Model save
Browse files- README.md +2 -2
- all_results.json +9 -0
- generation_config.json +14 -0
- train_results.json +9 -0
- trainer_state.json +2076 -0
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -4,8 +4,8 @@ library_name: transformers
|
|
| 4 |
model_name: Qwen2.5-14B-Instruct-ultrafeedback-spin-iter1-RPO
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
| 7 |
-
- trl
|
| 8 |
- dpo
|
|
|
|
| 9 |
licence: license
|
| 10 |
---
|
| 11 |
|
|
@@ -27,7 +27,7 @@ print(output["generated_text"])
|
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
-
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/yifanwang/huggingface/runs/
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
|
|
|
|
| 4 |
model_name: Qwen2.5-14B-Instruct-ultrafeedback-spin-iter1-RPO
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
|
|
|
| 7 |
- dpo
|
| 8 |
+
- trl
|
| 9 |
licence: license
|
| 10 |
---
|
| 11 |
|
|
|
|
| 27 |
|
| 28 |
## Training procedure
|
| 29 |
|
| 30 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/yifanwang/huggingface/runs/xyh1s3qx)
|
| 31 |
|
| 32 |
|
| 33 |
This model was trained with DPO, a method introduced in [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://huggingface.co/papers/2305.18290).
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.0,
|
| 5 |
+
"train_runtime": 0.984,
|
| 6 |
+
"train_samples": 9999,
|
| 7 |
+
"train_samples_per_second": 10161.128,
|
| 8 |
+
"train_steps_per_second": 1270.268
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 151643,
|
| 9 |
+
"repetition_penalty": 1.05,
|
| 10 |
+
"temperature": 0.7,
|
| 11 |
+
"top_k": 20,
|
| 12 |
+
"top_p": 0.8,
|
| 13 |
+
"transformers_version": "4.53.3"
|
| 14 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.0,
|
| 5 |
+
"train_runtime": 0.984,
|
| 6 |
+
"train_samples": 9999,
|
| 7 |
+
"train_samples_per_second": 10161.128,
|
| 8 |
+
"train_steps_per_second": 1270.268
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2076 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1250,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0008,
|
| 14 |
+
"grad_norm": 226.62270447693234,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"logits/chosen": -0.91796875,
|
| 17 |
+
"logits/rejected": -1.0625,
|
| 18 |
+
"logps/chosen": -448.0,
|
| 19 |
+
"logps/rejected": -404.0,
|
| 20 |
+
"loss": 0.6934,
|
| 21 |
+
"nll_loss": 2.625,
|
| 22 |
+
"rewards/accuracies": 0.0,
|
| 23 |
+
"rewards/chosen": 0.0,
|
| 24 |
+
"rewards/margins": 0.0,
|
| 25 |
+
"rewards/rejected": 0.0,
|
| 26 |
+
"step": 1
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.008,
|
| 30 |
+
"grad_norm": 183.46648450042892,
|
| 31 |
+
"learning_rate": 3.6e-08,
|
| 32 |
+
"logits/chosen": -0.9357638955116272,
|
| 33 |
+
"logits/rejected": -1.0073784589767456,
|
| 34 |
+
"logps/chosen": -528.4444580078125,
|
| 35 |
+
"logps/rejected": -375.77777099609375,
|
| 36 |
+
"loss": 0.7075,
|
| 37 |
+
"nll_loss": 1.7126736640930176,
|
| 38 |
+
"rewards/accuracies": 0.3055555522441864,
|
| 39 |
+
"rewards/chosen": -0.0180528424680233,
|
| 40 |
+
"rewards/margins": -0.006869846023619175,
|
| 41 |
+
"rewards/rejected": -0.011135525070130825,
|
| 42 |
+
"step": 10
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.016,
|
| 46 |
+
"grad_norm": 195.8050525947057,
|
| 47 |
+
"learning_rate": 7.599999999999999e-08,
|
| 48 |
+
"logits/chosen": -0.8521484136581421,
|
| 49 |
+
"logits/rejected": -1.0,
|
| 50 |
+
"logps/chosen": -412.79998779296875,
|
| 51 |
+
"logps/rejected": -408.6000061035156,
|
| 52 |
+
"loss": 0.6575,
|
| 53 |
+
"nll_loss": 2.35546875,
|
| 54 |
+
"rewards/accuracies": 0.5,
|
| 55 |
+
"rewards/chosen": -0.005053711123764515,
|
| 56 |
+
"rewards/margins": 0.09255371242761612,
|
| 57 |
+
"rewards/rejected": -0.09763183444738388,
|
| 58 |
+
"step": 20
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.024,
|
| 62 |
+
"grad_norm": 141.87495839689677,
|
| 63 |
+
"learning_rate": 1.16e-07,
|
| 64 |
+
"logits/chosen": -0.848437488079071,
|
| 65 |
+
"logits/rejected": -0.8382812738418579,
|
| 66 |
+
"logps/chosen": -521.5999755859375,
|
| 67 |
+
"logps/rejected": -380.20001220703125,
|
| 68 |
+
"loss": 0.5052,
|
| 69 |
+
"nll_loss": 1.78515625,
|
| 70 |
+
"rewards/accuracies": 0.862500011920929,
|
| 71 |
+
"rewards/chosen": 0.14396972954273224,
|
| 72 |
+
"rewards/margins": 0.4627929627895355,
|
| 73 |
+
"rewards/rejected": -0.31914061307907104,
|
| 74 |
+
"step": 30
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.032,
|
| 78 |
+
"grad_norm": 125.39420097312498,
|
| 79 |
+
"learning_rate": 1.56e-07,
|
| 80 |
+
"logits/chosen": -0.8394531011581421,
|
| 81 |
+
"logits/rejected": -0.969921886920929,
|
| 82 |
+
"logps/chosen": -529.0,
|
| 83 |
+
"logps/rejected": -401.6000061035156,
|
| 84 |
+
"loss": 0.3009,
|
| 85 |
+
"nll_loss": 1.8390624523162842,
|
| 86 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 87 |
+
"rewards/chosen": 0.3280273377895355,
|
| 88 |
+
"rewards/margins": 1.265039086341858,
|
| 89 |
+
"rewards/rejected": -0.9378906488418579,
|
| 90 |
+
"step": 40
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.04,
|
| 94 |
+
"grad_norm": 51.38963253406093,
|
| 95 |
+
"learning_rate": 1.96e-07,
|
| 96 |
+
"logits/chosen": -0.811328113079071,
|
| 97 |
+
"logits/rejected": -0.971484363079071,
|
| 98 |
+
"logps/chosen": -459.20001220703125,
|
| 99 |
+
"logps/rejected": -411.6000061035156,
|
| 100 |
+
"loss": 0.179,
|
| 101 |
+
"nll_loss": 2.192187547683716,
|
| 102 |
+
"rewards/accuracies": 0.987500011920929,
|
| 103 |
+
"rewards/chosen": 0.33916014432907104,
|
| 104 |
+
"rewards/margins": 2.1109375953674316,
|
| 105 |
+
"rewards/rejected": -1.771875023841858,
|
| 106 |
+
"step": 50
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 0.048,
|
| 110 |
+
"grad_norm": 41.54844453932047,
|
| 111 |
+
"learning_rate": 2.3599999999999997e-07,
|
| 112 |
+
"logits/chosen": -0.710888683795929,
|
| 113 |
+
"logits/rejected": -0.852734386920929,
|
| 114 |
+
"logps/chosen": -450.20001220703125,
|
| 115 |
+
"logps/rejected": -423.3999938964844,
|
| 116 |
+
"loss": 0.0706,
|
| 117 |
+
"nll_loss": 2.2906250953674316,
|
| 118 |
+
"rewards/accuracies": 1.0,
|
| 119 |
+
"rewards/chosen": -0.0625,
|
| 120 |
+
"rewards/margins": 3.278125047683716,
|
| 121 |
+
"rewards/rejected": -3.340625047683716,
|
| 122 |
+
"step": 60
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.056,
|
| 126 |
+
"grad_norm": 2.8975377631106283,
|
| 127 |
+
"learning_rate": 2.7600000000000004e-07,
|
| 128 |
+
"logits/chosen": -0.756054699420929,
|
| 129 |
+
"logits/rejected": -0.7701171636581421,
|
| 130 |
+
"logps/chosen": -454.79998779296875,
|
| 131 |
+
"logps/rejected": -425.79998779296875,
|
| 132 |
+
"loss": 0.0485,
|
| 133 |
+
"nll_loss": 2.09375,
|
| 134 |
+
"rewards/accuracies": 1.0,
|
| 135 |
+
"rewards/chosen": -0.03144531324505806,
|
| 136 |
+
"rewards/margins": 4.420312404632568,
|
| 137 |
+
"rewards/rejected": -4.453125,
|
| 138 |
+
"step": 70
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.064,
|
| 142 |
+
"grad_norm": 1.9819883761762331,
|
| 143 |
+
"learning_rate": 3.1599999999999997e-07,
|
| 144 |
+
"logits/chosen": -0.6548827886581421,
|
| 145 |
+
"logits/rejected": -0.584765613079071,
|
| 146 |
+
"logps/chosen": -516.2000122070312,
|
| 147 |
+
"logps/rejected": -442.0,
|
| 148 |
+
"loss": 0.0136,
|
| 149 |
+
"nll_loss": 1.96875,
|
| 150 |
+
"rewards/accuracies": 1.0,
|
| 151 |
+
"rewards/chosen": 0.05791015550494194,
|
| 152 |
+
"rewards/margins": 5.881249904632568,
|
| 153 |
+
"rewards/rejected": -5.818749904632568,
|
| 154 |
+
"step": 80
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.072,
|
| 158 |
+
"grad_norm": 40.33622312322741,
|
| 159 |
+
"learning_rate": 3.5599999999999996e-07,
|
| 160 |
+
"logits/chosen": -0.5289062261581421,
|
| 161 |
+
"logits/rejected": -0.5241454839706421,
|
| 162 |
+
"logps/chosen": -460.3999938964844,
|
| 163 |
+
"logps/rejected": -450.3999938964844,
|
| 164 |
+
"loss": 0.0126,
|
| 165 |
+
"nll_loss": 2.077343702316284,
|
| 166 |
+
"rewards/accuracies": 1.0,
|
| 167 |
+
"rewards/chosen": 0.30329591035842896,
|
| 168 |
+
"rewards/margins": 6.993750095367432,
|
| 169 |
+
"rewards/rejected": -6.690625190734863,
|
| 170 |
+
"step": 90
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.08,
|
| 174 |
+
"grad_norm": 3.381840804687941,
|
| 175 |
+
"learning_rate": 3.96e-07,
|
| 176 |
+
"logits/chosen": -0.6162109375,
|
| 177 |
+
"logits/rejected": -0.5665038824081421,
|
| 178 |
+
"logps/chosen": -516.5999755859375,
|
| 179 |
+
"logps/rejected": -461.6000061035156,
|
| 180 |
+
"loss": 0.0103,
|
| 181 |
+
"nll_loss": 2.063281297683716,
|
| 182 |
+
"rewards/accuracies": 1.0,
|
| 183 |
+
"rewards/chosen": 0.47880858182907104,
|
| 184 |
+
"rewards/margins": 8.15625,
|
| 185 |
+
"rewards/rejected": -7.675000190734863,
|
| 186 |
+
"step": 100
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.088,
|
| 190 |
+
"grad_norm": 0.022971788286646016,
|
| 191 |
+
"learning_rate": 4.36e-07,
|
| 192 |
+
"logits/chosen": -0.555468738079071,
|
| 193 |
+
"logits/rejected": -0.558398425579071,
|
| 194 |
+
"logps/chosen": -509.6000061035156,
|
| 195 |
+
"logps/rejected": -469.3999938964844,
|
| 196 |
+
"loss": 0.0069,
|
| 197 |
+
"nll_loss": 1.841406226158142,
|
| 198 |
+
"rewards/accuracies": 1.0,
|
| 199 |
+
"rewards/chosen": 0.911425769329071,
|
| 200 |
+
"rewards/margins": 9.612500190734863,
|
| 201 |
+
"rewards/rejected": -8.693750381469727,
|
| 202 |
+
"step": 110
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.096,
|
| 206 |
+
"grad_norm": 0.6089138686903683,
|
| 207 |
+
"learning_rate": 4.76e-07,
|
| 208 |
+
"logits/chosen": -0.43085938692092896,
|
| 209 |
+
"logits/rejected": -0.44287109375,
|
| 210 |
+
"logps/chosen": -460.3999938964844,
|
| 211 |
+
"logps/rejected": -492.6000061035156,
|
| 212 |
+
"loss": 0.0024,
|
| 213 |
+
"nll_loss": 1.9765625,
|
| 214 |
+
"rewards/accuracies": 1.0,
|
| 215 |
+
"rewards/chosen": 1.300390601158142,
|
| 216 |
+
"rewards/margins": 10.537500381469727,
|
| 217 |
+
"rewards/rejected": -9.240625381469727,
|
| 218 |
+
"step": 120
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.104,
|
| 222 |
+
"grad_norm": 5.072145285053,
|
| 223 |
+
"learning_rate": 4.982222222222223e-07,
|
| 224 |
+
"logits/chosen": -0.39960938692092896,
|
| 225 |
+
"logits/rejected": -0.3809570372104645,
|
| 226 |
+
"logps/chosen": -470.6000061035156,
|
| 227 |
+
"logps/rejected": -472.20001220703125,
|
| 228 |
+
"loss": 0.0046,
|
| 229 |
+
"nll_loss": 2.043750047683716,
|
| 230 |
+
"rewards/accuracies": 1.0,
|
| 231 |
+
