Skip to content

a7b23/Hindsight-Experience-Replay

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Hindsight-Experience-Replay

This is an implementation of the bit-flipping experiment mentioned in paper Hindsight Experience Replay arXiv preprint arXiv:1707.01495 (2017).

To start training, simply run :-

python her_main.py

The goal state can be chosen by 3 different strategies(final, future, episode) as mentioned in the paper. To run with any strategy simply change the Her flag and run the code. For example, for the strategy 'final' simply run :-

python her_main.py --Her final

To get the results without using HER algorithm run :-

python her_main.py --Her None --bit_size 7

Below is the plot showing the comparison between HER+DQN and DQN for a bit size of 7.

However, for a bit size > 10 only using the DQN algorithm does not lead to any success.

Below is the plot comparing all the 3 different goal strategies as well as just DQN for a bit size of 30.

Acknowledgment

Code borrowed from hindsight-experience-replay

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages