README.md

---
license: mit
base_model: ZhangShenao/SELM-Zephyr-7B-iter-2
tags:
- alignment-handbook
- dpo
- trl
- selm
datasets:
- HuggingFaceH4/ultrafeedback_binarized
model-index:
- name: SELM-Zephyr-7B-iter-3
  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. -->

[Self-Exploring Language Models: Active Preference Elicitation for Online Alignment](https://arxiv.org/abs/2405.19332).

# SELM-Zephyr-7B-iter-3

This model is a fine-tuned version of [ZhangShenao/SELM-Zephyr-7B-iter-2](https://huggingface.co/ZhangShenao/SELM-Zephyr-7B-iter-2) using synthetic data based on on the HuggingFaceH4/ultrafeedback_binarized dataset.

## Model description

- Model type: A 7B parameter Zephyr-based Self-Exploring Language Models (SELM).
- License: MIT

## Results

|                                        | AlpacaEval 2.0 (LC WR) | MT-Bench (Average) |
|----------------------------------------|------------------------|--------------------|
| [SELM-Zephyr-7B-iter-3](https://huggingface.co/ZhangShenao/SELM-Zephyr-7B-iter-3) |     &emsp; &emsp; &emsp;&emsp;        24.00         |  &emsp; &emsp; &emsp;         7.48       |
| [SELM-Zephyr-7B-iter-2](https://huggingface.co/ZhangShenao/SELM-Zephyr-7B-iter-2) |    &emsp; &emsp; &emsp;&emsp;         23.40         |  &emsp; &emsp; &emsp;         7.72       |
| [SELM-Zephyr-7B-iter-1](https://huggingface.co/ZhangShenao/SELM-Zephyr-7B-iter-1) |    &emsp; &emsp; &emsp;&emsp;         20.28         |  &emsp; &emsp; &emsp;         7.42       |
| [DPO-Zephyr-7B](https://huggingface.co/ZhangShenao/DPO-Zephyr-7B) |    &emsp; &emsp; &emsp;&emsp;         14.45         |  &emsp; &emsp; &emsp;        7.28        |

Our model also ranks highly on [WildBench](https://huggingface.co/spaces/allenai/WildBench)! 🔥

### Training hyperparameters

The following hyperparameters were used during training:
- alpha: 0.001
- beta: 0.01
- train_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 4
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- num_epochs: 1

### Framework versions

- Transformers 4.40.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.19.1