This repository contains a PyTorch re-implementation of Offline Reinforcement Learning with Implicit Q-Learning by Ilya Kostrikov, Ashvin Nair, and Sergey Levine.
For the official repository, please use: https:/ikostrikov/implicit_q_learning
This code can be used for offline RL, or for offline RL followed by online finetuning. Negative epochs are offline RL, positive epochs are online (the agent is actively collecting data and adding it to the replay buffer).
If you use this code for your research, please consider citing the paper:
@article{kostrikov2021iql,
title={Offline Reinforcement Learning with Implicit Q-Learning},
author={Ilya Kostrikov and Ashvin Nair and Sergey Levine},
year={2021},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
To run quick versions of these experiments to test if the code matches exactly as the results below, you can run the tests in tests/regression/iql
