Model save
Browse files- README.md +6 -12
- 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 +789 -69
README.md
CHANGED
|
@@ -2,15 +2,10 @@
|
|
| 2 |
license: mit
|
| 3 |
base_model: HuggingFaceH4/mistral-7b-sft-beta
|
| 4 |
tags:
|
| 5 |
-
- alignment-handbook
|
| 6 |
-
- trl
|
| 7 |
-
- dpo
|
| 8 |
-
- generated_from_trainer
|
| 9 |
- trl
|
| 10 |
- dpo
|
|
|
|
| 11 |
- generated_from_trainer
|
| 12 |
-
datasets:
|
| 13 |
-
- HuggingFaceH4/ultrafeedback_binarized
|
| 14 |
model-index:
|
| 15 |
- name: DPO-Zephyr-7B
|
| 16 |
results: []
|
|
@@ -21,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 21 |
|
| 22 |
# DPO-Zephyr-7B
|
| 23 |
|
| 24 |
-
This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on
|
| 25 |
|
| 26 |
## Model description
|
| 27 |
|
|
@@ -41,14 +36,13 @@ More information needed
|
|
| 41 |
|
| 42 |
The following hyperparameters were used during training:
|
| 43 |
- learning_rate: 5e-07
|
| 44 |
-
- train_batch_size:
|
| 45 |
-
- eval_batch_size:
|
| 46 |
- seed: 42
|
| 47 |
- distributed_type: multi-GPU
|
| 48 |
- num_devices: 3
|
| 49 |
-
-
|
| 50 |
-
-
|
| 51 |
-
- total_eval_batch_size: 24
|
| 52 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 53 |
- lr_scheduler_type: cosine
|
| 54 |
- lr_scheduler_warmup_ratio: 0.1
|
|
|
|
| 2 |
license: mit
|
| 3 |
base_model: HuggingFaceH4/mistral-7b-sft-beta
|
| 4 |
tags:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
- trl
|
| 6 |
- dpo
|
| 7 |
+
- alignment-handbook
|
| 8 |
- generated_from_trainer
|
|
|
|
|
|
|
| 9 |
model-index:
|
| 10 |
- name: DPO-Zephyr-7B
|
| 11 |
results: []
|
|
|
|
| 16 |
|
| 17 |
# DPO-Zephyr-7B
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on an unknown dataset.
|
| 20 |
|
| 21 |
## Model description
|
| 22 |
|
|
|
|
| 36 |
|
| 37 |
The following hyperparameters were used during training:
|
| 38 |
- learning_rate: 5e-07
|
| 39 |
+
- train_batch_size: 3
|
| 40 |
+
- eval_batch_size: 4
|
| 41 |
- seed: 42
|
| 42 |
- distributed_type: multi-GPU
|
| 43 |
- num_devices: 3
|
| 44 |
+
- total_train_batch_size: 9
|
| 45 |
+
- total_eval_batch_size: 12
|
|
|
|
| 46 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
- lr_scheduler_type: cosine
|
| 48 |
- lr_scheduler_warmup_ratio: 0.1
|
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 0
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 15283,
|
| 7 |
-
"train_samples_per_second": 3.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.3818549155440398,
|
| 5 |
+
"train_runtime": 4714.6907,
|
| 6 |
"train_samples": 15283,
|
| 7 |
+
"train_samples_per_second": 3.242,
|
| 8 |
+
"train_steps_per_second": 0.135
|
| 9 |
}
|
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:ac26dbef8ff942f79ddbc3c5b131cdc9fa2d10ae43fab47f2377c9b7de2fccef
|
| 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:c927851423f252a01b03cf8090580227fde5515df1c249c6f6198324520d072c
|
| 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:f784959e0b5911b50bf26eb3e5c2da47b1203fe94661e2c520a44209a9af3fc1
|
| 3 |
size 4540516344
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 0
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 15283,
|
| 7 |
-
"train_samples_per_second": 3.
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.3818549155440398,
|
| 5 |
+
"train_runtime": 4714.6907,
|
| 6 |
"train_samples": 15283,
|
| 7 |
+
"train_samples_per_second": 3.242,
|
| 8 |
+
"train_steps_per_second": 0.135
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 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,
|
|
@@ -174,92 +174,812 @@
|
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
-
"epoch": 0.
|
| 178 |
-
"grad_norm":
|
| 179 |
-
"learning_rate":
|
| 180 |
-
"logits/chosen": -2.
|
| 181 |
-
"logits/rejected": -2.
|
| 182 |
-
"logps/chosen": -
|
| 183 |
-
"logps/rejected": -
|
| 184 |
-
"loss": 0.
|
| 185 |
-
"rewards/accuracies": 0.
|
| 186 |
-
"rewards/chosen": -0.
|
| 187 |
-
"rewards/margins": 0.
|
| 188 |
-
"rewards/rejected": -
|
| 189 |
"step": 110
|
| 190 |
},
|
| 191 |
{
|
| 192 |
-
"epoch": 0.
|
| 193 |
-
"grad_norm":
|
| 194 |
-
"learning_rate":
|
| 195 |
-
"logits/chosen": -2.
|
| 196 |
-
"logits/rejected": -2.
|
| 197 |
-
"logps/chosen": -
|
| 198 |
-
"logps/rejected": -
|
| 199 |
-
"loss": 0.
|
| 200 |
-
"rewards/accuracies": 0.
|
| 201 |
-
"rewards/chosen": -0.
|
| 202 |
-
"rewards/margins": 0.
|
| 203 |
-
"rewards/rejected": -
|
| 204 |
"step": 120
|
| 205 |
},
|
| 206 |
{
|
| 207 |
-
"epoch": 0.
|
| 208 |
-
"grad_norm":
|
| 209 |
-
"learning_rate": 4.
|
| 210 |
-
"logits/chosen": -2.
|
| 211 |
-
"logits/rejected": -2.
|
| 212 |
-
"logps/chosen": -
|
| 213 |
-
"logps/rejected": -
|
| 214 |
-
"loss": 0.
|
| 215 |
-
"rewards/accuracies": 0.
|
| 216 |
-
"rewards/chosen": -0.
|
| 217 |
-
"rewards/margins": 0.
|
| 218 |
-
"rewards/rejected": -
|
| 219 |
"step": 130
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"grad_norm":
|
| 224 |
-
"learning_rate":
|
| 225 |
-
"logits/chosen": -2.
