rarilurelo / pytorch_a3c

38 stars 4 forks source link

Asynchronous Advantage Actor-Critic in PyTorch

This is PyTorch implementation of A3C as described in Asynchronous Methods for Deep Reinforcement Learning.

Since PyTorch has a easy method to control shared memory within multiprocess, we can easily implement asynchronous method like A3C.

Requirement

Usage

training

python run_a3c.py --atari

In default settings, num_process is 8. Set it as python run_a3c --num_process 4 to fit your number of cpu's cores.

test

After training

python test_a3c.py --render --monitor