rate": 4.
|
| 140 |
-
"logits/chosen": -2.
|
| 141 |
-
"logits/rejected": -2.
|
| 142 |
-
"logps/chosen": -
|
| 143 |
-
"logps/rejected": -
|
| 144 |
-
"loss": 0.
|
| 145 |
-
"rewards/accuracies": 0.
|
| 146 |
-
"rewards/chosen": -0.
|
| 147 |
-
"rewards/margins":
|
| 148 |
-
"rewards/rejected": -
|
| 149 |
"step": 90
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"learning_rate": 4.
|
| 154 |
-
"logits/chosen": -2.
|
| 155 |
-
"logits/rejected": -2.
|
| 156 |
-
"logps/chosen": -
|
| 157 |
-
"logps/rejected": -
|
| 158 |
-
"loss": 0.
|
| 159 |
-
"rewards/accuracies": 0.
|
| 160 |
-
"rewards/chosen": -0.
|
| 161 |
-
"rewards/margins":
|
| 162 |
-
"rewards/rejected": -
|
| 163 |
-
"step": 100
|
| 164 |
-
},
|
| 165 |
-
{
|
| 166 |
-
"epoch": 0.21,
|
| 167 |
-
"eval_logits/chosen": -2.6766650676727295,
|
| 168 |
-
"eval_logits/rejected": -2.6611006259918213,
|
| 169 |
-
"eval_logps/chosen": -316.1246337890625,
|
| 170 |
-
"eval_logps/rejected": -354.2039489746094,
|
| 171 |
-
"eval_loss": 0.2232680767774582,
|
| 172 |
-
"eval_rewards/accuracies": 0.68359375,
|
| 173 |
-
"eval_rewards/chosen": -0.590849757194519,
|
| 174 |
-
"eval_rewards/margins": 0.3776569366455078,
|
| 175 |
-
"eval_rewards/rejected": -0.9685067534446716,
|
| 176 |
-
"eval_runtime": 53.2908,
|
| 177 |
-
"eval_samples_per_second": 37.53,
|
| 178 |
-
"eval_steps_per_second": 0.6,
|
| 179 |
"step": 100
|
| 180 |
},
|
| 181 |
{
|
| 182 |
-
"epoch": 0.
|
| 183 |
-
"learning_rate": 4.
|
| 184 |
-
"logits/chosen": -2.
|
| 185 |
-
"logits/rejected": -2.
|
| 186 |
-
"logps/chosen": -
|
| 187 |
-
"logps/rejected": -
|
| 188 |
-
"loss": 0.
|
| 189 |
-
"rewards/accuracies": 0.
|
| 190 |
-
"rewards/chosen": -0.
|
| 191 |
-
"rewards/margins":
|
| 192 |
-
"rewards/rejected": -
|
| 193 |
"step": 110
|
| 194 |
},
|
| 195 |
{
|
| 196 |
-
"epoch": 0.
|
| 197 |
-
"learning_rate": 4.
|
| 198 |
-
"logits/chosen": -2.
|
| 199 |
-
"logits/rejected": -2.
|
| 200 |
-
"logps/chosen": -
|
| 201 |
-
"logps/rejected": -
|
| 202 |
-
"loss": 0.
|
| 203 |
-
"rewards/accuracies": 0.
|
| 204 |
-
"rewards/chosen": -0.
|
| 205 |
-
"rewards/margins":
|
| 206 |
-
"rewards/rejected": -
|
| 207 |
"step": 120
|
| 208 |
},
|
| 209 |
{
|
| 210 |
-
"epoch": 0.
|
| 211 |
-
"learning_rate": 4.
|
| 212 |
-
"logits/chosen": -2.
|
| 213 |
-
"logits/rejected": -2.
|
| 214 |
-
"logps/chosen": -
|
| 215 |
-
"logps/rejected": -
|
| 216 |
-
"loss": 0.
|
| 217 |
-
"rewards/accuracies": 0.
|
| 218 |
-
"rewards/chosen": -0.
|
| 219 |
-
"rewards/margins":
|
| 220 |
-
"rewards/rejected": -
|
| 221 |
"step": 130
|
| 222 |
},
|
| 223 |
{
|
| 224 |
-
"epoch": 0.
|
| 225 |
-
"learning_rate": 4.
|
| 226 |
-
"logits/chosen": -2.
|
| 227 |
-
"logits/rejected": -2.
|
| 228 |
-
"logps/chosen": -
|
| 229 |
-
"logps/rejected": -
|
| 230 |
-
"loss": 0.
|
| 231 |
-
"rewards/accuracies": 0.
|
| 232 |
-
"rewards/chosen": -0.
|
| 233 |
-
"rewards/margins":
|
| 234 |
-
"rewards/rejected": -
|
| 235 |
"step": 140
|
| 236 |
},
|
| 237 |
{
|
| 238 |
-
"epoch": 0.
|
| 239 |
-
"learning_rate":
|
| 240 |
-
"logits/chosen": -2.
|
| 241 |
-
"logits/rejected": -2.
|
| 242 |
-
"logps/chosen": -
|
| 243 |
-
"logps/rejected": -
|
| 244 |
-
"loss": 0.
|
| 245 |
-
"rewards/accuracies": 0.
|
| 246 |
-
"rewards/chosen": -0.
|
| 247 |
-
"rewards/margins":
|
| 248 |
-
"rewards/rejected": -
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
-
"epoch": 0.
|
| 253 |
-
"learning_rate":
|
| 254 |
-
"logits/chosen": -2.
|
| 255 |
-
"logits/rejected": -2.
|
| 256 |
-
"logps/chosen": -
|
| 257 |
-
"logps/rejected": -
|
| 258 |
-
"loss": 0.
|
| 259 |
-
"rewards/accuracies": 0.
|
| 260 |
-
"rewards/chosen": -0.
|
| 261 |
-
"rewards/margins":
|
| 262 |
-
"rewards/rejected": -
|
| 263 |
"step": 160
|
| 264 |
},
|
| 265 |
{
|
| 266 |
-
"epoch": 0.
|
| 267 |
-
"learning_rate":
|
| 268 |
-
"logits/chosen": -2.
|
| 269 |
-
"logits/rejected": -2.
|
| 270 |
-
"logps/chosen": -
|
| 271 |
-
"logps/rejected": -
|
| 272 |
-
"loss": 0.
|
| 273 |
-
"rewards/accuracies": 0.
|
| 274 |
-
"rewards/chosen": -0.
|
| 275 |
-
"rewards/margins":
|
| 276 |
-
"rewards/rejected": -
|
| 277 |
"step": 170
|
| 278 |
},
|
| 279 |
{
|
| 280 |
-
"epoch": 0.
|
| 281 |
-
"learning_rate": 3.
|
| 282 |
-
"logits/chosen": -2.
|
| 283 |
-
"logits/rejected": -2.
|
| 284 |
-
"logps/chosen": -
|
| 285 |
-
"logps/rejected": -
|
| 286 |
-
"loss": 0.
|
| 287 |
-
"rewards/accuracies": 0.
|
| 288 |
-
"rewards/chosen": -0.
|
| 289 |
-
"rewards/margins":
|
| 290 |
-
"rewards/rejected": -
|
| 291 |
"step": 180
|
| 292 |
},
|
| 293 |
{
|
| 294 |
-
"epoch": 0.
|
| 295 |
-
"learning_rate": 3.
|
| 296 |
-
"logits/chosen": -2.
|
| 297 |
-
"logits/rejected": -2.
|
| 298 |
-
"logps/chosen": -
|
| 299 |
-
"logps/rejected": -
|
| 300 |
-
"loss": 0.
|
| 301 |
-
"rewards/accuracies": 0.
