This repository contains the official implementation of Online Pre-Training for Offline-to-Online RL (OPT).
OPT is implemented in Python 3.9 and built using PyTorch. To install the required dependencies, run:
pip install -r requriements.txt
The core implementation of OPT can be found in algorithms/opt.py, and the IQL + OPT variant is available in algorithms/iql_opt.py Example commands:
python algorithms/opt.py --config_path=configs/opt/walker2d/random_v2.yaml
python algorithms/opt.py --config_path=configs/opt/antmaze/umaze_v2.yaml
This codebase builds upon the implementation from CORL.
If you have any questions or inquiries, please feel free to contact us at:
yongjae.shin@kaist.ac.kr, kanghoon.lee@lgresearch.ai