This is an implementation of Soft Actor Crtitic (The official implementation can be found here) with Tensorflow 2.1. It is partially based off OpenAI's spinningup Soft Actor Critic.
The easiest way to setup is with Docker after cloning the repo.
docker image build -t sac:1.0 -f docker/Dockerfile.sac .
docker container run --detach -it --name sac sac:1.0
docker attach sac
python sac.py --env Pendulum-v0 --hid 256 --l 2 --gamma 0.99 --epochs 50 --exp_name magic_sac
Will run Soft Actor Critic on Pendulum-v0 with 2 hidden layers each with 256 units for 50 epochs. Logging will be stored under name "data/magic_sac"