"rewards/chosen": 2.507031202316284,
|
| 232 |
+
"rewards/margins": 10.509374618530273,
|
| 233 |
+
"rewards/rejected": -8.009374618530273,
|
| 234 |
+
"step": 130
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.112,
|
| 238 |
+
"grad_norm": 0.6864472420728707,
|
| 239 |
+
"learning_rate": 4.937777777777777e-07,
|
| 240 |
+
"logits/chosen": -0.26445311307907104,
|
| 241 |
+
"logits/rejected": -0.293212890625,
|
| 242 |
+
"logps/chosen": -411.0,
|
| 243 |
+
"logps/rejected": -460.79998779296875,
|
| 244 |
+
"loss": 0.0023,
|
| 245 |
+
"nll_loss": 1.904687523841858,
|
| 246 |
+
"rewards/accuracies": 1.0,
|
| 247 |
+
"rewards/chosen": 3.862499952316284,
|
| 248 |
+
"rewards/margins": 11.34375,
|
| 249 |
+
"rewards/rejected": -7.487500190734863,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.12,
|
| 254 |
+
"grad_norm": 0.04499814020001256,
|
| 255 |
+
"learning_rate": 4.893333333333333e-07,
|
| 256 |
+
"logits/chosen": -0.2533935606479645,
|
| 257 |
+
"logits/rejected": -0.30218505859375,
|
| 258 |
+
"logps/chosen": -407.20001220703125,
|
| 259 |
+
"logps/rejected": -467.20001220703125,
|
| 260 |
+
"loss": 0.003,
|
| 261 |
+
"nll_loss": 1.5851562023162842,
|
| 262 |
+
"rewards/accuracies": 1.0,
|
| 263 |
+
"rewards/chosen": 5.1015625,
|
| 264 |
+
"rewards/margins": 12.21875,
|
| 265 |
+
"rewards/rejected": -7.109375,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.128,
|
| 270 |
+
"grad_norm": 0.12456933423635125,
|
| 271 |
+
"learning_rate": 4.848888888888888e-07,
|
| 272 |
+
"logits/chosen": -0.5003906488418579,
|
| 273 |
+
"logits/rejected": -0.4310546815395355,
|
| 274 |
+
"logps/chosen": -409.79998779296875,
|
| 275 |
+
"logps/rejected": -465.6000061035156,
|
| 276 |
+
"loss": 0.0118,
|
| 277 |
+
"nll_loss": 1.7273437976837158,
|
| 278 |
+
"rewards/accuracies": 0.987500011920929,
|
| 279 |
+
"rewards/chosen": 5.496874809265137,
|
| 280 |
+
"rewards/margins": 12.262499809265137,
|
| 281 |
+
"rewards/rejected": -6.771874904632568,
|
| 282 |
+
"step": 160
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.136,
|
| 286 |
+
"grad_norm": 0.10810136696878427,
|
| 287 |
+
"learning_rate": 4.804444444444444e-07,
|
| 288 |
+
"logits/chosen": -0.4216064512729645,
|
| 289 |
+
"logits/rejected": -0.3792480528354645,
|
| 290 |
+
"logps/chosen": -384.0,
|
| 291 |
+
"logps/rejected": -446.3999938964844,
|
| 292 |
+
"loss": 0.0022,
|
| 293 |
+
"nll_loss": 1.6687500476837158,
|
| 294 |
+
"rewards/accuracies": 1.0,
|
| 295 |
+
"rewards/chosen": 6.856249809265137,
|
| 296 |
+
"rewards/margins": 12.15625,
|
| 297 |
+
"rewards/rejected": -5.293749809265137,
|
| 298 |
+
"step": 170
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.144,
|
| 302 |
+
"grad_norm": 2.1017761962617016,
|
| 303 |
+
"learning_rate": 4.76e-07,
|
| 304 |
+
"logits/chosen": -0.3038574159145355,
|
| 305 |
+
"logits/rejected": -0.32097166776657104,
|
| 306 |
+
"logps/chosen": -426.20001220703125,
|
| 307 |
+
"logps/rejected": -432.79998779296875,
|
| 308 |
+
"loss": 0.0031,
|
| 309 |
+
"nll_loss": 1.6843750476837158,
|
| 310 |
+
"rewards/accuracies": 1.0,
|
| 311 |
+
"rewards/chosen": 8.0,
|
| 312 |
+
"rewards/margins": 12.050000190734863,
|
| 313 |
+
"rewards/rejected": -4.051562309265137,
|
| 314 |
+
"step": 180
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.152,
|
| 318 |
+
"grad_norm": 0.03472137716422411,
|
| 319 |
+
"learning_rate": 4.7155555555555556e-07,
|
| 320 |
+
"logits/chosen": -0.27385252714157104,
|
| 321 |
+
"logits/rejected": -0.274169921875,
|
| 322 |
+
"logps/chosen": -383.3999938964844,
|
| 323 |
+
"logps/rejected": -433.0,
|
| 324 |
+
"loss": 0.002,
|
| 325 |
+
"nll_loss": 1.497656226158142,
|
| 326 |
+
"rewards/accuracies": 1.0,
|
| 327 |
+
"rewards/chosen": 9.46875,
|
| 328 |
+
"rewards/margins": 14.081250190734863,
|
| 329 |
+
"rewards/rejected": -4.620312690734863,
|
| 330 |
+
"step": 190
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.16,
|
| 334 |
+
"grad_norm": 0.05628090345177454,
|
| 335 |
+
"learning_rate": 4.6711111111111104e-07,
|
| 336 |
+
"logits/chosen": -0.3796752989292145,
|
| 337 |
+
"logits/rejected": -0.342529296875,
|
| 338 |
+
"logps/chosen": -361.0,
|
| 339 |
+
"logps/rejected": -430.0,
|
| 340 |
+
"loss": 0.0017,
|
| 341 |
+
"nll_loss": 1.515625,
|
| 342 |
+
"rewards/accuracies": 1.0,
|
| 343 |
+
"rewards/chosen": 9.209375381469727,
|
| 344 |
+
"rewards/margins": 13.231249809265137,
|
| 345 |
+
"rewards/rejected": -4.025781154632568,
|
| 346 |
+
"step": 200
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.168,
|
| 350 |
+
"grad_norm": 0.03469897920654829,
|
| 351 |
+
"learning_rate": 4.6266666666666663e-07,
|
| 352 |
+
"logits/chosen": -0.30029296875,
|
| 353 |
+
"logits/rejected": -0.2890380918979645,
|
| 354 |
+
"logps/chosen": -372.70001220703125,
|
| 355 |
+
"logps/rejected": -418.20001220703125,
|
| 356 |
+
"loss": 0.0019,
|
| 357 |
+
"nll_loss": 1.4328124523162842,
|
| 358 |
+
"rewards/accuracies": 1.0,
|
| 359 |
+
"rewards/chosen": 10.162500381469727,
|
| 360 |
+
"rewards/margins": 13.068750381469727,
|
| 361 |
+
"rewards/rejected": -2.887939453125,
|
| 362 |
+
"step": 210
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.176,
|
| 366 |
+
"grad_norm": 0.03297401557091858,
|
| 367 |
+
"learning_rate": 4.5822222222222216e-07,
|
| 368 |
+
"logits/chosen": -0.4486328065395355,
|
| 369 |
+
"logits/rejected": -0.3726562559604645,
|
| 370 |
+
"logps/chosen": -398.6000061035156,
|
| 371 |
+
"logps/rejected": -430.79998779296875,
|
| 372 |
+
"loss": 0.0016,
|
| 373 |
+
"nll_loss": 1.484375,
|
| 374 |
+
"rewards/accuracies": 1.0,
|
| 375 |
+
"rewards/chosen": 10.693750381469727,
|
| 376 |
+
"rewards/margins": 14.068750381469727,
|
| 377 |
+
"rewards/rejected": -3.3890624046325684,
|
| 378 |
+
"step": 220
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.184,
|
| 382 |
+
"grad_norm": 0.03928262686521911,
|
| 383 |
+
"learning_rate": 4.5377777777777775e-07,
|
| 384 |
+
"logits/chosen": -0.31196290254592896,
|
| 385 |
+
"logits/rejected": -0.2955078184604645,
|
| 386 |
+
"logps/chosen": -329.6000061035156,
|
| 387 |
+
"logps/rejected": -432.6000061035156,
|
| 388 |
+
"loss": 0.0013,
|
| 389 |
+
"nll_loss": 1.2960937023162842,
|
| 390 |
+
"rewards/accuracies": 1.0,
|
| 391 |
+
"rewards/chosen": 11.287500381469727,
|
| 392 |
+
"rewards/margins": 15.443750381469727,
|
| 393 |
+
"rewards/rejected": -4.154687404632568,
|
| 394 |
+
"step": 230
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.192,
|
| 398 |
+
"grad_norm": 0.012808641512081923,
|
| 399 |
+
"learning_rate": 4.493333333333333e-07,
|
| 400 |
+
"logits/chosen": -0.28032225370407104,
|
| 401 |
+
"logits/rejected": -0.2730468809604645,
|
| 402 |
+
"logps/chosen": -390.1000061035156,
|
| 403 |
+
"logps/rejected": -418.20001220703125,
|
| 404 |
+
"loss": 0.0017,
|
| 405 |
+
"nll_loss": 1.353124976158142,
|
| 406 |
+
"rewards/accuracies": 1.0,
|
| 407 |
+
"rewards/chosen": 11.774999618530273,
|
| 408 |
+
"rewards/margins": 14.8125,
|
| 409 |
+
"rewards/rejected": -3.0433592796325684,
|
| 410 |
+
"step": 240
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.2,
|
| 414 |
+
"grad_norm": 0.06025316567815392,
|
| 415 |
+
"learning_rate": 4.4488888888888887e-07,
|
| 416 |
+
"logits/chosen": -0.2566772401332855,
|
| 417 |
+
"logits/rejected": -0.19821777939796448,
|
| 418 |
+
"logps/chosen": -371.20001220703125,
|
| 419 |
+
"logps/rejected": -412.6000061035156,
|
| 420 |
+
"loss": 0.0014,
|
| 421 |
+
"nll_loss": 1.3125,
|
| 422 |
+
"rewards/accuracies": 1.0,
|
| 423 |
+
"rewards/chosen": 11.987500190734863,
|
| 424 |
+
"rewards/margins": 14.800000190734863,
|
| 425 |
+
"rewards/rejected": -2.801953077316284,
|
| 426 |
+
"step": 250
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.208,
|
| 430 |
+
"grad_norm": 0.018239198980973682,
|
| 431 |
+
"learning_rate": 4.4044444444444445e-07,
|
| 432 |
+
"logits/chosen": -0.3375000059604645,
|
| 433 |
+
"logits/rejected": -0.22773437201976776,
|
| 434 |
+
"logps/chosen": -367.70001220703125,
|
| 435 |
+
"logps/rejected": -419.79998779296875,
|
| 436 |
+
"loss": 0.0016,
|
| 437 |
+
"nll_loss": 1.3914062976837158,
|
| 438 |
+
"rewards/accuracies": 1.0,
|
| 439 |
+
"rewards/chosen": 13.081250190734863,
|
| 440 |
+
"rewards/margins": 15.800000190734863,
|
| 441 |
+
"rewards/rejected": -2.72265625,
|
| 442 |
+
"step": 260
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.216,
|
| 446 |
+
"grad_norm": 0.06735412914324118,
|
| 447 |
+
"learning_rate": 4.36e-07,
|
| 448 |
+
"logits/chosen": -0.30003660917282104,
|
| 449 |
+
"logits/rejected": -0.3238281309604645,
|
| 450 |
+
"logps/chosen": -345.8999938964844,
|
| 451 |
+
"logps/rejected": -427.0,
|
| 452 |
+
"loss": 0.01,
|
| 453 |
+
"nll_loss": 1.340234398841858,
|
| 454 |
+
"rewards/accuracies": 1.0,
|
| 455 |
+
"rewards/chosen": 11.768750190734863,
|
| 456 |
+
"rewards/margins": 15.125,
|
| 457 |
+
"rewards/rejected": -3.364453077316284,
|
| 458 |
+
"step": 270
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.224,
|
| 462 |
+
"grad_norm": 0.01881116895313125,
|
| 463 |
+
"learning_rate": 4.3155555555555557e-07,
|
| 464 |
+
"logits/chosen": -0.29327392578125,
|
| 465 |
+
"logits/rejected": -0.25737303495407104,
|
| 466 |
+
"logps/chosen": -365.0,
|
| 467 |
+
"logps/rejected": -431.3999938964844,
|
| 468 |
+
"loss": 0.0036,
|
| 469 |
+
"nll_loss": 1.2589843273162842,
|
| 470 |
+
"rewards/accuracies": 1.0,
|
| 471 |
+
"rewards/chosen": 12.668749809265137,
|
| 472 |
+
"rewards/margins": 16.03125,
|
| 473 |
+
"rewards/rejected": -3.3645873069763184,
|
| 474 |
+
"step": 280
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.232,
|
| 478 |
+
"grad_norm": 0.033150944419182336,
|
| 479 |
+
"learning_rate": 4.271111111111111e-07,
|
| 480 |
+
"logits/chosen": -0.24541015923023224,
|
| 481 |
+
"logits/rejected": -0.2640136778354645,
|
| 482 |
+
"logps/chosen": -323.0,
|
| 483 |
+
"logps/rejected": -439.0,
|
| 484 |
+
"loss": 0.0014,
|
| 485 |
+
"nll_loss": 1.3125,
|
| 486 |
+
"rewards/accuracies": 1.0,
|
| 487 |
+
"rewards/chosen": 12.925000190734863,
|
| 488 |
+
"rewards/margins": 16.3125,
|
| 489 |
+
"rewards/rejected": -3.383984327316284,
|
| 490 |
+
"step": 290
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.24,
|
| 494 |
+
"grad_norm": 0.07622395451753027,
|
| 495 |
+
"learning_rate": 4.226666666666667e-07,
|
| 496 |
+
"logits/chosen": -0.34736329317092896,
|
| 497 |
+
"logits/rejected": -0.31829530000686646,
|
| 498 |
+
"logps/chosen": -317.1000061035156,
|
| 499 |
+
"logps/rejected": -420.0,
|
| 500 |
+
"loss": 0.0014,
|
| 501 |
+
"nll_loss": 1.361718773841858,
|
| 502 |
