search

jcwleo / awr-pytorch

Advantage-Weighted Regression
MIT License
10 stars 2 forks source link

Continous branch #4

Closed MoMe36 closed 4 years ago

MoMe36 commented 4 years ago

Hi ! I'd like to add support for continuous environments so I used your implementation as a base. I added various things such as :

Despite my best efforts, my agent fails in the Pendulum environment. I thought maybe we could work together to solve this. That'd be really neat !

Let me know what you think ! Thanks ! PS: This is my first PR, hope I didn't do too many things wrong

jcwleo commented 4 years ago

@MoMe36 Thank you for PR. I will review this PR soon. :)

MoMe36 commented 4 years ago

Actually, it is working on Pendulum-v0, it just takes quite a lot of time. ((:

jcwleo commented 4 years ago

LGTM