search

Jiwonjeon9603 / MASER

This repository is an implementation of "MASER: Multi-Agent Reinforcement Learning with Subgoals Generated from Experience Replay Buffer" accepted to ICML 2022.
20 stars 7 forks source link

Meaning of block-wise operation in the paper? #6

Closed RZ-Q closed 11 months ago

RZ-Q commented 1 year ago

I am confused about the block-wise operation, why the length is the same as the episode's length, and why to use this block-wise operation.

Jiwonjeon9603 commented 1 year ago

Our algorithm is based on dqn which chooses episodes from experience replay buffer randomly. Therefore, the length which randomly chosen from experience replay buffer are the same length as the episode length. If you see the equations from our paper, you would understand this concept easily.