+
"rewards/accuracies": 1.0,
|
| 503 |
+
"rewards/chosen": 13.631250381469727,
|
| 504 |
+
"rewards/margins": 16.15625,
|
| 505 |
+
"rewards/rejected": -2.510937452316284,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.248,
|
| 510 |
+
"grad_norm": 0.018683331713432866,
|
| 511 |
+
"learning_rate": 4.1822222222222217e-07,
|
| 512 |
+
"logits/chosen": -0.3045898377895355,
|
| 513 |
+
"logits/rejected": -0.24216309189796448,
|
| 514 |
+
"logps/chosen": -341.8999938964844,
|
| 515 |
+
"logps/rejected": -413.0,
|
| 516 |
+
"loss": 0.002,
|
| 517 |
+
"nll_loss": 1.350000023841858,
|
| 518 |
+
"rewards/accuracies": 1.0,
|
| 519 |
+
"rewards/chosen": 13.699999809265137,
|
| 520 |
+
"rewards/margins": 16.631250381469727,
|
| 521 |
+
"rewards/rejected": -2.9205079078674316,
|
| 522 |
+
"step": 310
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.256,
|
| 526 |
+
"grad_norm": 0.023574936630402193,
|
| 527 |
+
"learning_rate": 4.1377777777777776e-07,
|
| 528 |
+
"logits/chosen": -0.21635742485523224,
|
| 529 |
+
"logits/rejected": -0.23710937798023224,
|
| 530 |
+
"logps/chosen": -434.3999938964844,
|
| 531 |
+
"logps/rejected": -407.3999938964844,
|
| 532 |
+
"loss": 0.0022,
|
| 533 |
+
"nll_loss": 1.337499976158142,
|
| 534 |
+
"rewards/accuracies": 1.0,
|
| 535 |
+
"rewards/chosen": 13.274999618530273,
|
| 536 |
+
"rewards/margins": 15.524999618530273,
|
| 537 |
+
"rewards/rejected": -2.2476563453674316,
|
| 538 |
+
"step": 320
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.264,
|
| 542 |
+
"grad_norm": 0.03038779144828818,
|
| 543 |
+
"learning_rate": 4.093333333333333e-07,
|
| 544 |
+
"logits/chosen": -0.1950538605451584,
|
| 545 |
+
"logits/rejected": -0.179931640625,
|
| 546 |
+
"logps/chosen": -350.8999938964844,
|
| 547 |
+
"logps/rejected": -419.20001220703125,
|
| 548 |
+
"loss": 0.0127,
|
| 549 |
+
"nll_loss": 1.1906249523162842,
|
| 550 |
+
"rewards/accuracies": 0.987500011920929,
|
| 551 |
+
"rewards/chosen": 13.899999618530273,
|
| 552 |
+
"rewards/margins": 16.506250381469727,
|
| 553 |
+
"rewards/rejected": -2.590625047683716,
|
| 554 |
+
"step": 330
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.272,
|
| 558 |
+
"grad_norm": 0.04436049431080709,
|
| 559 |
+
"learning_rate": 4.048888888888889e-07,
|
| 560 |
+
"logits/chosen": -0.24531249701976776,
|
| 561 |
+
"logits/rejected": -0.2109375,
|
| 562 |
+
"logps/chosen": -297.79998779296875,
|
| 563 |
+
"logps/rejected": -413.6000061035156,
|
| 564 |
+
"loss": 0.0016,
|
| 565 |
+
"nll_loss": 1.203125,
|
| 566 |
+
"rewards/accuracies": 1.0,
|
| 567 |
+
"rewards/chosen": 14.46875,
|
| 568 |
+
"rewards/margins": 16.325000762939453,
|
| 569 |
+
"rewards/rejected": -1.864843726158142,
|
| 570 |
+
"step": 340
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.28,
|
| 574 |
+
"grad_norm": 0.017501398014684644,
|
| 575 |
+
"learning_rate": 4.004444444444444e-07,
|
| 576 |
+
"logits/chosen": -0.329833984375,
|
| 577 |
+
"logits/rejected": -0.34770506620407104,
|
| 578 |
+
"logps/chosen": -308.5,
|
| 579 |
+
"logps/rejected": -421.6000061035156,
|
| 580 |
+
"loss": 0.0012,
|
| 581 |
+
"nll_loss": 1.205468773841858,
|
| 582 |
+
"rewards/accuracies": 1.0,
|
| 583 |
+
"rewards/chosen": 14.143750190734863,
|
| 584 |
+
"rewards/margins": 16.381250381469727,
|
| 585 |
+
"rewards/rejected": -2.240673780441284,
|
| 586 |
+
"step": 350
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.288,
|
| 590 |
+
"grad_norm": 0.028368395543364983,
|
| 591 |
+
"learning_rate": 3.96e-07,
|
| 592 |
+
"logits/chosen": -0.19560547173023224,
|
| 593 |
+
"logits/rejected": -0.0811767578125,
|
| 594 |
+
"logps/chosen": -310.3999938964844,
|
| 595 |
+
"logps/rejected": -401.6000061035156,
|
| 596 |
+
"loss": 0.0013,
|
| 597 |
+
"nll_loss": 1.228906273841858,
|
| 598 |
+
"rewards/accuracies": 1.0,
|
| 599 |
+
"rewards/chosen": 15.112500190734863,
|
| 600 |
+
"rewards/margins": 16.625,
|
| 601 |
+
"rewards/rejected": -1.511328101158142,
|
| 602 |
+
"step": 360
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.296,
|
| 606 |
+
"grad_norm": 0.03826354878861224,
|
| 607 |
+
"learning_rate": 3.9155555555555553e-07,
|
| 608 |
+
"logits/chosen": -0.21774902939796448,
|
| 609 |
+
"logits/rejected": -0.11984863132238388,
|
| 610 |
+
"logps/chosen": -343.70001220703125,
|
| 611 |
+
"logps/rejected": -396.79998779296875,
|
| 612 |
+
"loss": 0.0015,
|
| 613 |
+
"nll_loss": 1.181249976158142,
|
| 614 |
+
"rewards/accuracies": 1.0,
|
| 615 |
+
"rewards/chosen": 14.868749618530273,
|
| 616 |
+
"rewards/margins": 15.931249618530273,
|
| 617 |
+
"rewards/rejected": -1.049218773841858,
|
| 618 |
+
"step": 370
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.304,
|
| 622 |
+
"grad_norm": 0.015874061291675685,
|
| 623 |
+
"learning_rate": 3.871111111111111e-07,
|
| 624 |
+
"logits/chosen": -0.1663818359375,
|
| 625 |
+
"logits/rejected": -0.06330566108226776,
|
| 626 |
+
"logps/chosen": -333.70001220703125,
|
| 627 |
+
"logps/rejected": -423.0,
|
| 628 |
+
"loss": 0.0013,
|
| 629 |
+
"nll_loss": 1.303125023841858,
|
| 630 |
+
"rewards/accuracies": 1.0,
|
| 631 |
+
"rewards/chosen": 14.887499809265137,
|
| 632 |
+
"rewards/margins": 17.950000762939453,
|
| 633 |
+
"rewards/rejected": -3.07421875,
|
| 634 |
+
"step": 380
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.312,
|
| 638 |
+
"grad_norm": 0.019028625638946934,
|
| 639 |
+
"learning_rate": 3.8266666666666665e-07,
|
| 640 |
+
"logits/chosen": -0.13876953721046448,
|
| 641 |
+
"logits/rejected": -0.10498046875,
|
| 642 |
+
"logps/chosen": -383.70001220703125,
|
| 643 |
+
"logps/rejected": -405.3999938964844,
|
| 644 |
+
"loss": 0.0012,
|
| 645 |
+
"nll_loss": 1.1964843273162842,
|
| 646 |
+
"rewards/accuracies": 1.0,
|
| 647 |
+
"rewards/chosen": 14.25,
|
| 648 |
+
"rewards/margins": 16.649999618530273,
|
| 649 |
+
"rewards/rejected": -2.4046874046325684,
|
| 650 |
+
"step": 390
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.32,
|
| 654 |
+
"grad_norm": 0.019036097748468133,
|
| 655 |
+
"learning_rate": 3.7822222222222224e-07,
|
| 656 |
+
"logits/chosen": -0.3286499083042145,
|
| 657 |
+
"logits/rejected": -0.28095704317092896,
|
| 658 |
+
"logps/chosen": -329.79998779296875,
|
| 659 |
+
"logps/rejected": -426.3999938964844,
|
| 660 |
+
"loss": 0.0013,
|
| 661 |
+
"nll_loss": 1.2902343273162842,
|
| 662 |
+
"rewards/accuracies": 1.0,
|
| 663 |
+
"rewards/chosen": 16.049999237060547,
|
| 664 |
+
"rewards/margins": 18.825000762939453,
|
| 665 |
+
"rewards/rejected": -2.77069091796875,
|
| 666 |
+
"step": 400
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.328,
|
| 670 |
+
"grad_norm": 0.016730568465352296,
|
| 671 |
+
"learning_rate": 3.7377777777777777e-07,
|
| 672 |
+
"logits/chosen": -0.11003418266773224,
|
| 673 |
+
"logits/rejected": -0.12824706733226776,
|
| 674 |
+
"logps/chosen": -382.3999938964844,
|
| 675 |
+
"logps/rejected": -416.0,
|
| 676 |
+
"loss": 0.0026,
|
| 677 |
+
"nll_loss": 1.211328148841858,
|
| 678 |
+
"rewards/accuracies": 1.0,
|
| 679 |
+
"rewards/chosen": 15.206250190734863,
|
| 680 |
+
"rewards/margins": 17.387500762939453,
|
| 681 |
+
"rewards/rejected": -2.163281202316284,
|
| 682 |
+
"step": 410
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.336,
|
| 686 |
+
"grad_norm": 0.01850424460734669,
|
| 687 |
+
"learning_rate": 3.693333333333333e-07,
|
| 688 |
+
"logits/chosen": -0.16533203423023224,
|
| 689 |
+
"logits/rejected": -0.1209716796875,
|
| 690 |
+
"logps/chosen": -356.3999938964844,
|
| 691 |
+
"logps/rejected": -413.79998779296875,
|
| 692 |
+
"loss": 0.0012,
|
| 693 |
+
"nll_loss": 1.1242187023162842,
|
| 694 |
+
"rewards/accuracies": 1.0,
|
| 695 |
+
"rewards/chosen": 15.637499809265137,
|
| 696 |
+
"rewards/margins": 17.71875,
|
| 697 |
+
"rewards/rejected": -2.0859375,
|
| 698 |
+
"step": 420
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.344,
|
| 702 |
+
"grad_norm": 0.039652389470085536,
|
| 703 |
+
"learning_rate": 3.6488888888888884e-07,
|
| 704 |
+
"logits/chosen": -0.10117187350988388,
|
| 705 |
+
"logits/rejected": -0.05829467624425888,
|
| 706 |
+
"logps/chosen": -298.20001220703125,
|
| 707 |
+
"logps/rejected": -420.6000061035156,
|
| 708 |
+
"loss": 0.0012,
|
| 709 |
+
"nll_loss": 1.1785156726837158,
|
| 710 |
+
"rewards/accuracies": 1.0,
|
| 711 |
+
"rewards/chosen": 15.581250190734863,
|
| 712 |
+
"rewards/margins": 18.375,
|
| 713 |
+
"rewards/rejected": -2.7708983421325684,
|
| 714 |
+
"step": 430
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.352,
|
| 718 |
+
"grad_norm": 0.008419647898269315,
|
| 719 |
+
"learning_rate": 3.604444444444444e-07,
|
| 720 |
+
"logits/chosen": -0.24436035752296448,
|
| 721 |
+
"logits/rejected": -0.1181640625,
|
| 722 |
+
"logps/chosen": -296.29998779296875,
|
| 723 |
+
"logps/rejected": -423.20001220703125,
|
| 724 |
+
"loss": 0.0011,
|
| 725 |
+
"nll_loss": 1.058984398841858,
|
| 726 |
+
"rewards/accuracies": 1.0,
|
| 727 |
+
"rewards/chosen": 16.268749237060547,
|
| 728 |
+
"rewards/margins": 18.568750381469727,
|
| 729 |
+
"rewards/rejected": -2.2855467796325684,
|
| 730 |
+
"step": 440
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.36,
|
| 734 |
+
"grad_norm": 0.008531494272693656,
|
| 735 |
+
"learning_rate": 3.5599999999999996e-07,
|
| 736 |
+
"logits/chosen": -0.06640625,
|
| 737 |
+
"logits/rejected": -0.0052490234375,
|
| 738 |
+
"logps/chosen": -329.1000061035156,
|
| 739 |
+
"logps/rejected": -404.6000061035156,
|
| 740 |
+
"loss": 0.0016,
|
| 741 |
+
"nll_loss": 1.16796875,
|
| 742 |
+
"rewards/accuracies": 1.0,
|
| 743 |
+
"rewards/chosen": 16.049999237060547,
|
| 744 |
+
"rewards/margins": 18.012500762939453,
|
| 745 |
+
"rewards/rejected": -1.9445312023162842,
|
| 746 |
+
"step": 450
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.368,
|
| 750 |
+
"grad_norm": 0.021562461095979032,
|
| 751 |
+
"learning_rate": 3.5155555555555554e-07,
|
| 752 |
+
"logits/chosen": -0.17312011122703552,
|
| 753 |
+
"logits/rejected": -0.07587890326976776,
|
| 754 |
+
"logps/chosen": -316.8999938964844,
|
| 755 |
+
"logps/rejected": -412.79998779296875,
|
| 756 |
+
"loss": 0.0012,
|
| 757 |
+
"nll_loss": 1.1652343273162842,
|
| 758 |
+
"rewards/accuracies": 1.0,
|
| 759 |
+
"rewards/chosen": 15.381250381469727,
|
| 760 |
+
"rewards/margins": 17.799999237060547,
|
| 761 |
+
"rewards/rejected": -2.4085936546325684,
|
| 762 |
+
"step": 460
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.376,
|
| 766 |
+
"grad_norm": 0.07020322384992712,
|
| 767 |
+
"learning_rate": 3.471111111111111e-07,
|
| 768 |
+
"logits/chosen": -0.14414063096046448,