|
| 302 |
-
"rewards/chosen": -0.
|
| 303 |
-
"rewards/margins":
|
| 304 |
-
"rewards/rejected": -
|
| 305 |
"step": 190
|
| 306 |
},
|
| 307 |
{
|
| 308 |
-
"epoch": 0.
|
| 309 |
-
"learning_rate":
|
| 310 |
-
"logits/chosen": -2.
|
| 311 |
-
"logits/rejected": -2.
|
| 312 |
-
"logps/chosen": -
|
| 313 |
-
"logps/rejected": -
|
| 314 |
-
"loss": 0.
|
| 315 |
-
"rewards/accuracies": 0.
|
| 316 |
-
"rewards/chosen": -0.
|
| 317 |
-
"rewards/margins":
|
| 318 |
-
"rewards/rejected": -
|
| 319 |
"step": 200
|
| 320 |
},
|
| 321 |
{
|
| 322 |
-
"epoch": 0.
|
| 323 |
-
"
|
| 324 |
-
"
|
| 325 |
-
"
|
| 326 |
-
"
|
| 327 |
-
"
|
| 328 |
-
"
|
| 329 |
-
"
|
| 330 |
-
"
|
| 331 |
-
"
|
| 332 |
-
"
|
| 333 |
-
"eval_samples_per_second": 37.503,
|
| 334 |
-
"eval_steps_per_second": 0.6,
|
| 335 |
-
"step": 200
|
| 336 |
-
},
|
| 337 |
-
{
|
| 338 |
-
"epoch": 0.44,
|
| 339 |
-
"learning_rate": 3.4440382358952115e-07,
|
| 340 |
-
"logits/chosen": -2.408134937286377,
|
| 341 |
-
"logits/rejected": -2.4148223400115967,
|
| 342 |
-
"logps/chosen": -372.76190185546875,
|
| 343 |
-
"logps/rejected": -418.78070068359375,
|
| 344 |
-
"loss": 0.2046,
|
| 345 |
-
"rewards/accuracies": 0.824999988079071,
|
| 346 |
-
"rewards/chosen": -0.6905978918075562,
|
| 347 |
-
"rewards/margins": 0.6037817597389221,
|
| 348 |
-
"rewards/rejected": -1.294379711151123,
|
| 349 |
"step": 210
|
| 350 |
},
|
| 351 |
{
|
| 352 |
-
"epoch": 0.
|
| 353 |
-
"learning_rate":
|
| 354 |
-
"logits/chosen": -2.
|
| 355 |
-
"logits/rejected": -2.
|
| 356 |
-
"logps/chosen": -
|
| 357 |
-
"logps/rejected": -
|
| 358 |
-
"loss": 0.
|
| 359 |
-
"rewards/accuracies": 0.
|
| 360 |
-
"rewards/chosen": -0.
|
| 361 |
-
"rewards/margins":
|
| 362 |
-
"rewards/rejected": -
|
| 363 |
"step": 220
|
| 364 |
},
|
| 365 |
{
|
| 366 |
-
"epoch": 0.
|
| 367 |
-
"learning_rate":
|
| 368 |
-
"logits/chosen": -2.
|
| 369 |
-
"logits/rejected": -2.
|
| 370 |
-
"logps/chosen": -
|
| 371 |
-
"logps/rejected": -
|
| 372 |
-
"loss": 0.
|
| 373 |
-
"rewards/accuracies": 0.
|
| 374 |
-
"rewards/chosen": -
|
| 375 |
-
"rewards/margins":
|
| 376 |
-
"rewards/rejected": -
|
| 377 |
"step": 230
|
| 378 |
},
|
| 379 |
{
|
| 380 |
-
"epoch": 0.
|
| 381 |
-
"learning_rate": 2.
|
| 382 |
-
"logits/chosen": -2.
|
| 383 |
-
"logits/rejected": -2.
|
| 384 |
-
"logps/chosen": -
|
| 385 |
-
"logps/rejected": -
|
| 386 |
-
"loss": 0.
|
| 387 |
-
"rewards/accuracies": 0.
|
| 388 |
-
"rewards/chosen": -0.
|
| 389 |
-
"rewards/margins":
|
| 390 |
-
"rewards/rejected": -
|
| 391 |
"step": 240
|
| 392 |
},
|
| 393 |
{
|
| 394 |
-
"epoch": 0.
|
| 395 |
-
"learning_rate":
|
| 396 |
-
"logits/chosen": -2.
|
| 397 |
-
"logits/rejected": -2.
|
| 398 |
-
"logps/chosen": -
|
| 399 |
-
"logps/rejected": -
|
| 400 |
-
"loss": 0.
|
| 401 |
-
"rewards/accuracies": 0.
|
| 402 |
-
"rewards/chosen": -0.
|
| 403 |
-
"rewards/margins":
|
| 404 |
-
"rewards/rejected": -
|
| 405 |
"step": 250
|
| 406 |
},
|
| 407 |
{
|
| 408 |
-
"epoch": 0.
|
| 409 |
-
"learning_rate":
|
| 410 |
-
"logits/chosen": -2.
|
| 411 |
-
"logits/rejected": -2.
|
| 412 |
-
"logps/chosen": -
|
| 413 |
-
"logps/rejected": -
|
| 414 |
-
"loss": 0.
|
| 415 |
-
"rewards/accuracies": 0.
|
| 416 |
-
"rewards/chosen": -0.
|
| 417 |
-
"rewards/margins":
|
| 418 |
-
"rewards/rejected": -
|
| 419 |
"step": 260
|
| 420 |
},
|
| 421 |
{
|
| 422 |
-
"epoch": 0.
|
| 423 |
-
"learning_rate":
|
| 424 |
-
"logits/chosen": -2.
|
| 425 |
-
"logits/rejected": -2.
|
| 426 |
-
"logps/chosen": -
|
| 427 |
-
"logps/rejected": -
|
| 428 |
-
"loss": 0.
|
| 429 |
-
"rewards/accuracies": 0.
|
| 430 |
-
"rewards/chosen": -0.
|
| 431 |
-
"rewards/margins":
|
| 432 |
-
"rewards/rejected": -
|
| 433 |
"step": 270
|
| 434 |
},
|
| 435 |
{
|
| 436 |
-
"epoch": 0.
|
| 437 |
-
"learning_rate":
|
| 438 |
-
"logits/chosen": -2.
|
| 439 |
-
"logits/rejected": -2.
|
| 440 |
-
"logps/chosen": -
|
| 441 |
-
"logps/rejected": -
|
| 442 |
-
"loss": 0.
|
| 443 |
-
"rewards/accuracies": 0.
|
| 444 |
-
"rewards/chosen": -0.
|
| 445 |
-
"rewards/margins":
|
| 446 |
-
"rewards/rejected": -
|
| 447 |
"step": 280
|
| 448 |
},
|
| 449 |
{
|
| 450 |
-
"epoch": 0.
|
| 451 |
-
"learning_rate":
|
| 452 |
-
"logits/chosen": -2.
|
| 453 |
-
"logits/rejected": -2.
|
| 454 |
-
"logps/chosen": -
|
| 455 |
-
"logps/rejected": -
|
| 456 |
-
"loss": 0.
|
| 457 |
-
"rewards/accuracies": 0.
|
| 458 |
-
"rewards/chosen": -0.
|
| 459 |
-
"rewards/margins":
|
| 460 |
-
"rewards/rejected": -
|
| 461 |
"step": 290
|
| 462 |
},
|
| 463 |
{
|
| 464 |
-
"epoch": 0.
|
| 465 |
-
"learning_rate":
|
| 466 |
-
"logits/chosen": -2.
|
| 467 |
-
"logits/rejected": -2.
|
| 468 |
-
"logps/chosen": -
|
| 469 |
-
"logps/rejected": -
|
| 470 |
-
"loss": 0.
|
| 471 |
-
"rewards/accuracies": 0.
|
| 472 |
-
"rewards/chosen": -
|
| 473 |
-
"rewards/margins":
|
| 474 |
-
"rewards/rejected": -
|
| 475 |
"step": 300
|
| 476 |
},
|
| 477 |
{
|
| 478 |
-
"epoch": 0.
|
| 479 |
-
"
|
| 480 |
-
"
|
| 481 |
-
"
|
| 482 |
-
"
|
| 483 |
-
"
|
| 484 |
-
"
|
| 485 |
-
"
|
| 486 |
-
"
|
| 487 |
-
"
|
| 488 |
-
"
|
| 489 |
-
"eval_samples_per_second": 37.48,
|
| 490 |
-
"eval_steps_per_second": 0.6,
|
| 491 |
-
"step": 300
|
| 492 |
-
},
|
| 493 |
-
{
|
| 494 |
-
"epoch": 0.65,
|
| 495 |
-
"learning_rate": 1.6583128063291573e-07,
|
| 496 |
-
"logits/chosen": -2.3479456901550293,
|
| 497 |
-
"logits/rejected": -2.308645248413086,
|
| 498 |
-
"logps/chosen": -374.3102111816406,
|
| 499 |
-
"logps/rejected": -374.8952941894531,
|
| 500 |
-
"loss": 0.234,
|
| 501 |
-
"rewards/accuracies": 0.6875,
|
| 502 |
-
"rewards/chosen": -0.8235334157943726,
|
| 503 |
-
"rewards/margins": 0.4518548846244812,
|
| 504 |
-
"rewards/rejected": -1.275388240814209,
|
| 505 |
"step": 310
|
| 506 |
},
|
| 507 |
{
|
| 508 |
-
"epoch": 0.
|
| 509 |
-
"learning_rate":
|
| 510 |
-
"logits/chosen": -2.
|
| 511 |
-
"logits/rejected": -2.
|
| 512 |
-
"logps/chosen": -
|
| 513 |
-
"logps/rejected": -
|
| 514 |
-
"loss": 0.
|
| 515 |
-
"rewards/accuracies": 0.
|
| 516 |
-
"rewards/chosen": -
|
| 517 |
-
"rewards/margins":
|
| 518 |
-
"rewards/rejected": -
|
| 519 |
"step": 320
|
| 520 |
},
|
| 521 |
{
|
| 522 |
-
"epoch": 0.
|
| 523 |
-
"learning_rate":
|
| 524 |
-
"logits/chosen": -2.
|
| 525 |
-
"logits/rejected": -2.
|
| 526 |
-
"logps/chosen": -
|
| 527 |
-
"logps/rejected": -
|
| 528 |
-
"loss": 0.
|
| 529 |
-
"rewards/accuracies": 0.
|
| 530 |
-
"rewards/chosen": -
|
| 531 |
-
"rewards/margins":
|
| 532 |
-
"rewards/rejected": -
|
| 533 |
"step": 330
|
| 534 |
},
|
| 535 |
{
|
| 536 |
-
"epoch": 0.
|
| 537 |
-
"learning_rate":
|
| 538 |
-
"logits/chosen": -2.
|
| 539 |
-
"logits/rejected": -2.
|
| 540 |
-
"logps/chosen": -
|
| 541 |
-
"logps/rejected": -
|
| 542 |
-
"loss": 0.
|
| 543 |
-
"rewards/accuracies": 0.
|
| 544 |
-
"rewards/chosen": -
|
| 545 |
-
"rewards/margins":
|
| 546 |
-
"rewards/rejected": -
|
| 547 |
"step": 340
|
| 548 |
},
|
| 549 |
{
|
| 550 |
-
"epoch": 0.
|
| 551 |
-
"learning_rate":
|
| 552 |
-
"logits/chosen": -2.
|
| 553 |
-
"logits/rejected": -2.
|
| 554 |
-
"logps/chosen": -
|
| 555 |
-
"logps/rejected": -
|
| 556 |
-
"loss": 0.
|
| 557 |
-
"rewards/accuracies": 0.
|
| 558 |
-
"rewards/chosen": -
|
| 559 |
-
"rewards/margins":
|
| 560 |
-
"rewards/rejected": -
|
| 561 |
"step": 350
|
| 562 |
},
|
| 563 |
{
|
| 564 |
-
"epoch": 0.
|
| 565 |
-
"learning_rate":
|
| 566 |
-
"logits/chosen": -2.
|
| 567 |
-
"logits/rejected": -2.
|
| 568 |
-
"logps/chosen": -
|
| 569 |
-
"logps/rejected": -
|
| 570 |
-
"loss": 0.
|
| 571 |
-
"rewards/accuracies": 0.
|
| 572 |
-
"rewards/chosen": -
|
| 573 |
-
"rewards/margins":
|
| 574 |
-
"rewards/rejected": -
|
| 575 |
"step": 360
|
| 576 |
},
|
| 577 |
{
|
| 578 |
-
"epoch": 0.
|
| 579 |
-
"learning_rate":
|
| 580 |
-
"logits/chosen": -2.
|
| 581 |
-
"logits/rejected": -2.
|
| 582 |
-
"logps/chosen": -
|
| 583 |
-
"logps/rejected": -
|
| 584 |
-
"loss": 0.
|
| 585 |
-
"rewards/accuracies": 0.
|
| 586 |
-
"rewards/chosen": -
|
| 587 |
-
"rewards/margins":
|
| 588 |
-
"rewards/rejected": -
|
| 589 |
"step": 370
|
| 590 |
},
|
| 591 |
{
|
| 592 |
-
"epoch": 0.
|
| 593 |
-
"learning_rate":
|
| 594 |
-
"logits/chosen": -2.
|
| 595 |
-
"logits/rejected": -2.
|
| 596 |
-
"logps/chosen": -
|
| 597 |
-
"logps/rejected": -
|
| 598 |
-
"loss": 0.
|
| 599 |
-
"rewards/accuracies": 0.
|
| 600 |
-
"rewards/chosen": -
|
| 601 |
-
"rewards/margins":
|
| 602 |
-
"rewards/rejected": -
|
| 603 |
"step": 380
|
| 604 |
},
|
| 605 |
{
|
| 606 |
-
"epoch": 0.
|
| 607 |
-
"learning_rate":
|
| 608 |
-
"logits/chosen": -2.
|
| 609 |
-
"logits/rejected": -2.
|
| 610 |
-
"logps/chosen": -
|
| 611 |
-
"logps/rejected": -
|
| 612 |
-
"loss": 0.
|
| 613 |
-
"rewards/accuracies": 0.
|
| 614 |
-
"rewards/chosen": -0.
|
| 615 |
-
"rewards/margins":
|
| 616 |
-
"rewards/rejected": -
|
| 617 |
"step": 390
|
| 618 |
},
|
| 619 |
{
|
| 620 |
-
"epoch": 0.
|
| 621 |
-
"learning_rate":
|
| 622 |
-
"logits/chosen": -2.
|
| 623 |
-
"logits/rejected": -2.
|
| 624 |
-
"logps/chosen": -
|
| 625 |
-
"logps/rejected": -
|
| 626 |
-
"loss": 0.
|
| 627 |
-
"rewards/accuracies": 0.
|
| 628 |
-
"rewards/chosen": -0.
|
| 629 |
-
"rewards/margins":
|
| 630 |
-
"rewards/rejected": -
|
| 631 |
-
"step": 400
|
| 632 |
-
},
|
| 633 |
-
{
|
| 634 |
-
"epoch": 0.84,
|
| 635 |
-
"eval_logits/chosen": -2.3910720348358154,
|
| 636 |
-
"eval_logits/rejected": -2.369123935699463,
|
| 637 |
-
"eval_logps/chosen": -344.2667236328125,
|
| 638 |
-
"eval_logps/rejected": -404.86376953125,
|
| 639 |
-
"eval_loss": 0.21091435849666595,
|
| 640 |
-
"eval_rewards/accuracies": 0.7109375,
|
| 641 |
-
"eval_rewards/chosen": -0.8722706437110901,
|
| 642 |
-
"eval_rewards/margins": 0.6028342247009277,
|
| 643 |
-
"eval_rewards/rejected": -1.475104808807373,
|
| 644 |
-
"eval_runtime": 53.3295,
|
| 645 |
-
"eval_samples_per_second": 37.503,
|
| 646 |
-
"eval_steps_per_second": 0.6,
|
| 647 |
"step": 400
|
| 648 |
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 0.86,
|
| 651 |
-
"learning_rate": 3.022313472693447e-08,
|
| 652 |
-
"logits/chosen": -2.35029935836792,
|
| 653 |
-
"logits/rejected": -2.3197388648986816,
|
| 654 |
-
"logps/chosen": -336.02435302734375,
|
| 655 |
-
"logps/rejected": -370.9861755371094,
|
| 656 |
-
"loss": 0.2113,
|
| 657 |
-
"rewards/accuracies": 0.65625,
|
| 658 |
-
"rewards/chosen": -0.8065496683120728,
|
| 659 |
-
"rewards/margins": 0.48399966955184937,
|
| 660 |
-
"rewards/rejected": -1.290549397468567,
|
| 661 |
-
"step": 410
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"epoch": 0.88,
|
| 665 |
-
"learning_rate": 2.2111614344599684e-08,
|
| 666 |
-
"logits/chosen": -2.384350299835205,
|
| 667 |
-
"logits/rejected": -2.343932867050171,
|
| 668 |
-
"logps/chosen": -353.1007385253906,
|
| 669 |
-
"logps/rejected": -373.4040222167969,
|
| 670 |
-
"loss": 0.2027,
|
| 671 |
-
"rewards/accuracies": 0.6499999761581421,
|
| 672 |
-
"rewards/chosen": -0.8209117650985718,
|
| 673 |
-
"rewards/margins": 0.5275887250900269,
|
| 674 |
-
"rewards/rejected": -1.3485008478164673,
|
| 675 |
-
"step": 420
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 0.9,
|
| 679 |
-
"learning_rate": 1.521597710086439e-08,
|
| 680 |
-
"logits/chosen": -2.332576036453247,
|
| 681 |
-
"logits/rejected": -2.319338321685791,
|
| 682 |
-
"logps/chosen": -381.9653015136719,
|
| 683 |
-
"logps/rejected": -392.8800964355469,
|
| 684 |
-
"loss": 0.207,
|
| 685 |
-
"rewards/accuracies": 0.7437499761581421,
|
| 686 |
-
"rewards/chosen": -0.9718208312988281,
|
| 687 |
-
"rewards/margins": 0.4558374285697937,
|
| 688 |
-
"rewards/rejected": -1.4276583194732666,
|
| 689 |
-
"step": 430
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 0.92,
|
| 693 |
-
"learning_rate": 9.57301420397924e-09,
|
| 694 |
-
"logits/chosen": -2.3502933979034424,
|
| 695 |
-
"logits/rejected": -2.3326973915100098,
|
| 696 |
-
"logps/chosen": -355.3913879394531,
|
| 697 |
-
"logps/rejected": -406.2868957519531,
|
| 698 |
-
"loss": 0.2031,
|
| 699 |
-
"rewards/accuracies": 0.7250000238418579,
|
| 700 |
-
"rewards/chosen": -0.955398678779602,
|
| 701 |
-
"rewards/margins": 0.5213964581489563,
|
| 702 |
-
"rewards/rejected": -1.4767951965332031,
|
| 703 |
-
"step": 440
|
| 704 |
-
},
|
| 705 |
-
{
|
| 706 |
-
"epoch": 0.94,
|
| 707 |
-
"learning_rate": 5.212833302556258e-09,
|
| 708 |
-
"logits/chosen": -2.3776185512542725,
|
| 709 |
-
"logits/rejected": -2.3594932556152344,
|
| 710 |
-
"logps/chosen": -356.1197509765625,
|
| 711 |
-
"logps/rejected": -393.2214050292969,
|
| 712 |
-
"loss": 0.214,
|
| 713 |
-
"rewards/accuracies": 0.6875,
|
| 714 |
-
"rewards/chosen": -0.9058281183242798,
|
| 715 |
-
"rewards/margins": 0.49910789728164673,
|
| 716 |
-
"rewards/rejected": -1.4049360752105713,
|
| 717 |
-
"step": 450
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.96,
|
| 721 |
-
"learning_rate": 2.158697848236607e-09,
|
| 722 |
-
"logits/chosen": -2.3594970703125,
|
| 723 |
-
"logits/rejected": -2.3184947967529297,
|
| 724 |
-
"logps/chosen": -365.49615478515625,
|
| 725 |
-
"logps/rejected": -374.89263916015625,
|
| 726 |
-
"loss": 0.2053,
|
| 727 |
-
"rewards/accuracies": 0.699999988079071,
|
| 728 |
-
"rewards/chosen": -0.9762248992919922,
|
| 729 |
-
"rewards/margins": 0.49983182549476624,
|
| 730 |
-
"rewards/rejected": -1.4760568141937256,
|
| 731 |
-
"step": 460
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.98,
|
| 735 |
-
"learning_rate": 4.269029751107489e-10,
|
| 736 |
-
"logits/chosen": -2.3517262935638428,
|
| 737 |
-
"logits/rejected": -2.3309903144836426,
|
| 738 |
-
"logps/chosen": -355.54058837890625,
|
| 739 |
-
"logps/rejected": -393.1492614746094,
|
| 740 |
-
"loss": 0.2037,
|
| 741 |
-
"rewards/accuracies": 0.706250011920929,
|
| 742 |
-
"rewards/chosen": -0.8432854413986206,
|
| 743 |
-
"rewards/margins": 0.583833634853363,
|
| 744 |
-
"rewards/rejected": -1.427119255065918,
|
| 745 |
-
"step": 470
|
| 746 |
-
},
|
| 747 |
{
|
| 748 |
"epoch": 1.0,
|
| 749 |
-
"step":
|
| 750 |
"total_flos": 0.0,
|
| 751 |
-
"train_loss": 0.
|
| 752 |
-
"train_runtime":
|
| 753 |
-
"train_samples_per_second":
|
| 754 |
-
"train_steps_per_second": 0.
