# Simple Preference Optimization (SimPO)

This repository contains the code and released models for our paper [SimPO: Simple Preference Optimization with a Reference-Free Reward](https://arxiv.org/abs/2405.14734). We propose a simpler and more effective preference optimization algorithm than DPO (Direct Preference Optimization) without using a reference model. SimPO outperforms DPO and its latest variants across AlpacaEval 2, MT-Bench, and Arena-Hard benchmarks under various settings.

<img src="./SimPO.png" width="1000px"></img>

## 🔗 Quick Links
- [SimPO: Simple Preference Optimization with a Reference-Free Reward](#simple-preference-optimization-simpo)
  - [Released Models](#released-models)
  - [Install Requirements](#install-requirements)
  - [Training scripts](#training-scripts)
  - [Evaluation](#evaluation)
  - [Bugs or Questions?](#bugs-or-questions)
  - [Citation](#citation)

## Released Models


## Install Requirements

Our codebase is built upon the [alignment-handbook repo](https://github.com/huggingface/alignment-handbook). The following steps will guide you through the installation process.

First, create a Python virtual environment using e.g. Conda:
```shell
conda create -n handbook python=3.10 && conda activate handbook
```

Next, install PyTorch `v2.2.2`. Since this is hardware-dependent, we
direct you to the [PyTorch Installation Page](https://pytorch.org/get-started/locally/).

You can then install the remaining package dependencies of [alignment-handbook](https://github.com/huggingface/alignment-handbook) as follows:

```shell
git clone https://github.com/huggingface/alignment-handbook.git
cd ./alignment-handbook/
python -m pip install .
```

You will also need Flash Attention 2 installed, which can be done by running:

```shell
python -m pip install flash-attn --no-build-isolation
```

## Training Scripts

We provide four training config files for the four training setups reported in our paper:

* Mistral-Base:
```shell
ACCELERATE_LOG_LEVEL=info accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml scripts/run_simpo.py training_configs/mistral-7b-base-simpo.yaml
```
* Mistral-Instruct:
```shell
ACCELERATE_LOG_LEVEL=info accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml scripts/run_simpo.py training_configs/mistral-7b-instruct-simpo.yaml
```
* Llama3-Base:
```shell
ACCELERATE_LOG_LEVEL=info accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml scripts/run_simpo.py training_configs/llama-3-8b-base-simpo.yaml
```
* Llama3-Instruct:
```shell
ACCELERATE_LOG_LEVEL=info accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml scripts/run_simpo.py training_configs/llama-3-8b-instruct-simpo.yaml
```

## Evaluation

We follow the official implementation for evaluation on AlpacaEval 2, Arena-Hard, and MT-Bench, as follows:

* AlpacaEval 2: Please refer to the [AlpacaEval repo](https://github.com/tatsu-lab/alpaca_eval) for evaluation.

* Arena-Hard: Please refer to to the [Arena-Hard-Auto repo](https://github.com/lm-sys/arena-hard-auto) for evaluation.

* MT-Bench: Please refer to the [FastChat repo](https://github.com/lm-sys/FastChat) for evaluation.

## Bugs or Questions?
If you have any questions related to the code or the paper, feel free to email Yu (yumeng5@virginia.edu). If you encounter any problems when using the code, or want to report a bug, feel free to open an issue! Please try to specify the problem with details so we can help you better and quicker!

## Citation
Please cite our paper if you find the repo helpful in your work:

```bibtex
@article{meng2024simpo,
  title={{SimPO}: Simple Preference Optimization with a Reference-Free Reward},
  author={Meng, Yu and Xia, Mengzhou and Chen, Danqi},
  year={2024}
}
```