|
| 769 |
+
"logits/rejected": -0.12218017876148224,
|
| 770 |
+
"logps/chosen": -299.70001220703125,
|
| 771 |
+
"logps/rejected": -428.0,
|
| 772 |
+
"loss": 0.0017,
|
| 773 |
+
"nll_loss": 1.25,
|
| 774 |
+
"rewards/accuracies": 1.0,
|
| 775 |
+
"rewards/chosen": 15.856249809265137,
|
| 776 |
+
"rewards/margins": 18.456249237060547,
|
| 777 |
+
"rewards/rejected": -2.5699219703674316,
|
| 778 |
+
"step": 470
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.384,
|
| 782 |
+
"grad_norm": 0.012596113248962468,
|
| 783 |
+
"learning_rate": 3.4266666666666666e-07,
|
| 784 |
+
"logits/chosen": -0.08417968451976776,
|
| 785 |
+
"logits/rejected": -0.03115234337747097,
|
| 786 |
+
"logps/chosen": -302.79998779296875,
|
| 787 |
+
"logps/rejected": -405.0,
|
| 788 |
+
"loss": 0.0012,
|
| 789 |
+
"nll_loss": 1.0558593273162842,
|
| 790 |
+
"rewards/accuracies": 1.0,
|
| 791 |
+
"rewards/chosen": 15.668749809265137,
|
| 792 |
+
"rewards/margins": 17.862499237060547,
|
| 793 |
+
"rewards/rejected": -2.1884765625,
|
| 794 |
+
"step": 480
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.392,
|
| 798 |
+
"grad_norm": 0.007929870216062266,
|
| 799 |
+
"learning_rate": 3.382222222222222e-07,
|
| 800 |
+
"logits/chosen": -0.10902099311351776,
|
| 801 |
+
"logits/rejected": -0.11533202975988388,
|
| 802 |
+
"logps/chosen": -397.20001220703125,
|
| 803 |
+
"logps/rejected": -404.6000061035156,
|
| 804 |
+
"loss": 0.0038,
|
| 805 |
+
"nll_loss": 1.135156273841858,
|
| 806 |
+
"rewards/accuracies": 1.0,
|
| 807 |
+
"rewards/chosen": 15.993749618530273,
|
| 808 |
+
"rewards/margins": 18.862499237060547,
|
| 809 |
+
"rewards/rejected": -2.874706983566284,
|
| 810 |
+
"step": 490
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.4,
|
| 814 |
+
"grad_norm": 0.008630639702085227,
|
| 815 |
+
"learning_rate": 3.337777777777778e-07,
|
| 816 |
+
"logits/chosen": -0.26744383573532104,
|
| 817 |
+
"logits/rejected": -0.14018554985523224,
|
| 818 |
+
"logps/chosen": -318.6000061035156,
|
| 819 |
+
"logps/rejected": -434.20001220703125,
|
| 820 |
+
"loss": 0.0012,
|
| 821 |
+
"nll_loss": 1.166406273841858,
|
| 822 |
+
"rewards/accuracies": 1.0,
|
| 823 |
+
"rewards/chosen": 16.037500381469727,
|
| 824 |
+
"rewards/margins": 19.012500762939453,
|
| 825 |
+
"rewards/rejected": -2.995312452316284,
|
| 826 |
+
"step": 500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 0.408,
|
| 830 |
+
"grad_norm": 0.016229293972070177,
|
| 831 |
+
"learning_rate": 3.293333333333333e-07,
|
| 832 |
+
"logits/chosen": -0.204833984375,
|
| 833 |
+
"logits/rejected": -0.13259276747703552,
|
| 834 |
+
"logps/chosen": -322.0,
|
| 835 |
+
"logps/rejected": -407.79998779296875,
|
| 836 |
+
"loss": 0.0022,
|
| 837 |
+
"nll_loss": 1.116796851158142,
|
| 838 |
+
"rewards/accuracies": 1.0,
|
| 839 |
+
"rewards/chosen": 15.512499809265137,
|
| 840 |
+
"rewards/margins": 18.28125,
|
| 841 |
+
"rewards/rejected": -2.77734375,
|
| 842 |
+
"step": 510
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.416,
|
| 846 |
+
"grad_norm": 0.5999767196042933,
|
| 847 |
+
"learning_rate": 3.248888888888889e-07,
|
| 848 |
+
"logits/chosen": -0.171142578125,
|
| 849 |
+
"logits/rejected": -0.15626220405101776,
|
| 850 |
+
"logps/chosen": -320.6000061035156,
|
| 851 |
+
"logps/rejected": -428.20001220703125,
|
| 852 |
+
"loss": 0.0013,
|
| 853 |
+
"nll_loss": 1.1687500476837158,
|
| 854 |
+
"rewards/accuracies": 1.0,
|
| 855 |
+
"rewards/chosen": 16.162500381469727,
|
| 856 |
+
"rewards/margins": 19.862499237060547,
|
| 857 |
+
"rewards/rejected": -3.703125,
|
| 858 |
+
"step": 520
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.424,
|
| 862 |
+
"grad_norm": 0.30677660404469975,
|
| 863 |
+
"learning_rate": 3.204444444444444e-07,
|
| 864 |
+
"logits/chosen": -0.20744629204273224,
|
| 865 |
+
"logits/rejected": -0.15806885063648224,
|
| 866 |
+
"logps/chosen": -359.0,
|
| 867 |
+
"logps/rejected": -433.20001220703125,
|
| 868 |
+
"loss": 0.0017,
|
| 869 |
+
"nll_loss": 1.25390625,
|
| 870 |
+
"rewards/accuracies": 1.0,
|
| 871 |
+
"rewards/chosen": 15.912500381469727,
|
| 872 |
+
"rewards/margins": 19.4375,
|
| 873 |
+
"rewards/rejected": -3.549999952316284,
|
| 874 |
+
"step": 530
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.432,
|
| 878 |
+
"grad_norm": 0.016384405120536898,
|
| 879 |
+
"learning_rate": 3.1599999999999997e-07,
|
| 880 |
+
"logits/chosen": -0.11655273288488388,
|
| 881 |
+
"logits/rejected": -0.13643798232078552,
|
| 882 |
+
"logps/chosen": -284.1000061035156,
|
| 883 |
+
"logps/rejected": -408.6000061035156,
|
| 884 |
+
"loss": 0.001,
|
| 885 |
+
"nll_loss": 1.033593773841858,
|
| 886 |
+
"rewards/accuracies": 1.0,
|
| 887 |
+
"rewards/chosen": 16.856250762939453,
|
| 888 |
+
"rewards/margins": 19.043750762939453,
|
| 889 |
+
"rewards/rejected": -2.197338819503784,
|
| 890 |
+
"step": 540
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.44,
|
| 894 |
+
"grad_norm": 0.02552403140799525,
|
| 895 |
+
"learning_rate": 3.115555555555555e-07,
|
| 896 |
+
"logits/chosen": -0.1761474609375,
|
| 897 |
+
"logits/rejected": -0.08870239555835724,
|
| 898 |
+
"logps/chosen": -308.8999938964844,
|
| 899 |
+
"logps/rejected": -428.6000061035156,
|
| 900 |
+
"loss": 0.0012,
|
| 901 |
+
"nll_loss": 1.185156226158142,
|
| 902 |
+
"rewards/accuracies": 1.0,
|
| 903 |
+
"rewards/chosen": 17.200000762939453,
|
| 904 |
+
"rewards/margins": 19.887500762939453,
|
| 905 |
+
"rewards/rejected": -2.680468797683716,
|
| 906 |
+
"step": 550
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.448,
|
| 910 |
+
"grad_norm": 0.012418272439184573,
|
| 911 |
+
"learning_rate": 3.071111111111111e-07,
|
| 912 |
+
"logits/chosen": -0.25361329317092896,
|
| 913 |
+
"logits/rejected": -0.15983887016773224,
|
| 914 |
+
"logps/chosen": -371.29998779296875,
|
| 915 |
+
"logps/rejected": -401.3999938964844,
|
| 916 |
+
"loss": 0.0012,
|
| 917 |
+
"nll_loss": 1.234375,
|
| 918 |
+
"rewards/accuracies": 1.0,
|
| 919 |
+
"rewards/chosen": 16.643749237060547,
|
| 920 |
+
"rewards/margins": 18.518749237060547,
|
| 921 |
+
"rewards/rejected": -1.90234375,
|
| 922 |
+
"step": 560
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 0.456,
|
| 926 |
+
"grad_norm": 0.008584798479513097,
|
| 927 |
+
"learning_rate": 3.026666666666666e-07,
|
| 928 |
+
"logits/chosen": -0.10489501804113388,
|
| 929 |
+
"logits/rejected": 0.01859130896627903,
|
| 930 |
+
"logps/chosen": -305.3999938964844,
|
| 931 |
+
"logps/rejected": -411.20001220703125,
|
| 932 |
+
"loss": 0.0012,
|
| 933 |
+
"nll_loss": 1.1484375,
|
| 934 |
+
"rewards/accuracies": 1.0,
|
| 935 |
+
"rewards/chosen": 17.762500762939453,
|
| 936 |
+
"rewards/margins": 19.325000762939453,
|
| 937 |
+
"rewards/rejected": -1.558447241783142,
|
| 938 |
+
"step": 570
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.464,
|
| 942 |
+
"grad_norm": 0.054564295459296766,
|
| 943 |
+
"learning_rate": 2.982222222222222e-07,
|
| 944 |
+
"logits/chosen": -0.12746581435203552,
|
| 945 |
+
"logits/rejected": 0.02424316480755806,
|
| 946 |
+
"logps/chosen": -309.5,
|
| 947 |
+
"logps/rejected": -403.6000061035156,
|
| 948 |
+
"loss": 0.0013,
|
| 949 |
+
"nll_loss": 1.109765648841858,
|
| 950 |
+
"rewards/accuracies": 1.0,
|
| 951 |
+
"rewards/chosen": 16.78125,
|
| 952 |
+
"rewards/margins": 18.862499237060547,
|
| 953 |
+
"rewards/rejected": -2.0740723609924316,
|
| 954 |
+
"step": 580
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.472,
|
| 958 |
+
"grad_norm": 0.021190418364761706,
|
| 959 |
+
"learning_rate": 2.937777777777778e-07,
|
| 960 |
+
"logits/chosen": -0.12922362983226776,
|
| 961 |
+
"logits/rejected": -0.007800293155014515,
|
| 962 |
+
"logps/chosen": -327.8999938964844,
|
| 963 |
+
"logps/rejected": -372.6000061035156,
|
| 964 |
+
"loss": 0.0012,
|
| 965 |
+
"nll_loss": 1.1261718273162842,
|
| 966 |
+
"rewards/accuracies": 1.0,
|
| 967 |
+
"rewards/chosen": 16.299999237060547,
|
| 968 |
+
"rewards/margins": 17.212499618530273,
|
| 969 |
+
"rewards/rejected": -0.9365234375,
|
| 970 |
+
"step": 590
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 0.48,
|
| 974 |
+
"grad_norm": 0.1273137285714855,
|
| 975 |
+
"learning_rate": 2.8933333333333333e-07,
|
| 976 |
+
"logits/chosen": -0.02729492262005806,
|
| 977 |
+
"logits/rejected": 0.03876953199505806,
|
| 978 |
+
"logps/chosen": -349.6000061035156,
|
| 979 |
+
"logps/rejected": -420.6000061035156,
|
| 980 |
+
"loss": 0.0013,
|
| 981 |
+
"nll_loss": 1.221093773841858,
|
| 982 |
+
"rewards/accuracies": 1.0,
|
| 983 |
+
"rewards/chosen": 16.512500762939453,
|
| 984 |
+
"rewards/margins": 20.143749237060547,
|
| 985 |
+
"rewards/rejected": -3.608593702316284,
|
| 986 |
+
"step": 600
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.488,
|
| 990 |
+
"grad_norm": 0.11206326927393837,
|
| 991 |
+
"learning_rate": 2.848888888888889e-07,
|
| 992 |
+
"logits/chosen": -0.05439453199505806,
|
| 993 |
+
"logits/rejected": 0.04978637769818306,
|
| 994 |
+
"logps/chosen": -288.8999938964844,
|
| 995 |
+
"logps/rejected": -419.0,
|
| 996 |
+
"loss": 0.0011,
|
| 997 |
+
"nll_loss": 1.078125,
|
| 998 |
+
"rewards/accuracies": 1.0,
|
| 999 |
+
"rewards/chosen": 16.268749237060547,
|
| 1000 |
+
"rewards/margins": 18.625,
|
| 1001 |
+
"rewards/rejected": -2.3518919944763184,
|
| 1002 |
+
"step": 610
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.496,
|
| 1006 |
+
"grad_norm": 0.006737027469415457,
|
| 1007 |
+
"learning_rate": 2.8044444444444445e-07,
|
| 1008 |
+
"logits/chosen": -0.06257323920726776,
|
| 1009 |
+
"logits/rejected": -0.02427978441119194,
|
| 1010 |
+
"logps/chosen": -282.6000061035156,
|
| 1011 |
+
"logps/rejected": -422.20001220703125,
|
| 1012 |
+
"loss": 0.0011,
|
| 1013 |
+
"nll_loss": 1.094140648841858,
|
| 1014 |
+
"rewards/accuracies": 1.0,
|
| 1015 |
+
"rewards/chosen": 16.34375,
|
| 1016 |
+
"rewards/margins": 19.587499618530273,
|
| 1017 |
+
"rewards/rejected": -3.24609375,
|
| 1018 |
+
"step": 620
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.504,
|
| 1022 |
+
"grad_norm": 0.007960696014384233,
|
| 1023 |
+
"learning_rate": 2.7600000000000004e-07,
|
| 1024 |
+
"logits/chosen": -0.26665037870407104,
|
| 1025 |
+
"logits/rejected": -0.16494140028953552,
|
| 1026 |
+
"logps/chosen": -271.0,
|
| 1027 |
+
"logps/rejected": -415.3999938964844,
|
| 1028 |
+
"loss": 0.0011,
|
| 1029 |
+
"nll_loss": 1.0574219226837158,
|
| 1030 |
+
"rewards/accuracies": 1.0,
|
| 1031 |
+
"rewards/chosen": 17.25,
|
| 1032 |
+
"rewards/margins": 19.387500762939453,
|
| 1033 |
+