|
| 755 |
}
|
| 756 |
],
|
| 757 |
"logging_steps": 10,
|
| 758 |
-
"max_steps":
|
| 759 |
"num_train_epochs": 1,
|
| 760 |
"save_steps": 1000,
|
| 761 |
"total_flos": 0.0,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.998766954377312,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 405,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 1.2195121951219512e-08,
|
| 14 |
+
"logits/chosen": -2.8088459968566895,
|
| 15 |
+
"logits/rejected": -2.7595884799957275,
|
| 16 |
+
"logps/chosen": -368.90777587890625,
|
| 17 |
+
"logps/rejected": -133.10202026367188,
|
| 18 |
+
"loss": 0.5469,
|
| 19 |
"rewards/accuracies": 0.0,
|
| 20 |
"rewards/chosen": 0.0,
|
| 21 |
"rewards/margins": 0.0,
|
|
|
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.02,
|
| 27 |
+
"learning_rate": 1.219512195121951e-07,
|
| 28 |
+
"logits/chosen": -2.838653802871704,
|
| 29 |
+
"logits/rejected": -2.824901819229126,
|
| 30 |
+
"logps/chosen": -433.81378173828125,
|
| 31 |
+
"logps/rejected": -114.705810546875,
|
| 32 |
+
"loss": 0.55,
|
| 33 |
+
"rewards/accuracies": 0.5,
|
| 34 |
+
"rewards/chosen": 0.0011021953541785479,
|
| 35 |
+
"rewards/margins": 0.0018466737819835544,
|
| 36 |
+
"rewards/rejected": -0.0007444784860126674,
|
| 37 |
"step": 10
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.05,
|
| 41 |
+
"learning_rate": 2.439024390243902e-07,
|
| 42 |
+
"logits/chosen": -2.797428607940674,
|
| 43 |
+
"logits/rejected": -2.7644600868225098,
|
| 44 |
+
"logps/chosen": -436.551025390625,
|
| 45 |
+
"logps/rejected": -109.42466735839844,
|
| 46 |
+
"loss": 0.5381,
|
| 47 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 48 |
+
"rewards/chosen": 0.02190575934946537,
|
| 49 |
+
"rewards/margins": 0.03880878537893295,
|
| 50 |
+
"rewards/rejected": -0.01690302975475788,
|
| 51 |
"step": 20
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.07,
|
| 55 |
+
"learning_rate": 3.6585365853658536e-07,
|
| 56 |
+
"logits/chosen": -2.7145814895629883,
|
| 57 |
+
"logits/rejected": -2.6879427433013916,
|
| 58 |
+
"logps/chosen": -422.17218017578125,
|
| 59 |
+
"logps/rejected": -128.7976531982422,
|
| 60 |
+
"loss": 0.4635,
|
| 61 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 62 |
+
"rewards/chosen": 0.07188864052295685,
|
| 63 |
+
"rewards/margins": 0.20736002922058105,
|
| 64 |
+
"rewards/rejected": -0.1354713886976242,
|
| 65 |
"step": 30
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.1,
|
| 69 |
+
"learning_rate": 4.878048780487804e-07,
|
| 70 |
+
"logits/chosen": -2.588033676147461,
|
| 71 |
+
"logits/rejected": -2.5705184936523438,
|
| 72 |
+
"logps/chosen": -397.0795593261719,
|
| 73 |
+
"logps/rejected": -140.2168426513672,
|
| 74 |
+
"loss": 0.3735,
|
| 75 |
+
"rewards/accuracies": 0.762499988079071,
|
| 76 |
+
"rewards/chosen": 0.016153398901224136,
|
| 77 |
+
"rewards/margins": 0.42458558082580566,
|
| 78 |
+
"rewards/rejected": -0.40843215584754944,
|
| 79 |
"step": 40
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.12,
|
| 83 |
+
"learning_rate": 4.992461696250783e-07,
|
| 84 |
+
"logits/chosen": -2.4239001274108887,
|
| 85 |
+
"logits/rejected": -2.3979756832122803,
|
| 86 |
+
"logps/chosen": -445.3406677246094,
|
| 87 |
+
"logps/rejected": -201.51806640625,
|
| 88 |
+
"loss": 0.2783,
|
| 89 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 90 |
+
"rewards/chosen": -0.030274126678705215,
|
| 91 |
+
"rewards/margins": 0.8344534635543823,
|
| 92 |
+
"rewards/rejected": -0.8647276163101196,
|
| 93 |
"step": 50
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.15,
|
| 97 |
+
"learning_rate": 4.966461721767899e-07,
|
| 98 |
+
"logits/chosen": -2.383493661880493,
|
| 99 |
+
"logits/rejected": -2.3332314491271973,
|
| 100 |
+
"logps/chosen": -423.5155334472656,
|
| 101 |
+
"logps/rejected": -253.60073852539062,
|
| 102 |
+
"loss": 0.2228,
|
| 103 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 104 |
+
"rewards/chosen": -0.36428865790367126,
|
| 105 |
+
"rewards/margins": 0.920581042766571,
|
| 106 |
+
"rewards/rejected": -1.2848697900772095,
|
| 107 |
"step": 60
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.17,
|
| 111 |
+
"learning_rate": 4.922100518015975e-07,
|
| 112 |
+
"logits/chosen": -2.4096710681915283,
|
| 113 |
+
"logits/rejected": -2.364241123199463,
|
| 114 |
+
"logps/chosen": -422.70513916015625,
|
| 115 |
+
"logps/rejected": -279.1031188964844,
|
| 116 |
+
"loss": 0.1891,
|
| 117 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 118 |
+
"rewards/chosen": -0.3884132504463196,
|
| 119 |
+
"rewards/margins": 1.2157753705978394,
|
| 120 |
+
"rewards/rejected": -1.6041886806488037,
|
| 121 |
"step": 70
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.2,
|
| 125 |
+
"learning_rate": 4.859708325770919e-07,
|
| 126 |
+
"logits/chosen": -2.38008451461792,
|
| 127 |
+
"logits/rejected": -2.344496011734009,
|
| 128 |
+
"logps/chosen": -455.79339599609375,
|
| 129 |
+
"logps/rejected": -303.01690673828125,
|
| 130 |
+
"loss": 0.1269,
|
| 131 |
+
"rewards/accuracies": 0.706250011920929,
|
| 132 |
+
"rewards/chosen": -0.647363543510437,
|
| 133 |
+
"rewards/margins": 1.3260728120803833,
|
| 134 |
+
"rewards/rejected": -1.9734363555908203,
|
| 135 |
"step": 80
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.22,
|
| 139 |
+
"learning_rate": 4.779749614980225e-07,
|
| 140 |
+
"logits/chosen": -2.374379873275757,
|
| 141 |
+
"logits/rejected": -2.3444790840148926,
|
| 142 |
+
"logps/chosen": -532.1400146484375,
|
| 143 |
+
"logps/rejected": -380.60955810546875,
|
| 144 |
+
"loss": 0.1095,
|
| 145 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 146 |
+
"rewards/chosen": -0.560777485370636,
|
| 147 |
+
"rewards/margins": 1.9465181827545166,
|
| 148 |
+
"rewards/rejected": -2.507295846939087,
|
| 149 |
"step": 90
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.25,
|
| 153 |
+
"learning_rate": 4.682819627081427e-07,
|
| 154 |
+
"logits/chosen": -2.3339896202087402,
|
| 155 |
+
"logits/rejected": -2.2830748558044434,
|
| 156 |
+
"logps/chosen": -467.65374755859375,
|
| 157 |