|
| 226 |
-
"logits/rejected": -2.
|
| 227 |
-
"logps/chosen": -
|
| 228 |
-
"logps/rejected": -
|
| 229 |
-
"loss": 0.
|
| 230 |
-
"rewards/accuracies": 0.
|
| 231 |
-
"rewards/chosen": -0.
|
| 232 |
-
"rewards/margins": 0.
|
| 233 |
-
"rewards/rejected": -
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"grad_norm":
|
| 239 |
-
"learning_rate": 4.
|
| 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": 0.
|
| 248 |
-
"rewards/rejected": -
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
-
"epoch": 0.
|
| 253 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 254 |
"total_flos": 0.0,
|
| 255 |
-
"train_loss": 0.
|
| 256 |
-
"train_runtime":
|
| 257 |
-
"train_samples_per_second": 3.
|
| 258 |
-
"train_steps_per_second": 0.
|
| 259 |
}
|
| 260 |
],
|
| 261 |
"logging_steps": 10,
|
| 262 |
-
"max_steps":
|
| 263 |
"num_input_tokens_seen": 0,
|
| 264 |
"num_train_epochs": 1,
|
| 265 |
"save_steps": 100,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 637,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 174 |
"step": 100
|
| 175 |
},
|
| 176 |
{
|
| 177 |
+
"epoch": 0.1726844583987441,
|
| 178 |
+
"grad_norm": 31.95579631076908,
|
| 179 |
+
"learning_rate": 4.920911573406924e-07,
|
| 180 |
+
"logits/chosen": -2.737246513366699,
|
| 181 |
+
"logits/rejected": -2.694401979446411,
|
| 182 |
+
"logps/chosen": -324.00128173828125,
|
| 183 |
+
"logps/rejected": -355.02178955078125,
|
| 184 |
+
"loss": 0.4353,
|
| 185 |
+
"rewards/accuracies": 0.887499988079071,
|
| 186 |
+
"rewards/chosen": -0.3587312698364258,
|
| 187 |
+
"rewards/margins": 0.7477051615715027,
|
| 188 |
+
"rewards/rejected": -1.1064363718032837,
|
| 189 |
"step": 110
|
| 190 |
},
|
| 191 |
{
|
| 192 |
+
"epoch": 0.18838304552590268,
|
| 193 |
+
"grad_norm": 35.71890678681308,
|
| 194 |
+
"learning_rate": 4.883087164434672e-07,
|
| 195 |
+
"logits/chosen": -2.626753568649292,
|
| 196 |
+
"logits/rejected": -2.6289310455322266,
|
| 197 |
+
"logps/chosen": -319.3797302246094,
|
| 198 |
+
"logps/rejected": -375.67864990234375,
|
| 199 |
+
"loss": 0.4172,
|
| 200 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 201 |
+
"rewards/chosen": -0.4024926722049713,
|
| 202 |
+
"rewards/margins": 0.8855185508728027,
|
| 203 |
+
"rewards/rejected": -1.2880111932754517,
|
| 204 |
"step": 120
|
| 205 |
},
|
| 206 |
{
|
| 207 |
+
"epoch": 0.20408163265306123,
|
| 208 |
+
"grad_norm": 44.90656496644609,
|
| 209 |
+
"learning_rate": 4.838100964592904e-07,
|
| 210 |
+
"logits/chosen": -2.5899486541748047,
|
| 211 |
+
"logits/rejected": -2.5745112895965576,
|
| 212 |
+
"logps/chosen": -272.218994140625,
|
| 213 |
+
"logps/rejected": -347.6167297363281,
|
| 214 |
+
"loss": 0.4832,
|
| 215 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 216 |
+
"rewards/chosen": -0.4248787760734558,
|
| 217 |
+
"rewards/margins": 0.8124195337295532,
|
| 218 |
+
"rewards/rejected": -1.2372982501983643,
|
| 219 |
"step": 130
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.21978021978021978,
|
| 223 |
+
"grad_norm": 76.16900405992267,
|
| 224 |
+
"learning_rate": 4.786088169001671e-07,
|
| 225 |
+
"logits/chosen": -2.5720114707946777,
|
| 226 |
+
"logits/rejected": -2.5637192726135254,
|
| 227 |
+
"logps/chosen": -289.96759033203125,
|
| 228 |
+
"logps/rejected": -435.85308837890625,
|
| 229 |
+
"loss": 0.3981,
|
| 230 |
+
"rewards/accuracies": 0.824999988079071,
|
| 231 |
+
"rewards/chosen": -0.3887278437614441,
|
| 232 |
+
"rewards/margins": 0.9565574526786804,
|
| 233 |
+
"rewards/rejected": -1.3452852964401245,
|
| 234 |
"step": 140
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.23547880690737832,
|
| 238 |
+
"grad_norm": 70.20692587613685,
|
| 239 |
+
"learning_rate": 4.727205089511466e-07,
|
| 240 |
+
"logits/chosen": -2.5425446033477783,
|
| 241 |
+
"logits/rejected": -2.525315523147583,
|
| 242 |
+
"logps/chosen": -313.8908996582031,
|
| 243 |
+
"logps/rejected": -377.4201354980469,
|
| 244 |
+
"loss": 0.4486,
|
| 245 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 246 |
+
"rewards/chosen": -0.7373483180999756,
|
| 247 |
+
"rewards/margins": 0.8309048414230347,
|
| 248 |
+
"rewards/rejected": -1.5682532787322998,
|
| 249 |
"step": 150
|
| 250 |
},
|
| 251 |
{
|
| 252 |
+
"epoch": 0.25117739403453687,
|
| 253 |
+
"grad_norm": 49.06765782836144,
|
| 254 |
+
"learning_rate": 4.661628684945851e-07,
|
| 255 |
+
"logits/chosen": -2.573646068572998,
|
| 256 |
+
"logits/rejected": -2.5104339122772217,
|
| 257 |
+
"logps/chosen": -370.089599609375,
|
| 258 |
+
"logps/rejected": -390.0255126953125,
|
| 259 |
+
"loss": 0.4023,
|
| 260 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 261 |
+
"rewards/chosen": -0.8807455897331238,
|
| 262 |
+
"rewards/margins": 0.7309385538101196,
|
| 263 |
+
"rewards/rejected": -1.6116840839385986,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.2668759811616955,
|
| 268 |
+
"grad_norm": 54.18143326196735,
