Skip to content
/ d2rl Public

Code for the paper "D2RL: Deep Dense Architectures for Reinforcement Learning"

License

Notifications You must be signed in to change notification settings

pairlab/d2rl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

24 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

D2RL

Official PyTorch code for D2RL: Deep Dense Architectures in Reinforcement Learning. Details on an independently reproduced TF2 implementation is listed below.

Paper: http://arxiv.org/abs/2010.09163

Blog: https://sites.google.com/view/d2rl/home

The code includes the code to train SAC-D2RL, TD3-D2RL, and CURL-D2RL agents.

If there are any issues or questions related to the code, send an email at: [email protected]

To try D2RL on other environments, the main parameters to be tuned are the learning rate of the actors and critics. To try D2RL with other algorithms, we also include a pseudo-code for the architecture changes in the main paper. Kindly let us know if there are any questions.

Installation details, dependencies, and instructions for training are included in the individual sub-folders.

TF2 version of the D2RL which was independently reproduced:

https://github.com/keiohta/tf2rl

More details on D2RL and how to run the code can be found in

Acknowledgement

The codebase is built upon other these previous codebases:

SAC: https://github.com/denisyarats/pytorch_sac

TD3: https://github.com/sfujim/TD3

CURL: https://github.com/MishaLaskin/curl

About

Code for the paper "D2RL: Deep Dense Architectures for Reinforcement Learning"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages