onlytailei / A3C-PyTorch

PyTorch implementation of Advantage async actor-critic Algorithms (A3C) in PyTorch

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Have you benchmarked this?

ethancaballero opened this issue · comments

How well does it do on Breakout, etc.?

I'm still testing it. There will be a result soon.

Train it on Breakout for a day. Pong's too easy.

one quick question: what's the different btn this repo and ikostrikov's

Have to say that now it is very similar.
At the beginning, I want to calculate the loss of former steps as a batch. But somehow, it does not work.