|
| 269 |
+
"learning_rate": 4.5895560292945996e-07,
|
| 270 |
+
"logits/chosen": -2.5094046592712402,
|
| 271 |
+
"logits/rejected": -2.519862413406372,
|
| 272 |
+
"logps/chosen": -328.95098876953125,
|
| 273 |
+
"logps/rejected": -409.6293029785156,
|
| 274 |
+
"loss": 0.4229,
|
| 275 |
+
"rewards/accuracies": 0.887499988079071,
|
| 276 |
+
"rewards/chosen": -0.8252763748168945,
|
| 277 |
+
"rewards/margins": 0.8684083223342896,
|
| 278 |
+
"rewards/rejected": -1.6936845779418945,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.282574568288854,
|
| 283 |
+
"grad_norm": 47.86539813882727,
|
| 284 |
+
"learning_rate": 4.5112037194555876e-07,
|
| 285 |
+
"logits/chosen": -2.4061732292175293,
|
| 286 |
+
"logits/rejected": -2.336141586303711,
|
| 287 |
+
"logps/chosen": -340.967529296875,
|
| 288 |
+
"logps/rejected": -416.18438720703125,
|
| 289 |
+
"loss": 0.4129,
|
| 290 |
+
"rewards/accuracies": 0.862500011920929,
|
| 291 |
+
"rewards/chosen": -0.9469585418701172,
|
| 292 |
+
"rewards/margins": 1.003477692604065,
|
| 293 |
+
"rewards/rejected": -1.9504363536834717,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.29827315541601257,
|
| 298 |
+
"grad_norm": 47.40203923720411,
|
| 299 |
+
"learning_rate": 4.426807224305315e-07,
|
| 300 |
+
"logits/chosen": -2.434738874435425,
|
| 301 |
+
"logits/rejected": -2.4182467460632324,
|
| 302 |
+
"logps/chosen": -390.0809631347656,
|
| 303 |
+
"logps/rejected": -541.5980224609375,
|
| 304 |
+
"loss": 0.39,
|
| 305 |
+
"rewards/accuracies": 0.925000011920929,
|
| 306 |
+
"rewards/chosen": -0.907140851020813,
|
| 307 |
+
"rewards/margins": 1.477752447128296,
|
| 308 |
+
"rewards/rejected": -2.3848931789398193,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.3139717425431711,
|
| 313 |
+
"grad_norm": 59.480022233916614,
|
| 314 |
+
"learning_rate": 4.3366201770542687e-07,
|
| 315 |
+
"logits/chosen": -2.3925621509552,
|
| 316 |
+
"logits/rejected": -2.3594958782196045,
|
| 317 |
+
"logps/chosen": -399.2648010253906,
|
| 318 |
+
"logps/rejected": -485.82061767578125,
|
| 319 |
+
"loss": 0.3722,
|
| 320 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 321 |
+
"rewards/chosen": -1.242919683456421,
|
| 322 |
+
"rewards/margins": 1.0100038051605225,
|
| 323 |
+
"rewards/rejected": -2.2529234886169434,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.32967032967032966,
|
| 328 |
+
"grad_norm": 54.533637965345,
|
| 329 |
+
"learning_rate": 4.2409136130137845e-07,
|
| 330 |
+
"logits/chosen": -2.391516923904419,
|
| 331 |
+
"logits/rejected": -2.3367998600006104,
|
| 332 |
+
"logps/chosen": -410.27752685546875,
|
| 333 |
+
"logps/rejected": -488.3935546875,
|
| 334 |
+
"loss": 0.3723,
|
| 335 |
+
"rewards/accuracies": 0.824999988079071,
|
| 336 |
+
"rewards/chosen": -1.2089805603027344,
|
| 337 |
+
"rewards/margins": 1.3162455558776855,
|
| 338 |
+
"rewards/rejected": -2.52522611618042,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.3453689167974882,
|
| 343 |
+
"grad_norm": 134.3137263377747,
|
| 344 |
+
"learning_rate": 4.1399751550651084e-07,
|
| 345 |
+
"logits/chosen": -2.2474474906921387,
|
| 346 |
+
"logits/rejected": -2.273486852645874,
|
| 347 |
+
"logps/chosen": -422.23419189453125,
|
| 348 |
+
"logps/rejected": -493.7196350097656,
|
| 349 |
+
"loss": 0.4143,
|
| 350 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 351 |
+
"rewards/chosen": -1.3034000396728516,
|
| 352 |
+
"rewards/margins": 1.1734570264816284,
|
| 353 |
+
"rewards/rejected": -2.4768571853637695,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.36106750392464676,
|
| 358 |
+
"grad_norm": 51.06322946002842,
|
| 359 |
+
"learning_rate": 4.034108149278543e-07,
|
| 360 |
+
"logits/chosen": -2.246279716491699,
|
| 361 |
+
"logits/rejected": -2.218812942504883,
|
| 362 |
+
"logps/chosen": -396.13433837890625,
|
| 363 |
+
"logps/rejected": -494.9988708496094,
|
| 364 |
+
"loss": 0.3583,
|
| 365 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 366 |
+
"rewards/chosen": -1.4342329502105713,
|
| 367 |
+
"rewards/margins": 1.3676201105117798,
|
| 368 |
+
"rewards/rejected": -2.8018529415130615,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.37676609105180536,
|
| 373 |
+
"grad_norm": 53.106943652386406,
|
| 374 |
+
"learning_rate": 3.923630753280357e-07,
|
| 375 |
+
"logits/chosen": -2.296950578689575,
|
| 376 |
+
"logits/rejected": -2.193643808364868,
|
| 377 |
+
"logps/chosen": -446.6080017089844,
|
| 378 |
+
"logps/rejected": -504.3912658691406,
|
| 379 |
+
"loss": 0.439,
|
| 380 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 381 |
+
"rewards/chosen": -1.5059573650360107,
|
| 382 |
+
"rewards/margins": 1.111060380935669,
|
| 383 |
+
"rewards/rejected": -2.617017984390259,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.3924646781789639,
|
| 388 |
+
"grad_norm": 31.823002355070333,
|
| 389 |
+
"learning_rate": 3.8088749801071496e-07,
|
| 390 |
+