"rewards/rejected": -2.118847608566284,
|
| 1034 |
+
"step": 630
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.512,
|
| 1038 |
+
"grad_norm": 0.012795263557863883,
|
| 1039 |
+
"learning_rate": 2.715555555555555e-07,
|
| 1040 |
+
"logits/chosen": -0.03950195387005806,
|
| 1041 |
+
"logits/rejected": 0.03125,
|
| 1042 |
+
"logps/chosen": -325.8999938964844,
|
| 1043 |
+
"logps/rejected": -407.79998779296875,
|
| 1044 |
+
"loss": 0.0111,
|
| 1045 |
+
"nll_loss": 1.0187499523162842,
|
| 1046 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1047 |
+
"rewards/chosen": 17.393749237060547,
|
| 1048 |
+
"rewards/margins": 19.674999237060547,
|
| 1049 |
+
"rewards/rejected": -2.2586913108825684,
|
| 1050 |
+
"step": 640
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.52,
|
| 1054 |
+
"grad_norm": 0.01127355606086418,
|
| 1055 |
+
"learning_rate": 2.671111111111111e-07,
|
| 1056 |
+
"logits/chosen": -0.078369140625,
|
| 1057 |
+
"logits/rejected": -0.0323486328125,
|
| 1058 |
+
"logps/chosen": -289.0,
|
| 1059 |
+
"logps/rejected": -417.0,
|
| 1060 |
+
"loss": 0.001,
|
| 1061 |
+
"nll_loss": 1.0125000476837158,
|
| 1062 |
+
"rewards/accuracies": 1.0,
|
| 1063 |
+
"rewards/chosen": 17.318750381469727,
|
| 1064 |
+
"rewards/margins": 19.424999237060547,
|
| 1065 |
+
"rewards/rejected": -2.0975098609924316,
|
| 1066 |
+
"step": 650
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.528,
|
| 1070 |
+
"grad_norm": 0.019808530454758836,
|
| 1071 |
+
"learning_rate": 2.6266666666666664e-07,
|
| 1072 |
+
"logits/chosen": -0.04782714694738388,
|
| 1073 |
+
"logits/rejected": 0.09812011569738388,
|
| 1074 |
+
"logps/chosen": -323.6000061035156,
|
| 1075 |
+
"logps/rejected": -409.6000061035156,
|
| 1076 |
+
"loss": 0.0011,
|
| 1077 |
+
"nll_loss": 1.1015625,
|
| 1078 |
+
"rewards/accuracies": 1.0,
|
| 1079 |
+
"rewards/chosen": 17.774999618530273,
|
| 1080 |
+
"rewards/margins": 20.549999237060547,
|
| 1081 |
+
"rewards/rejected": -2.7822265625,
|
| 1082 |
+
"step": 660
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.536,
|
| 1086 |
+
"grad_norm": 0.021732913992860835,
|
| 1087 |
+
"learning_rate": 2.582222222222222e-07,
|
| 1088 |
+
"logits/chosen": 0.01522216759622097,
|
| 1089 |
+
"logits/rejected": 0.11748047173023224,
|
| 1090 |
+
"logps/chosen": -285.20001220703125,
|
| 1091 |
+
"logps/rejected": -419.20001220703125,
|
| 1092 |
+
"loss": 0.0011,
|
| 1093 |
+
"nll_loss": 1.078125,
|
| 1094 |
+
"rewards/accuracies": 1.0,
|
| 1095 |
+
"rewards/chosen": 17.850000381469727,
|
| 1096 |
+
"rewards/margins": 19.487499237060547,
|
| 1097 |
+
"rewards/rejected": -1.6281249523162842,
|
| 1098 |
+
"step": 670
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 0.544,
|
| 1102 |
+
"grad_norm": 0.005724681559177148,
|
| 1103 |
+
"learning_rate": 2.5377777777777776e-07,
|
| 1104 |
+
"logits/chosen": 0.0006347656017169356,
|
| 1105 |
+
"logits/rejected": 0.07487793266773224,
|
| 1106 |
+
"logps/chosen": -316.0,
|
| 1107 |
+
"logps/rejected": -392.0,
|
| 1108 |
+
"loss": 0.0024,
|
| 1109 |
+
"nll_loss": 0.9683593511581421,
|
| 1110 |
+
"rewards/accuracies": 1.0,
|
| 1111 |
+
"rewards/chosen": 17.106250762939453,
|
| 1112 |
+
"rewards/margins": 18.868749618530273,
|
| 1113 |
+
"rewards/rejected": -1.749609351158142,
|
| 1114 |
+
"step": 680
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 0.552,
|
| 1118 |
+
"grad_norm": 0.00968898581539461,
|
| 1119 |
+
"learning_rate": 2.493333333333333e-07,
|
| 1120 |
+
"logits/chosen": -0.03961181640625,
|
| 1121 |
+
"logits/rejected": 0.116943359375,
|
| 1122 |
+
"logps/chosen": -282.0,
|
| 1123 |
+
"logps/rejected": -428.20001220703125,
|
| 1124 |
+
"loss": 0.0011,
|
| 1125 |
+
"nll_loss": 1.0832030773162842,
|
| 1126 |
+
"rewards/accuracies": 1.0,
|
| 1127 |
+
"rewards/chosen": 17.96875,
|
| 1128 |
+
"rewards/margins": 21.024999618530273,
|
| 1129 |
+
"rewards/rejected": -3.087207078933716,
|
| 1130 |
+
"step": 690
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.56,
|
| 1134 |
+
"grad_norm": 0.02206940002406898,
|
| 1135 |
+
"learning_rate": 2.448888888888889e-07,
|
| 1136 |
+
"logits/chosen": 0.05800781399011612,
|
| 1137 |
+
"logits/rejected": 0.04348144680261612,
|
| 1138 |
+
"logps/chosen": -305.8999938964844,
|
| 1139 |
+
"logps/rejected": -411.20001220703125,
|
| 1140 |
+
"loss": 0.0012,
|
| 1141 |
+
"nll_loss": 1.172265648841858,
|
| 1142 |
+
"rewards/accuracies": 1.0,
|
| 1143 |
+
"rewards/chosen": 17.575000762939453,
|
| 1144 |
+
"rewards/margins": 19.975000381469727,
|
| 1145 |
+
"rewards/rejected": -2.412109375,
|
| 1146 |
+
"step": 700
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.568,
|
| 1150 |
+
"grad_norm": 0.01284076138380296,
|
| 1151 |
+
"learning_rate": 2.404444444444444e-07,
|
| 1152 |
+
"logits/chosen": -0.03164062649011612,
|
| 1153 |
+
"logits/rejected": 0.05482788011431694,
|
| 1154 |
+
"logps/chosen": -294.6000061035156,
|
| 1155 |
+
"logps/rejected": -401.6000061035156,
|
| 1156 |
+
"loss": 0.0011,
|
| 1157 |
+
"nll_loss": 1.074609398841858,
|
| 1158 |
+
"rewards/accuracies": 1.0,
|
| 1159 |
+
"rewards/chosen": 18.200000762939453,
|
| 1160 |
+
"rewards/margins": 19.387500762939453,
|
| 1161 |
+
"rewards/rejected": -1.1970703601837158,
|
| 1162 |
+
"step": 710
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.576,
|
| 1166 |
+
"grad_norm": 0.007678737859862069,
|
| 1167 |
+
"learning_rate": 2.3599999999999997e-07,
|
| 1168 |
+
"logits/chosen": -0.03760986402630806,
|
| 1169 |
+
"logits/rejected": 0.05534667894244194,
|
| 1170 |
+
"logps/chosen": -310.20001220703125,
|
| 1171 |
+
"logps/rejected": -413.79998779296875,
|
| 1172 |
+
"loss": 0.0013,
|
| 1173 |
+
"nll_loss": 1.169531226158142,
|
| 1174 |
+
"rewards/accuracies": 1.0,
|
| 1175 |
+
"rewards/chosen": 17.600000381469727,
|
| 1176 |
+
"rewards/margins": 20.212499618530273,
|
| 1177 |
+
"rewards/rejected": -2.596874952316284,
|
| 1178 |
+
"step": 720
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.584,
|
| 1182 |
+
"grad_norm": 0.020744736487327976,
|
| 1183 |
+
"learning_rate": 2.3155555555555553e-07,
|
| 1184 |
+
"logits/chosen": -0.02609863318502903,
|
| 1185 |
+
"logits/rejected": 0.0072021484375,
|
| 1186 |
+
"logps/chosen": -287.0,
|
| 1187 |
+
"logps/rejected": -421.79998779296875,
|
| 1188 |
+
"loss": 0.001,
|
| 1189 |
+
"nll_loss": 1.001953125,
|
| 1190 |
+
"rewards/accuracies": 1.0,
|
| 1191 |
+
"rewards/chosen": 17.375,
|
| 1192 |
+
"rewards/margins": 20.887500762939453,
|
| 1193 |
+
"rewards/rejected": -3.515942335128784,
|
| 1194 |
+
"step": 730
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.592,
|
| 1198 |
+
"grad_norm": 0.010296671008794668,
|
| 1199 |
+
"learning_rate": 2.2711111111111112e-07,
|
| 1200 |
+
"logits/chosen": -0.06264648586511612,
|
| 1201 |
+
"logits/rejected": -0.03441772609949112,
|
| 1202 |
+
"logps/chosen": -281.20001220703125,
|
| 1203 |
+
"logps/rejected": -420.79998779296875,
|
| 1204 |
+
"loss": 0.001,
|
| 1205 |
+
"nll_loss": 0.9945312738418579,
|
| 1206 |
+
"rewards/accuracies": 1.0,
|
| 1207 |
+
"rewards/chosen": 17.787500381469727,
|
| 1208 |
+
"rewards/margins": 20.075000762939453,
|
| 1209 |
+
"rewards/rejected": -2.292163133621216,
|
| 1210 |
+
"step": 740
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 0.6,
|
| 1214 |
+
"grad_norm": 0.01984847155435085,
|
| 1215 |
+
"learning_rate": 2.2266666666666668e-07,
|
| 1216 |
+
"logits/chosen": -0.0015625000232830644,
|
| 1217 |
+
"logits/rejected": 0.05145263671875,
|
| 1218 |
+
"logps/chosen": -328.6000061035156,
|
| 1219 |
+
"logps/rejected": -402.20001220703125,
|
| 1220 |
+
"loss": 0.0013,
|
| 1221 |
+
"nll_loss": 0.98046875,
|
| 1222 |
+
"rewards/accuracies": 1.0,
|
| 1223 |
+
"rewards/chosen": 17.600000381469727,
|
| 1224 |
+
"rewards/margins": 19.737499237060547,
|
| 1225 |
+
"rewards/rejected": -2.1128907203674316,
|
| 1226 |
+
"step": 750
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.608,
|
| 1230 |
+
"grad_norm": 0.011414762899716836,
|
| 1231 |
+
"learning_rate": 2.1822222222222224e-07,
|
| 1232 |
+
"logits/chosen": 0.06040038913488388,
|
| 1233 |
+
"logits/rejected": -0.010241699405014515,
|
| 1234 |
+
"logps/chosen": -287.70001220703125,
|
| 1235 |
+
"logps/rejected": -402.79998779296875,
|
| 1236 |
+
"loss": 0.001,
|
| 1237 |
+
"nll_loss": 1.03125,
|
| 1238 |
+
"rewards/accuracies": 1.0,
|
| 1239 |
+
"rewards/chosen": 17.362499237060547,
|
| 1240 |
+
"rewards/margins": 19.549999237060547,
|
| 1241 |
+
"rewards/rejected": -2.1845703125,
|
| 1242 |
+
"step": 760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.616,
|
| 1246 |
+
"grad_norm": 0.011751018637081315,
|
| 1247 |
+
"learning_rate": 2.1377777777777777e-07,
|
| 1248 |
+
"logits/chosen": -0.10244140774011612,
|
| 1249 |
+
"logits/rejected": -0.09792480617761612,
|
| 1250 |
+
"logps/chosen": -279.29998779296875,
|
| 1251 |
+
"logps/rejected": -444.20001220703125,
|
| 1252 |
+
"loss": 0.0011,
|
| 1253 |
+
"nll_loss": 1.045312523841858,
|
| 1254 |
+
"rewards/accuracies": 1.0,
|
| 1255 |
+
"rewards/chosen": 18.112499237060547,
|
| 1256 |
+
"rewards/margins": 21.924999237060547,
|
| 1257 |
+
"rewards/rejected": -3.8414063453674316,
|
| 1258 |
+
"step": 770
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 0.624,
|
| 1262 |
+
"grad_norm": 0.010739426686833121,
|
| 1263 |
+
"learning_rate": 2.0933333333333333e-07,
|
| 1264 |
+
"logits/chosen": -0.06641845405101776,
|
| 1265 |
+
"logits/rejected": -0.03793945163488388,
|
| 1266 |
+
"logps/chosen": -283.6000061035156,
|
| 1267 |
+
"logps/rejected": -433.6000061035156,
|
| 1268 |
+
"loss": 0.001,
|
| 1269 |
+
"nll_loss": 1.0441405773162842,
|
| 1270 |
+
"rewards/accuracies": 1.0,
|
| 1271 |
+
"rewards/chosen": 18.043750762939453,
|
| 1272 |
+
"rewards/margins": 22.350000381469727,
|
| 1273 |
+
"rewards/rejected": -4.30859375,
|
| 1274 |
+
"step": 780
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.632,
|
| 1278 |
+
"grad_norm": 0.008894669609579957,
|
| 1279 |
+
"learning_rate": 2.048888888888889e-07,
|
| 1280 |
+
"logits/chosen": -0.13934326171875,
|
| 1281 |
+
"logits/rejected": 0.03299560397863388,
|
| 1282 |
+
"logps/chosen": -269.0,
|
| 1283 |
+
"logps/rejected": -435.0,
|
| 1284 |
+
"loss": 0.0026,
|
| 1285 |
+
"nll_loss": 1.0304687023162842,
|
| 1286 |
+
"rewards/accuracies": 1.0,
|
| 1287 |
+
"rewards/chosen": 17.850000381469727,
|
| 1288 |
+
"rewards/margins": 21.799999237060547,
|
| 1289 |
+
"rewards/rejected": -3.9339842796325684,
|
| 1290 |
+
"step": 790
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.64,
|
| 1294 |
+
"grad_norm": 0.00781796371528786,
|
| 1295 |
+
"learning_rate": 2.0044444444444445e-07,
|
| 1296 |
+
"logits/chosen": -0.136871337890625,
|
| 1297 |
+
"logits/rejected": -0.06427001953125,
|
| 1298 |
+
"logps/chosen": -301.79998779296875,
|
| 1299 |
+
"logps/rejected": -441.20001220703125,
|
| 1300 |
+
"loss": 0.0011,
|
| 1301 |
+
"nll_loss": 1.0535156726837158,
|
| 1302 |
+
"rewards/accuracies": 1.0,
|
| 1303 |
+
"rewards/chosen": 17.856250762939453,
|
| 1304 |
+
"rewards/margins": 22.362499237060547,
|
| 1305 |
+
"rewards/rejected": -4.500781059265137,
|
| 1306 |
+
"step": 800
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.648,
|
| 1310 |
+
"grad_norm": 0.006941095936510498,
|
| 1311 |
+
"learning_rate": 1.96e-07,
|
| 1312 |
+
"logits/chosen": -0.06254883110523224,
|
| 1313 |
+
"logits/rejected": -0.03428955003619194,
|
| 1314 |
+
"logps/chosen": -317.79998779296875,
|
| 1315 |
+
"logps/rejected": -424.6000061035156,
|
| 1316 |
+
"loss": 0.001,
|
| 1317 |
+
"nll_loss": 1.033593773841858,
|
| 1318 |
+
"rewards/accuracies": 1.0,
|
| 1319 |
+
"rewards/chosen": 17.875,
|
| 1320 |
+
"rewards/margins": 20.450000762939453,
|
| 1321 |
+
"rewards/rejected": -2.5609374046325684,
|
| 1322 |
+
"step": 810
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.656,
|
| 1326 |
+
"grad_norm": 0.01161314226877997,
|
| 1327 |
+
"learning_rate": 1.9155555555555554e-07,
|
| 1328 |
+
"logits/chosen": -0.06083984300494194,
|
| 1329 |
+
"logits/rejected": 0.07171630859375,
|
| 1330 |
+
"logps/chosen": -300.0,
|
| 1331 |
+
"logps/rejected": -426.3999938964844,
|
| 1332 |
+
"loss": 0.0011,
|
| 1333 |
+
"nll_loss": 1.101171851158142,
|
| 1334 |
+
"rewards/accuracies": 1.0,
|
| 1335 |
+
"rewards/chosen": 18.8125,
|
| 1336 |
+
"rewards/margins": 22.337499618530273,
|
| 1337 |
+
"rewards/rejected": -3.5337891578674316,
|
| 1338 |
+
"step": 820
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 0.664,
|
| 1342 |
+
"grad_norm": 0.04491917017356749,
|
| 1343 |
+
"learning_rate": 1.871111111111111e-07,
|
| 1344 |
+
"logits/chosen": -0.02180175855755806,
|
| 1345 |
+
"logits/rejected": -0.0004760742303915322,
|
| 1346 |
+
"logps/chosen": -272.1000061035156,
|
| 1347 |
+
"logps/rejected": -424.20001220703125,
|
| 1348 |
+
"loss": 0.001,
|
| 1349 |
+
"nll_loss": 1.0390625,
|
| 1350 |
+
"rewards/accuracies": 1.0,
|
| 1351 |
+
"rewards/chosen": 18.962499618530273,
|
| 1352 |
+
"rewards/margins": 22.612499237060547,
|
| 1353 |
+
"rewards/rejected": -3.6480469703674316,
|
| 1354 |
+
"step": 830
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.672,
|
| 1358 |
+
"grad_norm": 0.014327809455407186,
|
| 1359 |
+
"learning_rate": 1.8266666666666666e-07,
|
| 1360 |
+
"logits/chosen": -0.06074218824505806,
|
| 1361 |
+
"logits/rejected": -0.0008728027460165322,
|
| 1362 |
+
"logps/chosen": -304.8999938964844,
|
| 1363 |
+
"logps/rejected": -395.3999938964844,
|
| 1364 |
+
"loss": 0.0016,
|
| 1365 |
+
"nll_loss": 1.0378906726837158,
|
| 1366 |
+
"rewards/accuracies": 1.0,
|
| 1367 |
+
"rewards/chosen": 18.612499237060547,
|
| 1368 |
+
"rewards/margins": 19.75,
|
| 1369 |
+
"rewards/rejected": -1.1486327648162842,
|
| 1370 |
+
"step": 840
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.68,
|
| 1374 |
+
"grad_norm": 0.010011390264830125,
|
| 1375 |
+
"learning_rate": 1.7822222222222222e-07,
|
| 1376 |
+
"logits/chosen": -0.143798828125,
|
| 1377 |
+
"logits/rejected": -0.04060058668255806,
|
| 1378 |
+
"logps/chosen": -308.1000061035156,
|
| 1379 |
+
"logps/rejected": -433.6000061035156,
|
| 1380 |
+
"loss": 0.0031,
|
| 1381 |
+
"nll_loss": 1.0398437976837158,
|
| 1382 |
+
"rewards/accuracies": 1.0,
|
| 1383 |
+
"rewards/chosen": 18.887500762939453,
|
| 1384 |
+
"rewards/margins": 21.774999618530273,
|
| 1385 |
+
"rewards/rejected": -2.8993163108825684,
|
| 1386 |
+
"step": 850
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 0.688,
|
| 1390 |
+
"grad_norm": 0.006838199221939372,
|
| 1391 |
+
"learning_rate": 1.7377777777777778e-07,
|
| 1392 |
+
"logits/chosen": 0.041259765625,
|
| 1393 |
+
"logits/rejected": 0.13017578423023224,
|
| 1394 |
+
"logps/chosen": -293.0,
|
| 1395 |
+
"logps/rejected": -432.0,
|
| 1396 |
+
"loss": 0.001,
|
| 1397 |
+
"nll_loss": 1.0363280773162842,
|
| 1398 |
+
"rewards/accuracies": 1.0,
|
| 1399 |
+
"rewards/chosen": 18.375,
|
| 1400 |
+
"rewards/margins": 21.825000762939453,
|
| 1401 |
+
"rewards/rejected": -3.4496092796325684,
|
| 1402 |
+
"step": 860
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.696,
|
| 1406 |
+
"grad_norm": 0.011644152757164477,
|
| 1407 |
+
"learning_rate": 1.6933333333333334e-07,
|
| 1408 |
+
"logits/chosen": 0.08931884914636612,
|
| 1409 |
+
"logits/rejected": 0.07216797024011612,
|
| 1410 |
+
"logps/chosen": -299.3999938964844,
|
| 1411 |
+
"logps/rejected": -411.0,
|
| 1412 |
+
"loss": 0.001,
|
| 1413 |
+
"nll_loss": 0.9429687261581421,
|
| 1414 |
+
"rewards/accuracies": 1.0,
|
| 1415 |
+
"rewards/chosen": 19.487499237060547,
|
| 1416 |
+
"rewards/margins": 21.487499237060547,
|
| 1417 |
+
"rewards/rejected": -2.010546922683716,
|
| 1418 |
+
"step": 870
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.704,
|
| 1422 |
+
"grad_norm": 0.01033541957175912,
|
| 1423 |
+
"learning_rate": 1.6488888888888887e-07,
|
| 1424 |
+
"logits/chosen": 0.09145507961511612,
|
| 1425 |
+
"logits/rejected": 0.15330810844898224,
|
| 1426 |
+
"logps/chosen": -304.79998779296875,
|
| 1427 |
+
"logps/rejected": -400.20001220703125,
|
| 1428 |
+
"loss": 0.0009,
|
| 1429 |
+
"nll_loss": 0.8675781488418579,
|
| 1430 |
+
"rewards/accuracies": 1.0,
|
| 1431 |
+
"rewards/chosen": 18.4375,
|
| 1432 |
+
"rewards/margins": 20.612499237060547,
|
| 1433 |
+
"rewards/rejected": -2.170703172683716,
|
| 1434 |
+
"step": 880
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.712,
|
| 1438 |
+
"grad_norm": 0.008717665163684966,
|
| 1439 |
+
"learning_rate": 1.6044444444444443e-07,
|
| 1440 |
+
"logits/chosen": 0.10676269233226776,
|
| 1441 |
+
"logits/rejected": 0.12166748195886612,
|
| 1442 |
+
"logps/chosen": -274.5,
|
| 1443 |
+
"logps/rejected": -426.6000061035156,
|
| 1444 |
+
"loss": 0.001,
|
| 1445 |
+
"nll_loss": 0.9921875,
|
| 1446 |
+
"rewards/accuracies": 1.0,
|
| 1447 |
+
"rewards/chosen": 19.149999618530273,
|
| 1448 |
+
"rewards/margins": 22.774999618530273,
|
| 1449 |
+
"rewards/rejected": -3.625,
|
| 1450 |
+
"step": 890
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.72,
|
| 1454 |
+
"grad_norm": 0.007924558579110246,
|
| 1455 |
+
"learning_rate": 1.56e-07,
|
| 1456 |
+
"logits/chosen": -0.11391601711511612,
|
| 1457 |
+
"logits/rejected": -0.02934570237994194,
|
| 1458 |
+
"logps/chosen": -266.8999938964844,
|
| 1459 |
+
"logps/rejected": -427.20001220703125,
|
| 1460 |
+
"loss": 0.0035,
|
| 1461 |
+
"nll_loss": 0.981249988079071,
|
| 1462 |
+
"rewards/accuracies": 1.0,
|
| 1463 |
+
"rewards/chosen": 18.662500381469727,
|
| 1464 |
+
"rewards/margins": 21.200000762939453,
|
| 1465 |
+
"rewards/rejected": -2.541015625,
|
| 1466 |
+
"step": 900
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.728,
|
| 1470 |
+
"grad_norm": 0.07578643720533651,
|
| 1471 |
+
"learning_rate": 1.5155555555555555e-07,
|
| 1472 |
+
"logits/chosen": 0.07387695461511612,
|
| 1473 |
+
"logits/rejected": 0.15861816704273224,
|
| 1474 |
+
"logps/chosen": -239.5500030517578,
|
| 1475 |
+
"logps/rejected": -417.3999938964844,
|
| 1476 |
+
"loss": 0.0011,
|
| 1477 |
+
"nll_loss": 1.0226562023162842,
|
| 1478 |
+
"rewards/accuracies": 1.0,
|
| 1479 |
+
"rewards/chosen": 18.8125,
|
| 1480 |
+
"rewards/margins": 21.987499237060547,
|
| 1481 |
+
"rewards/rejected": -3.1800780296325684,
|
| 1482 |
+
"step": 910
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 0.736,
|
| 1486 |
+
"grad_norm": 0.009494584129256885,
|
| 1487 |
+
"learning_rate": 1.4711111111111111e-07,
|
| 1488 |
+
"logits/chosen": -0.004748535342514515,
|
| 1489 |
+
"logits/rejected": 0.05303344875574112,
|
| 1490 |
+
"logps/chosen": -327.79998779296875,
|
| 1491 |
+
"logps/rejected": -390.79998779296875,
|
| 1492 |
+
"loss": 0.001,
|
| 1493 |
+
"nll_loss": 1.0050780773162842,
|
| 1494 |
+
"rewards/accuracies": 1.0,
|
| 1495 |
+
"rewards/chosen": 19.056249618530273,
|
| 1496 |
+
"rewards/margins": 20.575000762939453,
|
| 1497 |
+
"rewards/rejected": -1.507421851158142,
|
| 1498 |
+
"step": 920
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 0.744,
|
| 1502 |
+
"grad_norm": 0.017667443139199403,
|
| 1503 |
+
"learning_rate": 1.4266666666666665e-07,
|
| 1504 |
+
"logits/chosen": -0.054931640625,
|
| 1505 |
+
"logits/rejected": 0.02968749962747097,
|
| 1506 |
+
"logps/chosen": -261.0,
|
| 1507 |
+
"logps/rejected": -418.0,
|
| 1508 |
+
"loss": 0.001,
|
| 1509 |
+
"nll_loss": 0.9671875238418579,
|
| 1510 |
+
"rewards/accuracies": 1.0,
|
| 1511 |
+
"rewards/chosen": 19.137500762939453,
|
| 1512 |
+
"rewards/margins": 21.331249237060547,
|
| 1513 |
+
"rewards/rejected": -2.1927733421325684,
|
| 1514 |
+
"step": 930
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.752,
|
| 1518 |
+
"grad_norm": 0.010899226100416153,
|
| 1519 |
+
"learning_rate": 1.382222222222222e-07,
|
| 1520 |
+
"logits/chosen": -0.0184326171875,
|
| 1521 |
+
"logits/rejected": 0.11514892429113388,
|
| 1522 |
+
"logps/chosen": -288.6000061035156,
|
| 1523 |
+
"logps/rejected": -408.3999938964844,
|
| 1524 |
+
"loss": 0.0465,
|
| 1525 |
+
"nll_loss": 1.074609398841858,
|
| 1526 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1527 |
+
"rewards/chosen": 19.024999618530273,
|
| 1528 |
+
"rewards/margins": 21.262500762939453,
|
| 1529 |
+
"rewards/rejected": -2.2400145530700684,
|
| 1530 |
+
"step": 940
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.76,
|
| 1534 |
+
"grad_norm": 0.025779260274360533,
|
| 1535 |
+
"learning_rate": 1.3377777777777777e-07,
|
| 1536 |
+
"logits/chosen": 0.02338867262005806,
|
| 1537 |
+
"logits/rejected": 0.10743407905101776,
|
| 1538 |
+
"logps/chosen": -330.6000061035156,
|
| 1539 |
+
"logps/rejected": -398.20001220703125,
|
| 1540 |
+
"loss": 0.001,
|
| 1541 |
+
"nll_loss": 1.0070312023162842,
|
| 1542 |
+
"rewards/accuracies": 1.0,
|
| 1543 |
+
"rewards/chosen": 18.756250381469727,
|
| 1544 |
+
"rewards/margins": 20.049999237060547,
|
| 1545 |
+
"rewards/rejected": -1.2770264148712158,
|
| 1546 |
+
"step": 950
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 0.768,
|
| 1550 |
+
"grad_norm": 0.025413812683072193,
|
| 1551 |
+
"learning_rate": 1.2933333333333333e-07,
|
| 1552 |
+
"logits/chosen": 0.05156249925494194,
|
| 1553 |
+
"logits/rejected": 0.17954102158546448,
|
| 1554 |
+
"logps/chosen": -291.3999938964844,
|
| 1555 |
+
"logps/rejected": -413.0,
|
| 1556 |
+
"loss": 0.001,
|
| 1557 |
+
"nll_loss": 0.958984375,
|
| 1558 |
+
"rewards/accuracies": 1.0,
|
| 1559 |
+
"rewards/chosen": 19.924999237060547,
|
| 1560 |
+
"rewards/margins": 21.637500762939453,
|
| 1561 |
+
"rewards/rejected": -1.715429663658142,
|
| 1562 |
+
"step": 960
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.776,
|
| 1566 |
+
"grad_norm": 0.008806349125691143,
|
| 1567 |
+
"learning_rate": 1.2488888888888889e-07,
|
| 1568 |
+
"logits/chosen": 0.02752685546875,
|
| 1569 |
+
"logits/rejected": 0.08073730766773224,
|
| 1570 |
+
"logps/chosen": -264.6000061035156,
|
| 1571 |
+
"logps/rejected": -407.6000061035156,
|
| 1572 |
+
"loss": 0.0009,
|
| 1573 |
+
"nll_loss": 0.8984375,
|
| 1574 |
+
"rewards/accuracies": 1.0,
|
| 1575 |
+
"rewards/chosen": 19.0,
|
| 1576 |
+
"rewards/margins": 20.237499237060547,
|
| 1577 |
+
"rewards/rejected": -1.255761742591858,
|
| 1578 |
+
"step": 970
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.784,
|
| 1582 |
+
"grad_norm": 0.0098690579475145,
|
| 1583 |
+
"learning_rate": 1.2044444444444445e-07,
|
| 1584 |
+
"logits/chosen": 0.01823730394244194,
|
| 1585 |
+
"logits/rejected": 0.11467285454273224,
|
| 1586 |
+
"logps/chosen": -310.20001220703125,
|
| 1587 |
+
"logps/rejected": -413.6000061035156,
|
| 1588 |
+
"loss": 0.001,
|
| 1589 |
+
"nll_loss": 0.9722656011581421,
|
| 1590 |
+
"rewards/accuracies": 1.0,
|
| 1591 |
+
"rewards/chosen": 20.237499237060547,
|
| 1592 |
+
"rewards/margins": 22.325000762939453,
|
| 1593 |
+
"rewards/rejected": -2.0926756858825684,
|
| 1594 |
+
"step": 980
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.792,
|
| 1598 |
+
"grad_norm": 0.010627117006645067,
|
| 1599 |
+
"learning_rate": 1.16e-07,
|
| 1600 |
+
"logits/chosen": -0.07388915866613388,
|
| 1601 |
+
"logits/rejected": 0.07421875,
|
| 1602 |
+
"logps/chosen": -342.8999938964844,
|
| 1603 |
+
"logps/rejected": -412.79998779296875,
|
| 1604 |
+
"loss": 0.001,
|
| 1605 |
+
"nll_loss": 1.040624976158142,
|
| 1606 |
+
"rewards/accuracies": 1.0,
|
| 1607 |
+
"rewards/chosen": 18.96875,
|
| 1608 |
+
"rewards/margins": 21.049999237060547,
|
| 1609 |
+
"rewards/rejected": -2.0562500953674316,
|
| 1610 |
+
"step": 990
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.8,
|
| 1614 |
+
"grad_norm": 0.04218956204057828,
|
| 1615 |
+
"learning_rate": 1.1155555555555555e-07,
|
| 1616 |
+
"logits/chosen": -0.0010253905784338713,
|
| 1617 |
+
"logits/rejected": 0.06143798679113388,
|
| 1618 |
+
"logps/chosen": -290.70001220703125,
|
| 1619 |
+
"logps/rejected": -412.6000061035156,
|
| 1620 |
+
"loss": 0.0011,
|
| 1621 |
+
"nll_loss": 1.09375,
|
| 1622 |
+
"rewards/accuracies": 1.0,
|
| 1623 |
+
"rewards/chosen": 19.962499618530273,
|
| 1624 |
+
"rewards/margins": 22.0,
|
| 1625 |
+
"rewards/rejected": -2.0201172828674316,
|
| 1626 |
+
"step": 1000
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.808,
|
| 1630 |
+
"grad_norm": 0.014065236545009229,
|
| 1631 |
+
"learning_rate": 1.0711111111111111e-07,
|
| 1632 |
+
"logits/chosen": 0.20627442002296448,
|
| 1633 |
+
"logits/rejected": 0.24067382514476776,
|
| 1634 |
+
"logps/chosen": -263.79998779296875,
|
| 1635 |
+
"logps/rejected": -406.0,
|
| 1636 |
+
"loss": 0.0009,
|
| 1637 |
+
"nll_loss": 0.910937488079071,
|
| 1638 |
+
"rewards/accuracies": 1.0,
|
| 1639 |
+
"rewards/chosen": 19.4375,
|
| 1640 |
+
"rewards/margins": 21.524999618530273,
|
| 1641 |
+
"rewards/rejected": -2.0770020484924316,
|
| 1642 |
+
"step": 1010
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.816,
|
| 1646 |
+
"grad_norm": 0.008101550660192351,
|
| 1647 |
+
"learning_rate": 1.0266666666666666e-07,
|
| 1648 |
+
"logits/chosen": 0.13178710639476776,
|
| 1649 |
+
"logits/rejected": 0.11772461235523224,
|
| 1650 |
+
"logps/chosen": -241.75,
|
| 1651 |
+
"logps/rejected": -394.20001220703125,
|
| 1652 |
+
"loss": 0.0009,
|
| 1653 |
+
"nll_loss": 0.893359363079071,
|
| 1654 |
+
"rewards/accuracies": 1.0,
|
| 1655 |
+
"rewards/chosen": 20.100000381469727,
|
| 1656 |
+
"rewards/margins": 20.950000762939453,
|
| 1657 |
+
"rewards/rejected": -0.851611316204071,
|
| 1658 |
+
"step": 1020
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.824,
|
| 1662 |
+
"grad_norm": 0.020212312188331298,
|
| 1663 |
+
"learning_rate": 9.822222222222222e-08,
|
| 1664 |
+
"logits/chosen": 0.03201904147863388,
|
| 1665 |
+
"logits/rejected": 0.09085693210363388,
|
| 1666 |
+
"logps/chosen": -283.8999938964844,
|
| 1667 |
+
"logps/rejected": -410.79998779296875,
|
| 1668 |
+
"loss": 0.001,
|
| 1669 |
+
"nll_loss": 0.98828125,
|
| 1670 |
+
"rewards/accuracies": 1.0,
|
| 1671 |
+
"rewards/chosen": 20.662500381469727,
|
| 1672 |
+
"rewards/margins": 22.450000762939453,
|
| 1673 |
+
"rewards/rejected": -1.788671851158142,
|
| 1674 |
+
"step": 1030
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.832,
|
| 1678 |
+
"grad_norm": 0.01018779281956963,
|
| 1679 |
+
"learning_rate": 9.377777777777778e-08,
|
| 1680 |
+
"logits/chosen": 0.005297851748764515,
|
| 1681 |
+
"logits/rejected": 0.03730468824505806,
|
| 1682 |
+
"logps/chosen": -293.1000061035156,
|
| 1683 |
+
"logps/rejected": -418.20001220703125,
|
| 1684 |
+
"loss": 0.0391,
|
| 1685 |
+
"nll_loss": 1.017578125,
|
| 1686 |
+
"rewards/accuracies": 0.987500011920929,
|
| 1687 |
+
"rewards/chosen": 20.100000381469727,
|
| 1688 |
+
"rewards/margins": 21.225000381469727,
|
| 1689 |
+
"rewards/rejected": -1.11474609375,
|
| 1690 |
+
"step": 1040
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.84,
|
| 1694 |
+
"grad_norm": 0.02497328379034309,
|
| 1695 |
+
"learning_rate": 8.933333333333333e-08,
|
| 1696 |
+
"logits/chosen": 0.03041992150247097,
|
| 1697 |
+
"logits/rejected": 0.05394287034869194,
|
| 1698 |
+
"logps/chosen": -302.5,
|
| 1699 |
+
"logps/rejected": -412.79998779296875,
|
| 1700 |
+
"loss": 0.0011,
|
| 1701 |
+
"nll_loss": 1.077734351158142,
|
| 1702 |
+
"rewards/accuracies": 1.0,
|
| 1703 |
+
"rewards/chosen": 20.100000381469727,
|
| 1704 |
+
"rewards/margins": 21.3125,
|
| 1705 |
+
"rewards/rejected": -1.21630859375,
|
| 1706 |
+
"step": 1050
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.848,
|
| 1710 |
+
"grad_norm": 0.012861381907556958,
|
| 1711 |
+
"learning_rate": 8.488888888888889e-08,
|
| 1712 |
+
"logits/chosen": 0.03743896633386612,
|
| 1713 |
+
"logits/rejected": 0.131591796875,
|
| 1714 |
+
"logps/chosen": -272.79998779296875,
|
| 1715 |
+
"logps/rejected": -407.20001220703125,
|
| 1716 |
+
"loss": 0.0009,
|
| 1717 |
+
"nll_loss": 0.9085937738418579,
|
| 1718 |
+
"rewards/accuracies": 1.0,
|
| 1719 |
+
"rewards/chosen": 20.412500381469727,
|
| 1720 |
+
"rewards/margins": 21.924999237060547,
|
| 1721 |
+
"rewards/rejected": -1.5066406726837158,
|
| 1722 |
+
"step": 1060
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 0.856,
|
| 1726 |
+
"grad_norm": 0.009207668604585216,
|
| 1727 |
+
"learning_rate": 8.044444444444445e-08,
|
| 1728 |
+
"logits/chosen": 0.04277343675494194,
|
| 1729 |
+
"logits/rejected": 0.11362304538488388,
|
| 1730 |
+
"logps/chosen": -288.20001220703125,
|
| 1731 |
+
"logps/rejected": -404.6000061035156,
|
| 1732 |
+
"loss": 0.0011,
|
| 1733 |
+
"nll_loss": 1.010156273841858,
|
| 1734 |
+
"rewards/accuracies": 1.0,
|
| 1735 |
+
"rewards/chosen": 19.450000762939453,
|
| 1736 |
+
"rewards/margins": 20.737499237060547,
|
| 1737 |
+
"rewards/rejected": -1.2658202648162842,
|
| 1738 |
+
"step": 1070
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.864,
|
| 1742 |
+
"grad_norm": 0.014789555122865826,
|
| 1743 |
+
"learning_rate": 7.599999999999999e-08,
|
| 1744 |
+
"logits/chosen": -0.03659667819738388,
|
| 1745 |
+
"logits/rejected": 0.09859619289636612,
|
| 1746 |
+
"logps/chosen": -269.79998779296875,
|
| 1747 |
+
"logps/rejected": -416.20001220703125,
|
| 1748 |
+
"loss": 0.001,
|
| 1749 |
+
"nll_loss": 1.0222656726837158,
|
| 1750 |
+
"rewards/accuracies": 1.0,
|
| 1751 |
+
"rewards/chosen": 19.387500762939453,
|
| 1752 |
+
"rewards/margins": 21.487499237060547,
|
| 1753 |
+
"rewards/rejected": -2.081835985183716,
|
| 1754 |
+
"step": 1080
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.872,
|
| 1758 |
+
"grad_norm": 0.01967144963219391,
|
| 1759 |
+
"learning_rate": 7.155555555555555e-08,
|
| 1760 |
+
"logits/chosen": 0.050048828125,
|
| 1761 |
+
"logits/rejected": 0.12807616591453552,
|
| 1762 |
+
"logps/chosen": -283.79998779296875,
|
| 1763 |
+
"logps/rejected": -398.3999938964844,
|
| 1764 |
+
"loss": 0.001,
|
| 1765 |
+
"nll_loss": 0.9683593511581421,
|
| 1766 |
+
"rewards/accuracies": 1.0,
|
| 1767 |
+
"rewards/chosen": 19.975000381469727,
|
| 1768 |
+
"rewards/margins": 21.174999237060547,
|
| 1769 |
+
"rewards/rejected": -1.1923339366912842,
|
| 1770 |
+
"step": 1090
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 0.88,
|
| 1774 |
+
"grad_norm": 0.01262612387400558,
|
| 1775 |
+
"learning_rate": 6.71111111111111e-08,
|
| 1776 |
+
"logits/chosen": -0.08295898139476776,
|
| 1777 |
+
"logits/rejected": 0.06423339992761612,
|
| 1778 |
+
"logps/chosen": -340.5,
|
| 1779 |
+
"logps/rejected": -409.3999938964844,
|
| 1780 |
+
"loss": 0.001,
|
| 1781 |
+
"nll_loss": 1.01171875,
|
| 1782 |
+
"rewards/accuracies": 1.0,
|
| 1783 |
+
"rewards/chosen": 19.274999618530273,
|
| 1784 |
+
"rewards/margins": 21.024999618530273,
|
| 1785 |
+
"rewards/rejected": -1.7332031726837158,
|
| 1786 |
+
"step": 1100
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 0.888,
|
| 1790 |
+
"grad_norm": 0.013672600008616023,
|
| 1791 |
+
"learning_rate": 6.266666666666666e-08,
|
| 1792 |
+
"logits/chosen": 0.16115722060203552,
|
| 1793 |
+
"logits/rejected": 0.17141112685203552,
|
| 1794 |
+
"logps/chosen": -246.89999389648438,
|
| 1795 |
+
"logps/rejected": -415.20001220703125,
|
| 1796 |
+
"loss": 0.0009,
|
| 1797 |
+
"nll_loss": 0.9058593511581421,
|
| 1798 |
+
"rewards/accuracies": 1.0,
|
| 1799 |
+
"rewards/chosen": 19.475000381469727,
|
| 1800 |
+
"rewards/margins": 21.612499237060547,
|
| 1801 |
+
"rewards/rejected": -2.133984327316284,
|
| 1802 |
+
"step": 1110
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.896,
|
| 1806 |
+
"grad_norm": 0.01751993042439502,
|
| 1807 |
+
"learning_rate": 5.822222222222222e-08,
|
| 1808 |
+
"logits/chosen": 0.06224365159869194,
|
| 1809 |
+
"logits/rejected": 0.15957030653953552,
|
| 1810 |
+
"logps/chosen": -226.8000030517578,
|
| 1811 |
+
"logps/rejected": -421.79998779296875,
|
| 1812 |
+
"loss": 0.001,
|
| 1813 |
+
"nll_loss": 0.981249988079071,
|
| 1814 |
+
"rewards/accuracies": 1.0,
