---
library_name: transformers
base_model: W-61/llama-3-8b-base-sft-ultrachat-8xh200
tags:
- alignment-handbook
- epsilon-dpo
- generated_from_trainer
datasets:
- HuggingFaceH4/ultrafeedback_binarized
model-index:
- name: llama-3-8b-base-epsilon-dpo-ultrafeedback-8xh200-20260411-020915
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# llama-3-8b-base-epsilon-dpo-ultrafeedback-8xh200-20260411-020915

This model is a fine-tuned version of [W-61/llama-3-8b-base-sft-ultrachat-8xh200](https://huggingface.co/W-61/llama-3-8b-base-sft-ultrachat-8xh200) on the HuggingFaceH4/ultrafeedback_binarized dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6085
- Rewards/chosen: -0.6393
- Rewards/rejected: -0.8881
- Rewards/accuracies: 0.6905
- Rewards/margins: 0.2488
- Logps/chosen: -567.7599
- Logps/rejected: -657.1562
- Logps/ref Chosen: -287.9388
- Logps/ref Rejected: -266.7935
- Logits/chosen: -0.8106
- Logits/rejected: -0.7709
- Kl/p Epsilon Steps: 0.6734
- Kl/n Epsilon Steps: 0.3185

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 5e-07
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- total_eval_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/chosen | Logps/rejected | Logps/ref Chosen | Logps/ref Rejected | Logits/chosen | Logits/rejected | Kl/p Epsilon Steps | Kl/n Epsilon Steps |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:------------:|:--------------:|:----------------:|:------------------:|:-------------:|:---------------:|:------------------:|:------------------:|
| 2.3277        | 0.4188 | 200  | 0.5904          | -0.6331        | -0.9468          | 0.7011             | 0.3137          | -411.3474    | -452.2706      | -287.9388        | -266.7935          | -0.8135       | -0.7841         | 0.6885             | 0.3044             |
| 2.4805        | 0.8377 | 400  | 0.6085          | -0.6393        | -0.8881          | 0.6905             | 0.2488          | -567.7599    | -657.1562      | -287.9388        | -266.7935          | -0.8106       | -0.7709         | 0.6734             | 0.3185             |


### Framework versions

- Transformers 4.51.0
- Pytorch 2.3.1+cu121
- Datasets 2.21.0
- Tokenizers 0.21.4