"logits/chosen": -2.179251194000244,
|
| 391 |
+
"logits/rejected": -2.139425277709961,
|
| 392 |
+
"logps/chosen": -373.7418518066406,
|
| 393 |
+
"logps/rejected": -501.843505859375,
|
| 394 |
+
"loss": 0.369,
|
| 395 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 396 |
+
"rewards/chosen": -1.2382627725601196,
|
| 397 |
+
"rewards/margins": 1.2532708644866943,
|
| 398 |
+
"rewards/rejected": -2.4915337562561035,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.40816326530612246,
|
| 403 |
+
"grad_norm": 57.958160333752495,
|
| 404 |
+
"learning_rate": 3.6901857004211443e-07,
|
| 405 |
+
"logits/chosen": -2.2685585021972656,
|
| 406 |
+
"logits/rejected": -2.1836700439453125,
|
| 407 |
+
"logps/chosen": -438.3352966308594,
|
| 408 |
+
"logps/rejected": -474.6566467285156,
|
| 409 |
+
"loss": 0.3729,
|
| 410 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 411 |
+
"rewards/chosen": -1.387054443359375,
|
| 412 |
+
"rewards/margins": 1.218056559562683,
|
| 413 |
+
"rewards/rejected": -2.6051113605499268,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.423861852433281,
|
| 418 |
+
"grad_norm": 78.46346160026032,
|
| 419 |
+
"learning_rate": 3.5679196060850034e-07,
|
| 420 |
+
"logits/chosen": -2.194472312927246,
|
| 421 |
+
"logits/rejected": -2.1565122604370117,
|
| 422 |
+
"logps/chosen": -470.94024658203125,
|
| 423 |
+
"logps/rejected": -557.4560546875,
|
| 424 |
+
"loss": 0.3988,
|
| 425 |
+
"rewards/accuracies": 0.800000011920929,
|
| 426 |
+
"rewards/chosen": -1.681726098060608,
|
| 427 |
+
"rewards/margins": 1.0995349884033203,
|
| 428 |
+
"rewards/rejected": -2.7812609672546387,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.43956043956043955,
|
| 433 |
+
"grad_norm": 73.42817989092451,
|
| 434 |
+
"learning_rate": 3.4424441382108826e-07,
|
| 435 |
+
"logits/chosen": -2.0928378105163574,
|
| 436 |
+
"logits/rejected": -2.080376148223877,
|
| 437 |
+
"logps/chosen": -437.2099609375,
|
| 438 |
+
"logps/rejected": -575.6405029296875,
|
| 439 |
+
"loss": 0.3881,
|
| 440 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 441 |
+
"rewards/chosen": -1.617456078529358,
|
| 442 |
+
"rewards/margins": 1.3479641675949097,
|
| 443 |
+
"rewards/rejected": -2.9654202461242676,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.4552590266875981,
|
| 448 |
+
"grad_norm": 62.6558696654114,
|
| 449 |
+
"learning_rate": 3.314136382905234e-07,
|
| 450 |
+
"logits/chosen": -2.0131213665008545,
|
| 451 |
+
"logits/rejected": -1.9851818084716797,
|
| 452 |
+
"logps/chosen": -512.6057739257812,
|
| 453 |
+
"logps/rejected": -646.832275390625,
|
| 454 |
+
"loss": 0.4216,
|
| 455 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 456 |
+
"rewards/chosen": -2.3018593788146973,
|
| 457 |
+
"rewards/margins": 1.346274971961975,
|
| 458 |
+
"rewards/rejected": -3.648134708404541,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.47095761381475665,
|
| 463 |
+
"grad_norm": 78.12598055564801,
|
| 464 |
+
"learning_rate": 3.1833819380279023e-07,
|
| 465 |
+
"logits/chosen": -2.0949795246124268,
|
| 466 |
+
"logits/rejected": -1.9874067306518555,
|
| 467 |
+
"logps/chosen": -488.44384765625,
|
| 468 |
+
"logps/rejected": -571.2259521484375,
|
| 469 |
+
"loss": 0.386,
|
| 470 |
+
"rewards/accuracies": 0.8125,
|
| 471 |
+
"rewards/chosen": -2.06516695022583,
|
| 472 |
+
"rewards/margins": 1.3252384662628174,
|
| 473 |
+
"rewards/rejected": -3.3904056549072266,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.48665620094191525,
|
| 478 |
+
"grad_norm": 48.25836081307725,
|
| 479 |
+
"learning_rate": 3.0505737543712275e-07,
|
| 480 |
+
"logits/chosen": -2.1759631633758545,
|
| 481 |
+
"logits/rejected": -2.095853805541992,
|
| 482 |
+
"logps/chosen": -437.9419860839844,
|
| 483 |
+
"logps/rejected": -518.4359130859375,
|
| 484 |
+
"loss": 0.3903,
|
| 485 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 486 |
+
"rewards/chosen": -1.5794050693511963,
|
| 487 |
+
"rewards/margins": 1.1850736141204834,
|
| 488 |
+
"rewards/rejected": -2.7644786834716797,
|
| 489 |
+
"step": 310
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.5023547880690737,
|
| 493 |
+
"grad_norm": 60.78980701157502,
|
| 494 |
+
"learning_rate": 2.9161109547416667e-07,
|
| 495 |
+
"logits/chosen": -2.1254727840423584,
|
| 496 |
+
"logits/rejected": -2.0983071327209473,
|
| 497 |
+
"logps/chosen": -487.24383544921875,
|
| 498 |
+
"logps/rejected": -561.2527465820312,
|
| 499 |
+
"loss": 0.3585,
|
| 500 |
+
"rewards/accuracies": 0.862500011920929,
|
| 501 |
+
"rewards/chosen": -1.5642211437225342,
|
| 502 |
+
"rewards/margins": 1.3017809391021729,
|
| 503 |
+
"rewards/rejected": -2.866001844406128,
|
| 504 |
+
"step": 320
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.5180533751962323,
|
| 508 |
+
"grad_norm": 59.58938057948085,
|
| 509 |
+
"learning_rate": 2.780397634492949e-07,
|
| 510 |
+
"logits/chosen": -2.1095099449157715,
|
| 511 |
+
"logits/rejected": -2.0623815059661865,
|
| 512 |
+
"logps/chosen": -420.601318359375,
|
| 513 |
+
"logps/rejected": -561.5743408203125,
|
| 514 |
+
"loss": 0.4054,
|
| 515 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 516 |
+
"rewards/chosen": -1.7670698165893555,
|
| 517 |
+
"rewards/margins": 1.3623578548431396,
|
| 518 |
+
"rewards/rejected": -3.129427671432495,
|
| 519 |
+
"step": 330
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.533751962323391,
|
| 523 |
+
"grad_norm": 59.60421047453409,
|
| 524 |
+
"learning_rate": 2.6438416471154273e-07,
|
| 525 |
+
"logits/chosen": -2.105985641479492,
|
| 526 |
+
"logits/rejected": -2.1100611686706543,
|
| 527 |
+
"logps/chosen": -455.4019470214844,
|
| 528 |
+
"logps/rejected": -597.48583984375,
|
| 529 |
+
"loss": 0.3559,
|
| 530 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 531 |
+
"rewards/chosen": -1.826162338256836,
|
| 532 |
+
"rewards/margins": 1.1689088344573975,
|
| 533 |
+
"rewards/rejected": -2.9950711727142334,
|
| 534 |
+
"step": 340
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.5494505494505495,
|
| 538 |
+
"grad_norm": 64.80810897640455,
|
| 539 |
+
"learning_rate": 2.5068533785312666e-07,
|
| 540 |
+
"logits/chosen": -2.1501102447509766,
|
| 541 |
+
"logits/rejected": -2.098646640777588,
|
| 542 |
+
"logps/chosen": -398.44293212890625,
|
| 543 |
+
"logps/rejected": -512.3727416992188,
|
| 544 |
+
"loss": 0.3817,
|
| 545 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 546 |
+
"rewards/chosen": -1.6598011255264282,
|
| 547 |
+
"rewards/margins": 1.3402740955352783,
|
| 548 |
+
"rewards/rejected": -3.000075101852417,
|
| 549 |
+
"step": 350
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.565149136577708,
|
| 553 |
+
"grad_norm": 66.30622337743506,
|
| 554 |
+
"learning_rate": 2.3698445137790258e-07,
|
| 555 |
+
"logits/chosen": -2.183788776397705,
|
| 556 |
+
"logits/rejected": -2.1399548053741455,
|
| 557 |
+
"logps/chosen": -450.28790283203125,
|
| 558 |
+
"logps/rejected": -570.9381103515625,
|
| 559 |
+
"loss": 0.3441,
|
| 560 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 561 |
+
"rewards/chosen": -1.7852160930633545,
|
| 562 |
+
"rewards/margins": 1.4079563617706299,
|
| 563 |
+
"rewards/rejected": -3.1931726932525635,
|
| 564 |
+
"step": 360
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.5808477237048666,
|
| 568 |
+
"grad_norm": 74.385047879014,
|
| 569 |
+
"learning_rate": 2.2332267997940513e-07,
|
| 570 |
+
"logits/chosen": -1.9558976888656616,
|
| 571 |
+
"logits/rejected": -1.8889505863189697,
|
| 572 |
+
"logps/chosen": -472.29046630859375,
|
| 573 |
+
"logps/rejected": -564.1910400390625,
|
| 574 |
+
"loss": 0.4144,
|
| 575 |
+
"rewards/accuracies": 0.8125,
|
| 576 |
+
"rewards/chosen": -2.128002405166626,
|
| 577 |
+
"rewards/margins": 1.2228187322616577,
|
| 578 |
+
"rewards/rejected": -3.3508212566375732,
|
| 579 |
+
"step": 370
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.5965463108320251,
|
| 583 |
+
"grad_norm": 98.38326705187197,
|
| 584 |
+
"learning_rate": 2.0974108080028692e-07,
|
| 585 |
+
"logits/chosen": -2.1048827171325684,
|
| 586 |
+
"logits/rejected": -2.014101266860962,
|
| 587 |
+
"logps/chosen": -472.6429138183594,
|
| 588 |
+
"logps/rejected": -552.962646484375,
|
| 589 |
+
"loss": 0.362,
|
| 590 |
+
"rewards/accuracies": 0.887499988079071,
|
| 591 |
+
"rewards/chosen": -1.7709888219833374,
|
| 592 |
+
"rewards/margins": 1.5231949090957642,
|
| 593 |
+
"rewards/rejected": -3.2941837310791016,
|
| 594 |
+
"step": 380
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.6122448979591837,
|
| 598 |
+
"grad_norm": 28.798007236746653,
|
| 599 |
+
"learning_rate": 1.962804700450265e-07,
|
| 600 |
+
"logits/chosen": -2.229259490966797,
|
| 601 |
+
"logits/rejected": -2.136399269104004,
|
| 602 |
+
"logps/chosen": -404.88970947265625,
|
| 603 |
+
"logps/rejected": -605.4417724609375,
|
| 604 |
+
"loss": 0.2913,
|
| 605 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 606 |
+
"rewards/chosen": -1.31179940700531,
|
| 607 |
+
"rewards/margins": 2.022050142288208,
|
| 608 |
+
"rewards/rejected": -3.3338496685028076,
|
| 609 |
+
"step": 390
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.6279434850863422,
|
| 613 |
+
"grad_norm": 47.93461485983235,
|
| 614 |
+
"learning_rate": 1.8298130031671972e-07,
|
| 615 |
+
"logits/chosen": -2.066481351852417,
|
| 616 |
+
"logits/rejected": -2.0052218437194824,
|
| 617 |
+
"logps/chosen": -461.60711669921875,
|
| 618 |
+
"logps/rejected": -593.0902099609375,
|
| 619 |
+
"loss": 0.3182,
|
| 620 |
+
"rewards/accuracies": 0.887499988079071,
|
| 621 |
+
"rewards/chosen": -1.50743567943573,
|
| 622 |
+
"rewards/margins": 1.5388964414596558,
|
| 623 |
+
"rewards/rejected": -3.0463321208953857,
|
| 624 |
+
"step": 400
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.6436420722135008,
|
| 628 |
+
"grad_norm": 81.64181086704542,