|
| 1815 |
+
"rewards/chosen": 21.274999618530273,
|
| 1816 |
+
"rewards/margins": 22.875,
|
| 1817 |
+
"rewards/rejected": -1.5966796875,
|
| 1818 |
+
"step": 1120
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 0.904,
|
| 1822 |
+
"grad_norm": 0.014517125852295223,
|
| 1823 |
+
"learning_rate": 5.377777777777778e-08,
|
| 1824 |
+
"logits/chosen": -0.0188446044921875,
|
| 1825 |
+
"logits/rejected": 0.1414794921875,
|
| 1826 |
+
"logps/chosen": -283.5,
|
| 1827 |
+
"logps/rejected": -416.20001220703125,
|
| 1828 |
+
"loss": 0.001,
|
| 1829 |
+
"nll_loss": 0.995312511920929,
|
| 1830 |
+
"rewards/accuracies": 1.0,
|
| 1831 |
+
"rewards/chosen": 19.987499237060547,
|
| 1832 |
+
"rewards/margins": 22.587499618530273,
|
| 1833 |
+
"rewards/rejected": -2.6039061546325684,
|
| 1834 |
+
"step": 1130
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.912,
|
| 1838 |
+
"grad_norm": 0.02080473594604604,
|
| 1839 |
+
"learning_rate": 4.933333333333333e-08,
|
| 1840 |
+
"logits/chosen": 0.041839599609375,
|
| 1841 |
+
"logits/rejected": 0.14179687201976776,
|
| 1842 |
+
"logps/chosen": -338.70001220703125,
|
| 1843 |
+
"logps/rejected": -391.3999938964844,
|
| 1844 |
+
"loss": 0.0009,
|
| 1845 |
+
"nll_loss": 0.916796863079071,
|
| 1846 |
+
"rewards/accuracies": 1.0,
|
| 1847 |
+
"rewards/chosen": 19.575000762939453,
|
| 1848 |
+
"rewards/margins": 21.100000381469727,
|
| 1849 |
+
"rewards/rejected": -1.5293457508087158,
|
| 1850 |
+
"step": 1140
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.92,
|
| 1854 |
+
"grad_norm": 0.020719630349212197,
|
| 1855 |
+
"learning_rate": 4.4888888888888885e-08,
|
| 1856 |
+
"logits/chosen": 0.04411621019244194,
|
| 1857 |
+
"logits/rejected": 0.07084961235523224,
|
| 1858 |
+
"logps/chosen": -269.3999938964844,
|
| 1859 |
+
"logps/rejected": -404.79998779296875,
|
| 1860 |
+
"loss": 0.001,
|
| 1861 |
+
"nll_loss": 0.961718738079071,
|
| 1862 |
+
"rewards/accuracies": 1.0,
|
| 1863 |
+
"rewards/chosen": 19.725000381469727,
|
| 1864 |
+
"rewards/margins": 21.475000381469727,
|
| 1865 |
+
"rewards/rejected": -1.7581055164337158,
|
| 1866 |
+
"step": 1150
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.928,
|
| 1870 |
+
"grad_norm": 0.008502519454361718,
|
| 1871 |
+
"learning_rate": 4.044444444444444e-08,
|
| 1872 |
+
"logits/chosen": -0.06318359076976776,
|
| 1873 |
+
"logits/rejected": 0.095306396484375,
|
| 1874 |
+
"logps/chosen": -291.3999938964844,
|
| 1875 |
+
"logps/rejected": -408.0,
|
| 1876 |
+
"loss": 0.0009,
|
| 1877 |
+
"nll_loss": 0.875781238079071,
|
| 1878 |
+
"rewards/accuracies": 1.0,
|
| 1879 |
+
"rewards/chosen": 20.287500381469727,
|
| 1880 |
+
"rewards/margins": 22.037500381469727,
|
| 1881 |
+
"rewards/rejected": -1.7532227039337158,
|
| 1882 |
+
"step": 1160
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.936,
|
| 1886 |
+
"grad_norm": 0.010182005409104732,
|
| 1887 |
+
"learning_rate": 3.6e-08,
|
| 1888 |
+
"logits/chosen": -0.02708740159869194,
|
| 1889 |
+
"logits/rejected": 0.07539062201976776,
|
| 1890 |
+
"logps/chosen": -297.70001220703125,
|
| 1891 |
+
"logps/rejected": -411.3999938964844,
|
| 1892 |
+
"loss": 0.0011,
|
| 1893 |
+
"nll_loss": 1.0515625476837158,
|
| 1894 |
+
"rewards/accuracies": 1.0,
|
| 1895 |
+
"rewards/chosen": 19.549999237060547,
|
| 1896 |
+
"rewards/margins": 21.25,
|
| 1897 |
+
"rewards/rejected": -1.7091796398162842,
|
| 1898 |
+
"step": 1170
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.944,
|
| 1902 |
+
"grad_norm": 0.014883019162230187,
|
| 1903 |
+
"learning_rate": 3.155555555555556e-08,
|
| 1904 |
+
"logits/chosen": -0.02933349646627903,
|
| 1905 |
+
"logits/rejected": 0.08122558891773224,
|
| 1906 |
+
"logps/chosen": -297.70001220703125,
|
| 1907 |
+
"logps/rejected": -410.0,
|
| 1908 |
+
"loss": 0.0017,
|
| 1909 |
+
"nll_loss": 0.9140625,
|
| 1910 |
+
"rewards/accuracies": 1.0,
|
| 1911 |
+
"rewards/chosen": 19.962499618530273,
|
| 1912 |
+
"rewards/margins": 21.512500762939453,
|
| 1913 |
+
"rewards/rejected": -1.5568358898162842,
|
| 1914 |
+
"step": 1180
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.952,
|
| 1918 |
+
"grad_norm": 0.006673830289265544,
|
| 1919 |
+
"learning_rate": 2.7111111111111108e-08,
|
| 1920 |
+
"logits/chosen": -0.1202392578125,
|
| 1921 |
+
"logits/rejected": -0.01416015625,
|
| 1922 |
+
"logps/chosen": -267.1000061035156,
|
| 1923 |
+
"logps/rejected": -421.79998779296875,
|
| 1924 |
+
"loss": 0.0015,
|
| 1925 |
+
"nll_loss": 0.9007812738418579,
|
| 1926 |
+
"rewards/accuracies": 1.0,
|
| 1927 |
+
"rewards/chosen": 18.825000762939453,
|
| 1928 |
+
"rewards/margins": 21.350000381469727,
|
| 1929 |
+
"rewards/rejected": -2.530468702316284,
|
| 1930 |
+
"step": 1190
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 0.96,
|
| 1934 |
+
"grad_norm": 0.016858387886967145,
|
| 1935 |
+
"learning_rate": 2.2666666666666668e-08,
|
| 1936 |
+
"logits/chosen": 0.02562255784869194,
|
| 1937 |
+
"logits/rejected": 0.12241210788488388,
|
| 1938 |
+
"logps/chosen": -303.20001220703125,
|
| 1939 |
+
"logps/rejected": -397.0,
|
| 1940 |
+
"loss": 0.001,
|
| 1941 |
+
"nll_loss": 1.024999976158142,
|
| 1942 |
+
"rewards/accuracies": 1.0,
|
| 1943 |
+
"rewards/chosen": 20.774999618530273,
|
| 1944 |
+
"rewards/margins": 22.137500762939453,
|
| 1945 |
+
"rewards/rejected": -1.382421851158142,
|
| 1946 |
+
"step": 1200
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.968,
|
| 1950 |
+
"grad_norm": 0.014032183883427854,
|
| 1951 |
+
"learning_rate": 1.822222222222222e-08,
|
| 1952 |
+
"logits/chosen": 0.02207031287252903,
|
| 1953 |
+
"logits/rejected": 0.04111327975988388,
|
| 1954 |
+
"logps/chosen": -264.3999938964844,
|
| 1955 |
+
"logps/rejected": -418.20001220703125,
|
| 1956 |
+
"loss": 0.0009,
|
| 1957 |
+
"nll_loss": 0.9339843988418579,
|
| 1958 |
+
"rewards/accuracies": 1.0,
|
| 1959 |
+
"rewards/chosen": 19.412500381469727,
|
| 1960 |
+
"rewards/margins": 22.100000381469727,
|
| 1961 |
+
"rewards/rejected": -2.688281297683716,
|
| 1962 |
+
"step": 1210
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.976,
|
| 1966 |
+
"grad_norm": 0.013396264787771736,
|
| 1967 |
+
"learning_rate": 1.3777777777777778e-08,
|
| 1968 |
+
"logits/chosen": 0.06943359225988388,
|
| 1969 |
+
"logits/rejected": 0.160491943359375,
|
| 1970 |
+
"logps/chosen": -273.25,
|
| 1971 |
+
"logps/rejected": -424.6000061035156,
|
| 1972 |
+
"loss": 0.001,
|
| 1973 |
+
"nll_loss": 0.955859363079071,
|
| 1974 |
+
"rewards/accuracies": 1.0,
|
| 1975 |
+
"rewards/chosen": 20.075000762939453,
|
| 1976 |
+
"rewards/margins": 22.850000381469727,
|
| 1977 |
+
"rewards/rejected": -2.7822265625,
|
| 1978 |
+
"step": 1220
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 0.984,
|
| 1982 |
+
"grad_norm": 0.01902150518424847,
|
| 1983 |
+
"learning_rate": 9.333333333333334e-09,
|
| 1984 |
+
"logits/chosen": 0.12221679836511612,
|
| 1985 |
+
"logits/rejected": 0.22767333686351776,
|
| 1986 |
+
"logps/chosen": -282.20001220703125,
|
| 1987 |
+
"logps/rejected": -382.79998779296875,
|
| 1988 |
+
"loss": 0.0008,
|
| 1989 |
+
"nll_loss": 0.8285156488418579,
|
| 1990 |
+
"rewards/accuracies": 1.0,
|
| 1991 |
+
"rewards/chosen": 18.587499618530273,
|
| 1992 |
+
"rewards/margins": 19.037500381469727,
|
| 1993 |
+
"rewards/rejected": -0.44819337129592896,
|
| 1994 |
+
"step": 1230
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.992,
|
| 1998 |
+
"grad_norm": 0.008528899989414147,
|
| 1999 |
+
"learning_rate": 4.888888888888888e-09,
|
| 2000 |
+
"logits/chosen": 0.05415039137005806,
|
| 2001 |
+
"logits/rejected": 0.09858398139476776,
|
| 2002 |
+
"logps/chosen": -258.70001220703125,
|
| 2003 |
+
"logps/rejected": -415.20001220703125,
|
| 2004 |
+
"loss": 0.0009,
|
| 2005 |
+
"nll_loss": 0.880859375,
|
| 2006 |
+
"rewards/accuracies": 1.0,
|
| 2007 |
+
"rewards/chosen": 20.700000762939453,
|
| 2008 |
+
"rewards/margins": 22.612499237060547,
|
| 2009 |
+
"rewards/rejected": -1.9093749523162842,
|
| 2010 |
+
"step": 1240
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 1.0,
|
| 2014 |
+
"grad_norm": 0.012294810094826115,
|
| 2015 |
+
"learning_rate": 4.4444444444444443e-10,
|
| 2016 |
+
"logits/chosen": -0.02900390699505806,
|
| 2017 |
+
"logits/rejected": 0.1044921875,
|
| 2018 |
+
"logps/chosen": -266.1000061035156,
|
| 2019 |
+
"logps/rejected": -404.0,
|
| 2020 |
+
"loss": 0.0011,
|
| 2021 |
+
"nll_loss": 0.961718738079071,
|
| 2022 |
+
"rewards/accuracies": 1.0,
|
| 2023 |
+
"rewards/chosen": 20.8125,
|
| 2024 |
+
"rewards/margins": 21.975000381469727,
|
| 2025 |
+
"rewards/rejected": -1.18310546875,
|
| 2026 |
+
"step": 1250
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 1.0,
|
| 2030 |
+
"eval_logits/chosen": -0.11485877633094788,
|
| 2031 |
+
"eval_logits/rejected": 0.04485614597797394,
|
| 2032 |
+
"eval_logps/chosen": -333.0769348144531,
|
| 2033 |
+
"eval_logps/rejected": -407.5384521484375,
|
| 2034 |
+
"eval_loss": 0.0016917419852688909,
|
| 2035 |
+
"eval_nll_loss": 0.98046875,
|
| 2036 |
+
"eval_rewards/accuracies": 1.0,
|
| 2037 |
+
"eval_rewards/chosen": 18.94230842590332,
|
| 2038 |
+
"eval_rewards/margins": 20.269229888916016,
|
| 2039 |
+
"eval_rewards/rejected": -1.33984375,
|
| 2040 |
+
"eval_runtime": 8.6236,
|
| 2041 |
+
"eval_samples_per_second": 11.596,
|
| 2042 |
+
"eval_steps_per_second": 1.507,
|
| 2043 |
+
"step": 1250
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 1.0,
|
| 2047 |
+
"step": 1250,
|
| 2048 |
+
"total_flos": 0.0,
|
| 2049 |
+
"train_loss": 0.0,
|
| 2050 |
+
"train_runtime": 0.984,
|
| 2051 |
+
"train_samples_per_second": 10161.128,
|
| 2052 |
+
"train_steps_per_second": 1270.268
|
| 2053 |
+
}
|
| 2054 |
+
],
|
| 2055 |
+
"logging_steps": 10,
|
| 2056 |
+
"max_steps": 1250,
|
| 2057 |
+
"num_input_tokens_seen": 0,
|
| 2058 |
+
"num_train_epochs": 1,
|
| 2059 |
+
"save_steps": 500,
|
| 2060 |
+
"stateful_callbacks": {
|
| 2061 |
+
"TrainerControl": {
|
| 2062 |
+
"args": {
|
| 2063 |
+
"should_epoch_stop": false,
|
| 2064 |
+
"should_evaluate": false,
|
| 2065 |
+
"should_log": false,
|
| 2066 |
+
"should_save": true,
|
| 2067 |
+
"should_training_stop": true
|
| 2068 |
+
},
|
| 2069 |
+
"attributes": {}
|
| 2070 |
+
}
|
| 2071 |
+
},
|
| 2072 |
+
"total_flos": 0.0,
|
| 2073 |
+
"train_batch_size": 2,
|
| 2074 |
+
"trial_name": null,
|
| 2075 |
+
"trial_params": null
|
| 2076 |
+
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8401
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8bd212b751b93ee4c9fce82340c043532249ccf965d8401ecae766984808b692
|
| 3 |
size 8401
|