jongcheon-park / td3-gail-pytorch

TD3-GAIL pytorch

PyTorch implementation of TD3-GAIL tested on the following environments:

• Bipedal Walker v2

• Lunar Lander Continuous v2

• Walker2d v1 (Roboschool)

• HalfCheetah v1 (Roboschool)

TD3-GAIL is the recovered reward function based reinforcement learning that fusion of TD3 and GAIL. TD3-GAIL shows good learning performance compared with GAIL and DDPG-GAIL.

Usage

• To train the TD3-GAIL, run the 'train.py'
• To test the trained TD3-GAIL, run the 'test.py'

Requirements

Python 3.6
PyTorch 1.5.0
NumPy 1.19.5
gym 0.15.4
Roboschool 1.0.34
Pillow 7.0.0

References

[1] [Official TD3 Code] (https://github.com/sfujim/TD3)
[2] [DGAIL] (https://github.com/nikhilbarhate99/Deterministic-GAIL-PyTorch)