|
| 629 |
+
"learning_rate": 1.6988353904658492e-07,
|
| 630 |
+
"logits/chosen": -2.079695224761963,
|
| 631 |
+
"logits/rejected": -2.0234522819519043,
|
| 632 |
+
"logps/chosen": -435.9869079589844,
|
| 633 |
+
"logps/rejected": -548.7467041015625,
|
| 634 |
+
"loss": 0.5391,
|
| 635 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 636 |
+
"rewards/chosen": -1.8887078762054443,
|
| 637 |
+
"rewards/margins": 1.0673691034317017,
|
| 638 |
+
"rewards/rejected": -2.9560768604278564,
|
| 639 |
+
"step": 410
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.6593406593406593,
|
| 643 |
+
"grad_norm": 99.80504939653643,
|
| 644 |
+
"learning_rate": 1.570265483815364e-07,
|
| 645 |
+
"logits/chosen": -2.146688222885132,
|
| 646 |
+
"logits/rejected": -2.1056787967681885,
|
| 647 |
+
"logps/chosen": -441.99456787109375,
|
| 648 |
+
"logps/rejected": -486.8045959472656,
|
| 649 |
+
"loss": 0.6064,
|
| 650 |
+
"rewards/accuracies": 0.625,
|
| 651 |
+
"rewards/chosen": -1.862978219985962,
|
| 652 |
+
"rewards/margins": 0.5048335790634155,
|
| 653 |
+
"rewards/rejected": -2.367811679840088,
|
| 654 |
+
"step": 420
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.6750392464678179,
|
| 658 |
+
"grad_norm": 85.01436657258617,
|
| 659 |
+
"learning_rate": 1.444489668907914e-07,
|
| 660 |
+
"logits/chosen": -2.0182952880859375,
|
| 661 |
+
"logits/rejected": -2.0226969718933105,
|
| 662 |
+
"logps/chosen": -346.7263488769531,
|
| 663 |
+
"logps/rejected": -505.09619140625,
|
| 664 |
+
"loss": 0.5198,
|
| 665 |
+
"rewards/accuracies": 0.800000011920929,
|
| 666 |
+
"rewards/chosen": -1.6009681224822998,
|
| 667 |
+
"rewards/margins": 1.235584020614624,
|
| 668 |
+
"rewards/rejected": -2.836552143096924,
|
| 669 |
+
"step": 430
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.6907378335949764,
|
| 673 |
+
"grad_norm": 75.62069078299841,
|
| 674 |
+
"learning_rate": 1.3218859344701632e-07,
|
| 675 |
+
"logits/chosen": -2.2084121704101562,
|
| 676 |
+
"logits/rejected": -2.1915016174316406,
|
| 677 |
+
"logps/chosen": -412.80224609375,
|
| 678 |
+
"logps/rejected": -515.23095703125,
|
| 679 |
+
"loss": 0.5238,
|
| 680 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 681 |
+
"rewards/chosen": -1.4952690601348877,
|
| 682 |
+
"rewards/margins": 1.0581085681915283,
|
| 683 |
+
"rewards/rejected": -2.553377628326416,
|
| 684 |
+
"step": 440
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.706436420722135,
|
| 688 |
+
"grad_norm": 150.50609049436147,
|
| 689 |
+
"learning_rate": 1.202822736309758e-07,
|
| 690 |
+
"logits/chosen": -2.1147706508636475,
|
| 691 |
+
"logits/rejected": -2.0880684852600098,
|
| 692 |
+
"logps/chosen": -442.6726989746094,
|
| 693 |
+
"logps/rejected": -475.3531799316406,
|
| 694 |
+
"loss": 0.5647,
|
| 695 |
+
"rewards/accuracies": 0.75,
|
| 696 |
+
"rewards/chosen": -1.6886441707611084,
|
| 697 |
+
"rewards/margins": 0.8048279881477356,
|
| 698 |
+
"rewards/rejected": -2.4934723377227783,
|
| 699 |
+
"step": 450
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.7221350078492935,
|
| 703 |
+
"grad_norm": 76.56164006248804,
|
| 704 |
+
"learning_rate": 1.0876578900107053e-07,
|
| 705 |
+
"logits/chosen": -2.2221930027008057,
|
| 706 |
+
"logits/rejected": -2.1485908031463623,
|
| 707 |
+
"logps/chosen": -464.043701171875,
|
| 708 |
+
"logps/rejected": -508.6534118652344,
|
| 709 |
+
"loss": 0.522,
|
| 710 |
+
"rewards/accuracies": 0.6875,
|
| 711 |
+
"rewards/chosen": -1.763662338256836,
|
| 712 |
+
"rewards/margins": 0.7875553965568542,
|
| 713 |
+
"rewards/rejected": -2.551217555999756,
|
| 714 |
+
"step": 460
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.7378335949764521,
|
| 718 |
+
"grad_norm": 74.89323927902005,
|
| 719 |
+
"learning_rate": 9.767374956053584e-08,
|
| 720 |
+
"logits/chosen": -2.205277919769287,
|
| 721 |
+
"logits/rejected": -2.18218994140625,
|
| 722 |
+
"logps/chosen": -412.97686767578125,
|
| 723 |
+
"logps/rejected": -543.1309204101562,
|
| 724 |
+
"loss": 0.5596,
|
| 725 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 726 |
+
"rewards/chosen": -1.553432822227478,
|
| 727 |
+
"rewards/margins": 0.9423764944076538,
|
| 728 |
+
"rewards/rejected": -2.495809316635132,
|
| 729 |
+
"step": 470
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.7535321821036107,
|
| 733 |
+
"grad_norm": 87.37555939932336,
|
| 734 |
+
"learning_rate": 8.70394897454659e-08,
|
| 735 |
+
"logits/chosen": -2.2252628803253174,
|
| 736 |
+
"logits/rejected": -2.1848928928375244,
|
| 737 |
+
"logps/chosen": -382.27752685546875,
|
| 738 |
+
"logps/rejected": -457.525390625,
|
| 739 |
+
"loss": 0.547,
|
| 740 |
+
"rewards/accuracies": 0.612500011920929,
|
| 741 |
+
"rewards/chosen": -1.6248447895050049,
|
| 742 |
+
"rewards/margins": 0.4326212406158447,
|
| 743 |
+
"rewards/rejected": -2.0574657917022705,
|
| 744 |
+
"step": 480
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.7692307692307693,
