Offline-Reinforcement-Learning-Algorithms

Contains PyTorch Implementation of the following algorithms

• Off-Policy Deep Reinforcement Learning without Exploration (BCQ) : https://arxiv.org/abs/1812.02900
• Offline Reinforcement Learning with Implicit Q-Learning (IQL) : https://arxiv.org/abs/2110.06169
• A Minimalist Approach to Offline Reinforcement Learning (TD3_BC) : https://arxiv.org/pdf/2106.06860.pdf

Train agent

python main.py --algo_name='IQL' --env_name='halfcheetah-expert-v2'