On a Connection Between Imitation Learning and RLHF (ICLR 2025)

This repository contains the code for our ICLR 2025 paper On a Connection Between Imitation Learning and RLHF.

Environment

First, install PyTorch 2.1.2 from the PyTorch Installation Page.

Create a Python virtual environment using e.g. Conda:

conda create -n DIL python=3.10 && conda activate DIL

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

Training Scripts

• Mistral-Base:

bash Mistral-7B-Base.sh

• Llama3-Base:

bash Llama-8B-Base.sh

Evaluation Benchmarks

• Open LLM Leadboard v1

• Open LLM Leadboard v2

Reference

If you find our repo to be useful, please cite our paper:

@inproceedings{DIL2025,
  title={On a Connection Between Imitation Learning and RLHF},
  author={Xiao, Teng; Yuan, Yige; Li, Mingxiao; Chen, Zhengyu; Honavar, Vasant G.},
  booktitle={International Conference on Learning Representations (ICLR)},
  year={2025}
}