+
"logps/rejected": -348.66156005859375,
|
| 158 |
+
"loss": 0.1477,
|
| 159 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 160 |
+
"rewards/chosen": -0.54865962266922,
|
| 161 |
+
"rewards/margins": 1.753379464149475,
|
| 162 |
+
"rewards/rejected": -2.30203914642334,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 163 |
"step": 100
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 0.27,
|
| 167 |
+
"learning_rate": 4.569639943810477e-07,
|
| 168 |
+
"logits/chosen": -2.3383262157440186,
|
| 169 |
+
"logits/rejected": -2.2902047634124756,
|
| 170 |
+
"logps/chosen": -484.0919494628906,
|
| 171 |
+
"logps/rejected": -367.74505615234375,
|
| 172 |
+
"loss": 0.1193,
|
| 173 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 174 |
+
"rewards/chosen": -0.769287645816803,
|
| 175 |
+
"rewards/margins": 1.7386033535003662,
|
| 176 |
+
"rewards/rejected": -2.5078909397125244,
|
| 177 |
"step": 110
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 0.3,
|
| 181 |
+
"learning_rate": 4.4410531154874543e-07,
|
| 182 |
+
"logits/chosen": -2.3726840019226074,
|
| 183 |
+
"logits/rejected": -2.317364454269409,
|
| 184 |
+
"logps/chosen": -516.4107666015625,
|
| 185 |
+
"logps/rejected": -385.29571533203125,
|
| 186 |
+
"loss": 0.1169,
|
| 187 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 188 |
+
"rewards/chosen": -0.6457246541976929,
|
| 189 |
+
"rewards/margins": 1.9019176959991455,
|
| 190 |
+
"rewards/rejected": -2.547642469406128,
|
| 191 |
"step": 120
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 0.32,
|
| 195 |
+
"learning_rate": 4.298016388768561e-07,
|
| 196 |
+
"logits/chosen": -2.413625717163086,
|
| 197 |
+
"logits/rejected": -2.3727028369903564,
|
| 198 |
+
"logps/chosen": -501.69561767578125,
|
| 199 |
+
"logps/rejected": -364.65997314453125,
|
| 200 |
+
"loss": 0.1158,
|
| 201 |
+
"rewards/accuracies": 0.875,
|
| 202 |
+
"rewards/chosen": -0.36493119597435,
|
| 203 |
+
"rewards/margins": 2.0498671531677246,
|
| 204 |
+
"rewards/rejected": -2.4147982597351074,
|
| 205 |
"step": 130
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 0.35,
|
| 209 |
+
"learning_rate": 4.1415945805573005e-07,
|
| 210 |
+
"logits/chosen": -2.3137855529785156,
|
| 211 |
+
"logits/rejected": -2.2661328315734863,
|
| 212 |
+
"logps/chosen": -494.91546630859375,
|
| 213 |
+
"logps/rejected": -381.90924072265625,
|
| 214 |
+
"loss": 0.1181,
|
| 215 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 216 |
+
"rewards/chosen": -0.7394388914108276,
|
| 217 |
+
"rewards/margins": 1.7537353038787842,
|
| 218 |
+
"rewards/rejected": -2.4931740760803223,
|
| 219 |
"step": 140
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.37,
|
| 223 |
+
"learning_rate": 3.972952151123984e-07,
|
| 224 |
+
"logits/chosen": -2.3150975704193115,
|
| 225 |
+
"logits/rejected": -2.2541096210479736,
|
| 226 |
+
"logps/chosen": -429.13299560546875,
|
| 227 |
+
"logps/rejected": -321.36676025390625,
|
| 228 |
+
"loss": 0.1415,
|
| 229 |
+
"rewards/accuracies": 0.78125,
|
| 230 |
+
"rewards/chosen": -0.5441495180130005,
|
| 231 |
+
"rewards/margins": 1.660264253616333,
|
| 232 |
+
"rewards/rejected": -2.204413890838623,
|
| 233 |
"step": 150
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 0.39,
|
| 237 |
+
"learning_rate": 3.793344535444142e-07,
|
| 238 |
+
"logits/chosen": -2.2927441596984863,
|
| 239 |
+
"logits/rejected": -2.235689401626587,
|
| 240 |
+
"logps/chosen": -517.9212646484375,
|
| 241 |
+
"logps/rejected": -355.93096923828125,
|
| 242 |
+
"loss": 0.1116,
|
| 243 |
+
"rewards/accuracies": 0.793749988079071,
|
| 244 |
+
"rewards/chosen": -0.5073868036270142,
|
| 245 |
+
"rewards/margins": 1.9403215646743774,
|
| 246 |
+
"rewards/rejected": -2.4477083683013916,
|
| 247 |
"step": 160
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 0.42,
|
| 251 |
+
"learning_rate": 3.604108797288461e-07,
|
| 252 |
+
"logits/chosen": -2.2924787998199463,
|
| 253 |
+
"logits/rejected": -2.2269301414489746,
|
| 254 |
+
"logps/chosen": -528.6913452148438,
|
| 255 |
+
"logps/rejected": -407.18780517578125,
|
| 256 |
+
"loss": 0.0933,
|
| 257 |
+
"rewards/accuracies": 0.831250011920929,
|
| 258 |
+
"rewards/chosen": -0.8971579670906067,
|
| 259 |
+
"rewards/margins": 2.0700173377990723,
|
| 260 |
+
"rewards/rejected": -2.9671754837036133,
|
| 261 |
"step": 170
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 0.44,
|
| 265 |
+
"learning_rate": 3.40665367563858e-07,
|
| 266 |
+
"logits/chosen": -2.288649320602417,
|
| 267 |
+
"logits/rejected": -2.230454683303833,
|
| 268 |
+
"logps/chosen": -500.1395568847656,
|
| 269 |
+
"logps/rejected": -393.0097961425781,
|
| 270 |
+
"loss": 0.0995,
|
| 271 |
+
"rewards/accuracies": 0.78125,
|
| 272 |
+
"rewards/chosen": -0.9004061818122864,
|
| 273 |
+
"rewards/margins": 1.8742806911468506,
|
| 274 |
+
"rewards/rejected": -2.774686813354492,
|
| 275 |
"step": 180
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 0.47,
|
| 279 |
+
"learning_rate": 3.202449097526798e-07,
|
| 280 |
+
"logits/chosen": -2.3139257431030273,
|
| 281 |
+
"logits/rejected": -2.2718071937561035,
|
| 282 |
+
"logps/chosen": -465.964111328125,
|
| 283 |
+
"logps/rejected": -354.6561584472656,
|
| 284 |
+
"loss": 0.1167,
|
| 285 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 286 |
+
"rewards/chosen": -0.6382254362106323,
|
| 287 |
+
"rewards/margins": 1.8348891735076904,
|
| 288 |
+
"rewards/rejected": -2.4731147289276123,
|
| 289 |
"step": 190
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 0.49,
|
| 293 |
+
"learning_rate": 2.993015235369905e-07,
|
| 294 |
+
"logits/chosen": -2.2759037017822266,
|
| 295 |
+
"logits/rejected": -2.2043704986572266,
|
| 296 |
+
"logps/chosen": -518.3172607421875,
|
| 297 |
+
"logps/rejected": -417.5638122558594,
|
| 298 |
+
"loss": 0.0996,
|
| 299 |
+
"rewards/accuracies": 0.831250011920929,
|
| 300 |
+
"rewards/chosen": -0.7904411554336548,
|
| 301 |
+
"rewards/margins": 2.2067065238952637,
|
| 302 |
+
"rewards/rejected": -2.997147798538208,
|
| 303 |
"step": 200
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 0.52,
|
| 307 |
+
"learning_rate": 2.7799111902582693e-07,
|
| 308 |
+
"logits/chosen": -2.2700018882751465,
|
| 309 |
+
"logits/rejected": -2.207021951675415,
|
| 310 |
+
"logps/chosen": -489.39801025390625,
|
| 311 |
+
"logps/rejected": -369.2752685546875,
|
| 312 |
+
"loss": 0.1109,
|
| 313 |
+
"rewards/accuracies": 0.78125,
|
| 314 |
+
"rewards/chosen": -0.889120876789093,
|
| 315 |
+
"rewards/margins": 1.7826087474822998,
|
| 316 |
+
"rewards/rejected": -2.671729803085327,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 317 |
"step": 210
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 0.54,
|
| 321 |
+
"learning_rate": 2.564723385445869e-07,
|
| 322 |
+
"logits/chosen": -2.324565887451172,
|
| 323 |
+
"logits/rejected": -2.267853260040283,
|
| 324 |
+
"logps/chosen": -490.59130859375,
|
| 325 |
+
"logps/rejected": -388.76690673828125,
|
| 326 |
+
"loss": 0.1348,
|
| 327 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 328 |
+
"rewards/chosen": -0.7297108769416809,
|
| 329 |
+
"rewards/margins": 1.9148566722869873,
|
| 330 |
+
"rewards/rejected": -2.6445674896240234,
|
| 331 |
"step": 220
|
| 332 |
},
|
| 333 |
{
|
| 334 |
+
"epoch": 0.57,
|
| 335 |
+
"learning_rate": 2.3490537564442845e-07,
|
| 336 |
+
"logits/chosen": -2.267577648162842,
|
| 337 |
+
"logits/rejected": -2.186471462249756,
|
| 338 |
+
"logps/chosen": -497.12335205078125,
|
| 339 |
+
"logps/rejected": -387.3985900878906,
|
| 340 |
+
"loss": 0.1239,
|
| 341 |
+
"rewards/accuracies": 0.731249988079071,
|
| 342 |
+
"rewards/chosen": -1.051777958869934,
|
| 343 |
+
"rewards/margins": 1.7568168640136719,
|
| 344 |
+
"rewards/rejected": -2.8085949420928955,
|
| 345 |
"step": 230
|
| 346 |
},
|
| 347 |
{
|
| 348 |
+
"epoch": 0.59,
|
| 349 |
+
"learning_rate": 2.1345078256378801e-07,
|
| 350 |
+
"logits/chosen": -2.311931610107422,
|
| 351 |
+
"logits/rejected": -2.2356672286987305,
|
| 352 |
+
"logps/chosen": -485.4923400878906,
|
| 353 |
+
"logps/rejected": -404.1107177734375,
|
| 354 |
+
"loss": 0.1251,
|
| 355 |
+
"rewards/accuracies": 0.768750011920929,
|
| 356 |
+
"rewards/chosen": -0.7667996287345886,
|
| 357 |
+
"rewards/margins": 2.0932888984680176,
|
| 358 |
+
"rewards/rejected": -2.860088586807251,
|
| 359 |
"step": 240
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 0.62,
|
| 363 |
+
"learning_rate": 1.9226827501969865e-07,
|
| 364 |
+
"logits/chosen": -2.326636552810669,
|
| 365 |
+
"logits/rejected": -2.2703440189361572,
|
| 366 |
+
"logps/chosen": -509.9071350097656,
|
| 367 |
+
"logps/rejected": -414.39874267578125,
|
| 368 |
+
"loss": 0.1196,
|
| 369 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 370 |
+
"rewards/chosen": -0.6723321676254272,
|
| 371 |
+
"rewards/margins": 2.289646625518799,
|
| 372 |
+
"rewards/rejected": -2.9619784355163574,
|
| 373 |
"step": 250
|
| 374 |
},
|
| 375 |
{
|
| 376 |
+
"epoch": 0.64,
|
| 377 |
+
"learning_rate": 1.715155432264775e-07,
|
| 378 |
+
"logits/chosen": -2.3231379985809326,
|
| 379 |
+
"logits/rejected": -2.2714104652404785,
|
| 380 |
+
"logps/chosen": -511.625732421875,
|
| 381 |
+
"logps/rejected": -397.1949157714844,
|
| 382 |
+
"loss": 0.104,
|
| 383 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 384 |
+
"rewards/chosen": -0.7959054708480835,
|
| 385 |
+
"rewards/margins": 2.012528896331787,
|
| 386 |
+
"rewards/rejected": -2.80843448638916,
|
| 387 |
"step": 260
|
| 388 |
},
|
| 389 |
{
|
| 390 |
+
"epoch": 0.67,
|
| 391 |
+
"learning_rate": 1.51347077992983e-07,
|
| 392 |
+
"logits/chosen": -2.3125240802764893,
|
| 393 |
+
"logits/rejected": -2.269193410873413,
|
| 394 |
+
"logps/chosen": -496.95001220703125,
|
| 395 |
+
"logps/rejected": -402.70782470703125,
|
| 396 |
+
"loss": 0.0976,
|
| 397 |
+
"rewards/accuracies": 0.78125,
|
| 398 |
+
"rewards/chosen": -0.9325113296508789,
|
| 399 |
+
"rewards/margins": 1.8711084127426147,
|
| 400 |
+
"rewards/rejected": -2.803619861602783,
|
| 401 |
"step": 270
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 0.69,
|
| 405 |
+
"learning_rate": 1.3191302063739906e-07,
|
| 406 |
+
"logits/chosen": -2.282691717147827,
|
| 407 |
+
"logits/rejected": -2.233623743057251,
|
| 408 |
+
"logps/chosen": -469.02630615234375,
|
| 409 |
+
"logps/rejected": -393.0565185546875,
|
| 410 |
+
"loss": 0.0978,
|
| 411 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 412 |
+
"rewards/chosen": -0.8983050584793091,
|
| 413 |
+
"rewards/margins": 1.9013340473175049,
|
| 414 |
+
"rewards/rejected": -2.7996389865875244,
|
| 415 |
"step": 280
|
| 416 |
},
|
| 417 |
{
|
| 418 |
+
"epoch": 0.72,
|
| 419 |
+
"learning_rate": 1.1335804528119475e-07,
|
| 420 |
+
"logits/chosen": -2.3497660160064697,
|
| 421 |
+
"logits/rejected": -2.2719693183898926,
|
| 422 |
+
"logps/chosen": -515.9114990234375,
|
| 423 |
+
"logps/rejected": -393.408935546875,
|
| 424 |
+
"loss": 0.1062,
|
| 425 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 426 |
+
"rewards/chosen": -0.8015244603157043,
|
| 427 |
+
"rewards/margins": 2.1254332065582275,
|
| 428 |
+
"rewards/rejected": -2.926957607269287,
|
| 429 |
"step": 290
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 0.74,
|
| 433 |
+
"learning_rate": 9.582028184286423e-08,
|
| 434 |
+
"logits/chosen": -2.2509076595306396,
|
| 435 |
+
"logits/rejected": -2.2074227333068848,
|
| 436 |
+
"logps/chosen": -466.32305908203125,
|
| 437 |
+
"logps/rejected": -419.3622131347656,
|
| 438 |
+
"loss": 0.1048,
|
| 439 |
+
"rewards/accuracies": 0.75,
|
| 440 |
+
"rewards/chosen": -1.0646774768829346,
|
| 441 |
+
"rewards/margins": 1.9130761623382568,
|
| 442 |
+
"rewards/rejected": -2.9777536392211914,
|
| 443 |
"step": 300
|
| 444 |
},
|
| 445 |
{
|
| 446 |
+
"epoch": 0.76,
|
| 447 |
+
"learning_rate": 7.943028774907065e-08,
|
| 448 |
+
"logits/chosen": -2.250230312347412,
|
| 449 |
+
"logits/rejected": -2.195244789123535,
|
| 450 |
+
"logps/chosen": -476.79998779296875,
|
| 451 |
+
"logps/rejected": -408.64068603515625,
|
| 452 |
+
"loss": 0.1044,
|
| 453 |
+
"rewards/accuracies": 0.768750011920929,
|
| 454 |
+
"rewards/chosen": -0.923287034034729,
|
| 455 |
+
"rewards/margins": 1.9779703617095947,
|
| 456 |
+
"rewards/rejected": -2.901257276535034,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 457 |
"step": 310
|
| 458 |
},
|
| 459 |
{
|
| 460 |
+
"epoch": 0.79,
|
| 461 |
+
"learning_rate": 6.431007601814637e-08,
|
| 462 |
+
"logits/chosen": -2.2714295387268066,
|
| 463 |
+
"logits/rejected": -2.2263479232788086,
|
| 464 |
+
"logps/chosen": -431.47601318359375,
|
| 465 |
+
"logps/rejected": -397.78521728515625,
|
| 466 |
+
"loss": 0.0946,
|
| 467 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 468 |
+
"rewards/chosen": -1.0377047061920166,
|
| 469 |
+
"rewards/margins": 1.8672853708267212,
|
| 470 |
+
"rewards/rejected": -2.9049899578094482,
|
| 471 |
"step": 320
|
| 472 |
},
|
| 473 |
{
|
| 474 |
+
"epoch": 0.81,
|
| 475 |
+
"learning_rate": 5.0572206951246e-08,
|
| 476 |
+
"logits/chosen": -2.243610382080078,
|
| 477 |
+
"logits/rejected": -2.1739163398742676,
|
| 478 |
+
"logps/chosen": -469.3233337402344,
|
| 479 |
+
"logps/rejected": -400.6234130859375,
|
| 480 |
+
"loss": 0.1024,
|
| 481 |
+
"rewards/accuracies": 0.78125,
|
| 482 |
+
"rewards/chosen": -1.0177555084228516,
|
| 483 |
+
"rewards/margins": 1.9254471063613892,
|
| 484 |
+
"rewards/rejected": -2.943202495574951,
|
| 485 |
"step": 330
|
| 486 |
},
|
| 487 |
{
|
| 488 |
+
"epoch": 0.84,
|
| 489 |
+
"learning_rate": 3.831895019292897e-08,
|
| 490 |
+
"logits/chosen": -2.302597761154175,
|
| 491 |
+
"logits/rejected": -2.245525360107422,
|
| 492 |
+
"logps/chosen": -527.4336547851562,
|
| 493 |
+
"logps/rejected": -453.01116943359375,
|
| 494 |
+
"loss": 0.1089,
|
| 495 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 496 |
+
"rewards/chosen": -0.9302037954330444,
|
| 497 |
+
"rewards/margins": 2.4225857257843018,
|
| 498 |
+
"rewards/rejected": -3.3527894020080566,
|
| 499 |
"step": 340
|
| 500 |
},
|
| 501 |
{
|
| 502 |
+
"epoch": 0.86,
|
| 503 |
+
"learning_rate": 2.764152339909756e-08,
|
| 504 |
+
"logits/chosen": -2.2584733963012695,
|
| 505 |
+
"logits/rejected": -2.1962692737579346,
|
| 506 |
+
"logps/chosen": -499.47576904296875,
|
| 507 |
+
"logps/rejected": -363.5857238769531,
|
| 508 |
+
"loss": 0.1036,
|
| 509 |
+
"rewards/accuracies": 0.8125,
|
| 510 |
+
"rewards/chosen": -0.7420647144317627,
|
| 511 |
+
"rewards/margins": 1.9064794778823853,
|
| 512 |
+
"rewards/rejected": -2.6485438346862793,
|
| 513 |
"step": 350
|
| 514 |
},
|
| 515 |
{
|
| 516 |
+
"epoch": 0.89,
|
| 517 |
+
"learning_rate": 1.861941317991664e-08,
|
| 518 |
+
"logits/chosen": -2.300192356109619,
|
| 519 |
+
"logits/rejected": -2.2127695083618164,
|
| 520 |
+
"logps/chosen": -531.5907592773438,
|
| 521 |
+
"logps/rejected": -412.7438049316406,
|
| 522 |
+
"loss": 0.1013,
|
| 523 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 524 |
+
"rewards/chosen": -0.7529584765434265,
|
| 525 |
+
"rewards/margins": 2.228832960128784,
|
| 526 |
+
"rewards/rejected": -2.9817919731140137,
|
| 527 |
"step": 360
|
| 528 |
},
|
| 529 |
{
|
| 530 |
+
"epoch": 0.91,
|
| 531 |
+
"learning_rate": 1.13197833728636e-08,
|
| 532 |
+
"logits/chosen": -2.25828218460083,
|
| 533 |
+
"logits/rejected": -2.198098659515381,
|
| 534 |
+
"logps/chosen": -485.60052490234375,
|
| 535 |
+
"logps/rejected": -423.1136779785156,
|
| 536 |
+
"loss": 0.095,
|
| 537 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 538 |
+
"rewards/chosen": -0.8292403221130371,
|
| 539 |
+
"rewards/margins": 2.2823524475097656,
|
| 540 |
+
"rewards/rejected": -3.1115927696228027,
|
| 541 |
"step": 370
|
| 542 |
},
|
| 543 |
{
|
| 544 |
+
"epoch": 0.94,
|
| 545 |
+
"learning_rate": 5.79697505093521e-09,
|
| 546 |
+
"logits/chosen": -2.2520506381988525,
|
| 547 |
+
"logits/rejected": -2.1905932426452637,
|
| 548 |
+
"logps/chosen": -501.0511169433594,
|
| 549 |
+
"logps/rejected": -402.864501953125,
|
| 550 |
+
"loss": 0.1084,
|
| 551 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 552 |
+
"rewards/chosen": -0.9875413775444031,
|
| 553 |
+
"rewards/margins": 1.987672209739685,
|
| 554 |
+
"rewards/rejected": -2.9752135276794434,
|
| 555 |
"step": 380
|
| 556 |
},
|
| 557 |
{
|
| 558 |
+
"epoch": 0.96,
|
| 559 |
+
"learning_rate": 2.092101988131256e-09,
|
| 560 |
+
"logits/chosen": -2.2959237098693848,
|
| 561 |
+
"logits/rejected": -2.1943726539611816,
|
| 562 |
+
"logps/chosen": -542.3916015625,
|
| 563 |
+
"logps/rejected": -426.2535095214844,
|
| 564 |
+
"loss": 0.092,
|
| 565 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 566 |
+
"rewards/chosen": -0.7925306558609009,
|
| 567 |
+
"rewards/margins": 2.3787803649902344,
|
| 568 |
+
"rewards/rejected": -3.1713109016418457,
|
| 569 |
"step": 390
|
| 570 |
},
|
| 571 |
{
|
| 572 |
+
"epoch": 0.99,
|
| 573 |
+
"learning_rate": 2.327445937151673e-10,
|
| 574 |
+
"logits/chosen": -2.2838008403778076,
|
| 575 |
+
"logits/rejected": -2.2221412658691406,
|
| 576 |
+
"logps/chosen": -538.8756103515625,
|
| 577 |
+
"logps/rejected": -445.8970642089844,
|
| 578 |
+
"loss": 0.0998,
|
| 579 |
+
"rewards/accuracies": 0.824999988079071,
|
| 580 |
+
"rewards/chosen": -0.9025734066963196,
|
| 581 |
+
"rewards/margins": 2.2714860439300537,
|
| 582 |
+
"rewards/rejected": -3.1740598678588867,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 583 |
"step": 400
|
| 584 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 585 |
{
|
| 586 |
"epoch": 1.0,
|
| 587 |
+
"step": 405,
|
| 588 |
"total_flos": 0.0,
|
| 589 |
+
"train_loss": 0.1562451661368947,
|
| 590 |
+
"train_runtime": 3209.2418,
|
| 591 |
+
"train_samples_per_second": 16.17,
|
| 592 |
+
"train_steps_per_second": 0.126
|
| 593 |
}
|
| 594 |
],
|
| 595 |
"logging_steps": 10,
|
| 596 |
+
"max_steps": 405,
|
| 597 |
"num_train_epochs": 1,
|
| 598 |
"save_steps": 1000,
|
| 599 |
"total_flos": 0.0,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5944
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9ee5499d0ce2b903ad76a9c35cc9bf4c38d09435955bae297aa02a619c188930
|
| 3 |
size 5944
|