Entropy-Regularized-RL

A3C with Proper Entropy Bounses( Also GAE).

Run it by

python A3C.py

soft q learning

The gaussian kernel is from haarnoja. This is for atari, there exist some problems, I will fix them soon.

soft actor critic

After reading paper, I think sac can be almost like a3c...And because of the entropy, it's will not converge faster than a3c in my experiment.

Run it by

python sac.py

sac_new.py is ddpg style. fixed alpha, just for fun ~

Paper

Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor

Equivalence Between Policy Gradients and Soft Q-Learning

Soft Actor-Critic Algorithms and Applications