|
| 748 |
+
"grad_norm": 93.79826007492892,
|
| 749 |
+
"learning_rate": 7.689496824624525e-08,
|
| 750 |
+
"logits/chosen": -2.2677321434020996,
|
| 751 |
+
"logits/rejected": -2.212900161743164,
|
| 752 |
+
"logps/chosen": -403.1687316894531,
|
| 753 |
+
"logps/rejected": -456.62615966796875,
|
| 754 |
+
"loss": 0.5459,
|
| 755 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 756 |
+
"rewards/chosen": -1.4313546419143677,
|
| 757 |
+
"rewards/margins": 0.6928239464759827,
|
| 758 |
+
"rewards/rejected": -2.124178647994995,
|
| 759 |
+
"step": 490
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.7849293563579278,
|
| 763 |
+
"grad_norm": 56.75850068630019,
|
| 764 |
+
"learning_rate": 6.727067196345099e-08,
|
| 765 |
+
"logits/chosen": -2.3276655673980713,
|
| 766 |
+
"logits/rejected": -2.2189202308654785,
|
| 767 |
+
"logps/chosen": -438.74267578125,
|
| 768 |
+
"logps/rejected": -478.62298583984375,
|
| 769 |
+
"loss": 0.5453,
|
| 770 |
+
"rewards/accuracies": 0.75,
|
| 771 |
+
"rewards/chosen": -1.3515713214874268,
|
| 772 |
+
"rewards/margins": 1.0120915174484253,
|
| 773 |
+
"rewards/rejected": -2.3636624813079834,
|
| 774 |
+
"step": 500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.8006279434850864,
|
| 778 |
+
"grad_norm": 97.05649014340479,
|
| 779 |
+
"learning_rate": 5.8195524386862374e-08,
|
| 780 |
+
"logits/chosen": -2.311316967010498,
|
| 781 |
+
"logits/rejected": -2.242011547088623,
|
| 782 |
+
"logps/chosen": -362.28399658203125,
|
| 783 |
+
"logps/rejected": -465.5743103027344,
|
| 784 |
+
"loss": 0.4852,
|
| 785 |
+
"rewards/accuracies": 0.800000011920929,
|
| 786 |
+
"rewards/chosen": -1.409887433052063,
|
| 787 |
+
"rewards/margins": 0.9168035387992859,
|
| 788 |
+
"rewards/rejected": -2.326690912246704,
|
| 789 |
+
"step": 510
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.8163265306122449,
|
| 793 |
+
"grad_norm": 127.5887173764836,
|
| 794 |
+
"learning_rate": 4.969679867292276e-08,
|
| 795 |
+
"logits/chosen": -2.2007040977478027,
|
| 796 |
+
"logits/rejected": -2.138988971710205,
|
| 797 |
+
"logps/chosen": -413.65447998046875,
|
| 798 |
+
"logps/rejected": -489.22540283203125,
|
| 799 |
+
"loss": 0.4985,
|
| 800 |
+
"rewards/accuracies": 0.762499988079071,
|
| 801 |
+
"rewards/chosen": -1.3157858848571777,
|
| 802 |
+
"rewards/margins": 1.0105135440826416,
|
| 803 |
+
"rewards/rejected": -2.3262994289398193,
|
| 804 |
+
"step": 520
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.8320251177394035,
|
| 808 |
+
"grad_norm": 92.31568060854485,
|
| 809 |
+
"learning_rate": 4.180003568187776e-08,
|
| 810 |
+
"logits/chosen": -2.2751996517181396,
|
| 811 |
+
"logits/rejected": -2.2204182147979736,
|
| 812 |
+
"logps/chosen": -437.9576110839844,
|
| 813 |
+
"logps/rejected": -478.0186462402344,
|
| 814 |
+
"loss": 0.5193,
|
| 815 |
+
"rewards/accuracies": 0.699999988079071,
|
| 816 |
+
"rewards/chosen": -1.3832905292510986,
|
| 817 |
+
"rewards/margins": 0.656178891658783,
|
| 818 |
+
"rewards/rejected": -2.0394692420959473,
|
| 819 |
+
"step": 530
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.847723704866562,
|
| 823 |
+
"grad_norm": 93.73800900430595,
|
| 824 |
+
"learning_rate": 3.452896722091128e-08,
|
| 825 |
+
"logits/chosen": -2.2862913608551025,
|
| 826 |
+
"logits/rejected": -2.1965749263763428,
|
| 827 |
+
"logps/chosen": -418.68218994140625,
|
| 828 |
+
"logps/rejected": -488.07208251953125,
|
| 829 |
+
"loss": 0.549,
|
| 830 |
+
"rewards/accuracies": 0.762499988079071,
|
| 831 |
+
"rewards/chosen": -1.4643903970718384,
|
| 832 |
+
"rewards/margins": 0.8227276802062988,
|
| 833 |
+
"rewards/rejected": -2.2871181964874268,
|
| 834 |
+
"step": 540
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.8634222919937206,
|
| 838 |
+
"grad_norm": 74.68277239397396,
|
| 839 |
+
"learning_rate": 2.7905444723949762e-08,
|
| 840 |
+
"logits/chosen": -2.2326064109802246,
|
| 841 |
+
"logits/rejected": -2.1876773834228516,
|
| 842 |
+
"logps/chosen": -434.93377685546875,
|
| 843 |
+
"logps/rejected": -486.3379821777344,
|
| 844 |
+
"loss": 0.5464,
|
| 845 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 846 |
+
"rewards/chosen": -1.5688042640686035,
|
| 847 |
+
"rewards/margins": 0.6943384408950806,
|
| 848 |
+
"rewards/rejected": -2.2631428241729736,
|
| 849 |
+
"step": 550
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.8791208791208791,
|
| 853 |
+
"grad_norm": 98.95496114421825,
|
| 854 |
+
"learning_rate": 2.194937358247506e-08,
|
| 855 |
+
"logits/chosen": -2.314251661300659,
|
| 856 |
+
"logits/rejected": -2.282222032546997,
|
| 857 |
+
"logps/chosen": -444.4891662597656,
|
| 858 |
+
"logps/rejected": -482.2521057128906,
|
| 859 |
+
"loss": 0.5058,
|
| 860 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 861 |
+
"rewards/chosen": -1.4890687465667725,
|
| 862 |
+
"rewards/margins": 0.7982088327407837,
|
