rainbow icon indicating copy to clipboard operation
rainbow copied to clipboard

Add PER

Open Ashutosh-Adhikari opened this issue 8 years ago • 3 comments

Hi,

The reference code for PER additions : https://github.com/Kaixhin/Rainbow.git.

It is segment tree based implementation of PER.

Ashutosh-Adhikari avatar Apr 23 '18 07:04 Ashutosh-Adhikari

Quick questions. Prioritized experience replay is a just sampling method. It should only affect the way we sample from replay buffer. Why changing the loss function to be a weighted one? Have you tested the performance?

hengyuan-hu avatar Apr 23 '18 08:04 hengyuan-hu

Please have a look at the Line 13 of the algorithm described in PER paper. I had only checked for Breakout. Gave a slight performance improvement over DQN. Please have a look at Kaixhin/Rainbow#15 and let me know.

Ashutosh-Adhikari avatar Apr 23 '18 09:04 Ashutosh-Adhikari

Hi, Did you get time to check the code through tests? :)

Ashutosh-Adhikari avatar May 28 '18 19:05 Ashutosh-Adhikari