Skip to content

PyTorch implementation of Deterministic Generative Adversarial Imitation Learning (GAIL) for Off Policy learning

License

Notifications You must be signed in to change notification settings

nikhilbarhate99/Deterministic-GAIL-PyTorch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

50 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Deterministic-GAIL-PyTorch

This is an attempt to implement Generative Adversarial Imitation Learning (GAIL) for deterministic policies with off Policy learning on static data. The policy never interacts with the environment (except for evaluation), instead it is trained on policy state-action pair, where policy only selects actions for states sampled from expert data.

Results

Although it works sometimes (depending on the type of environment), the algorithm has high variance, and the results are inconsistent.

BipedalWalker-v2

Expert Policy Recovered Policy (10 expert episodes)
Epochs vs rewards

Releases

No releases published

Packages

No packages published

Languages