| 863 |
+
"rewards/rejected": -2.2872776985168457,
|
| 864 |
+
"step": 560
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.8948194662480377,
|
| 868 |
+
"grad_norm": 93.93456697380334,
|
| 869 |
+
"learning_rate": 1.6678653324693787e-08,
|
| 870 |
+
"logits/chosen": -2.271334171295166,
|
| 871 |
+
"logits/rejected": -2.243417263031006,
|
| 872 |
+
"logps/chosen": -379.41131591796875,
|
| 873 |
+
"logps/rejected": -474.8644104003906,
|
| 874 |
+
"loss": 0.5554,
|
| 875 |
+
"rewards/accuracies": 0.737500011920929,
|
| 876 |
+
"rewards/chosen": -1.2463254928588867,
|
| 877 |
+
"rewards/margins": 1.0109448432922363,
|
| 878 |
+
"rewards/rejected": -2.257270336151123,
|
| 879 |
+
"step": 570
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.9105180533751962,
|
| 883 |
+
"grad_norm": 63.908735732483215,
|
| 884 |
+
"learning_rate": 1.2109123822844653e-08,
|
| 885 |
+
"logits/chosen": -2.2195041179656982,
|
| 886 |
+
"logits/rejected": -2.152104139328003,
|
| 887 |
+
"logps/chosen": -424.376220703125,
|
| 888 |
+
"logps/rejected": -493.041748046875,
|
| 889 |
+
"loss": 0.5407,
|
| 890 |
+
"rewards/accuracies": 0.75,
|
| 891 |
+
"rewards/chosen": -1.484989047050476,
|
| 892 |
+
"rewards/margins": 0.8357736468315125,
|
| 893 |
+
"rewards/rejected": -2.3207626342773438,
|
| 894 |
+
"step": 580
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.9262166405023547,
|
| 898 |
+
"grad_norm": 77.29245256601979,
|
| 899 |
+
"learning_rate": 8.254517690300944e-09,
|
| 900 |
+
"logits/chosen": -2.1921186447143555,
|
| 901 |
+
"logits/rejected": -2.160266876220703,
|
| 902 |
+
"logps/chosen": -358.38665771484375,
|
| 903 |
+
"logps/rejected": -419.24053955078125,
|
| 904 |
+
"loss": 0.5065,
|
| 905 |
+
"rewards/accuracies": 0.699999988079071,
|
| 906 |
+
"rewards/chosen": -1.4891427755355835,
|
| 907 |
+
"rewards/margins": 0.6564360857009888,
|
| 908 |
+
"rewards/rejected": -2.1455788612365723,
|
| 909 |
+
"step": 590
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.9419152276295133,
|
| 913 |
+
"grad_norm": 85.49976100967837,
|
| 914 |
+
"learning_rate": 5.126419011529992e-09,
|
| 915 |
+
"logits/chosen": -2.247999668121338,
|
| 916 |
+
"logits/rejected": -2.1932740211486816,
|
| 917 |
+
"logps/chosen": -437.494384765625,
|
| 918 |
+
"logps/rejected": -485.1595764160156,
|
| 919 |
+
"loss": 0.4877,
|
| 920 |
+
"rewards/accuracies": 0.737500011920929,
|
| 921 |
+
"rewards/chosen": -1.302431583404541,
|
| 922 |
+
"rewards/margins": 0.9456756711006165,
|
| 923 |
+
"rewards/rejected": -2.2481071949005127,
|
| 924 |
+
"step": 600
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.957613814756672,
|
| 928 |
+
"grad_norm": 107.98710002534753,
|
| 929 |
+
"learning_rate": 2.734228528934679e-09,
|
| 930 |
+
"logits/chosen": -2.124533176422119,
|
| 931 |
+
"logits/rejected": -2.1044251918792725,
|
| 932 |
+
"logps/chosen": -403.1171569824219,
|
| 933 |
+
"logps/rejected": -488.7078552246094,
|
| 934 |
+
"loss": 0.5816,
|
| 935 |
+
"rewards/accuracies": 0.6875,
|
| 936 |
+
"rewards/chosen": -1.370384931564331,
|
| 937 |
+
"rewards/margins": 0.671573281288147,
|
| 938 |
+
"rewards/rejected": -2.0419580936431885,
|
| 939 |
+
"step": 610
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.9733124018838305,
|
| 943 |
+
"grad_norm": 62.944406149509454,
|
| 944 |
+
"learning_rate": 1.0851353912008642e-09,
|
| 945 |
+
"logits/chosen": -2.2792160511016846,
|
| 946 |
+
"logits/rejected": -2.1857845783233643,
|
| 947 |
+
"logps/chosen": -420.2891540527344,
|
| 948 |
+
"logps/rejected": -431.2037658691406,
|
| 949 |
+
"loss": 0.5459,
|
| 950 |
+
"rewards/accuracies": 0.737500011920929,
|
| 951 |
+
"rewards/chosen": -1.3576745986938477,
|
| 952 |
+
"rewards/margins": 0.5345790982246399,
|
| 953 |
+
"rewards/rejected": -1.8922535181045532,
|
| 954 |
+
"step": 620
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.989010989010989,
|
| 958 |
+
"grad_norm": 81.15926557617168,
|
| 959 |
+
"learning_rate": 1.840955480532924e-10,
|
| 960 |
+
"logits/chosen": -2.2514820098876953,
|
| 961 |
+
"logits/rejected": -2.2048940658569336,
|
| 962 |
+
"logps/chosen": -372.85504150390625,
|
| 963 |
+
"logps/rejected": -425.1669921875,
|
| 964 |
+
"loss": 0.5177,
|
| 965 |
+
"rewards/accuracies": 0.762499988079071,
|
| 966 |
+
"rewards/chosen": -1.1938732862472534,
|
| 967 |
+
"rewards/margins": 0.8971937894821167,
|
| 968 |
+
"rewards/rejected": -2.091067314147949,
|
| 969 |
+
"step": 630
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.0,
|
| 973 |
+
"step": 637,
|
| 974 |
"total_flos": 0.0,
|
| 975 |
+
"train_loss": 0.3818549155440398,
|
| 976 |
+
"train_runtime": 4714.6907,
|
| 977 |
+
"train_samples_per_second": 3.242,
|
| 978 |
+
"train_steps_per_second": 0.135
|
| 979 |
}
|
| 980 |
],
|
| 981 |
"logging_steps": 10,
|
| 982 |
+
"max_steps": 637,
|
| 983 |
"num_input_tokens_seen": 0,
|
| 984 |
"num_train_epochs": 1,
|
| 985 |
"save_steps": 100,
|