PyTorch implementation of Advantage Actor Critic (A2C), Proximal Policy Optimization (PPO), Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation (ACKTR) and Generative Adversarial Imitation Learning (GAIL).
Added a command line argument --disable-fork-safety to disable fork safety which sometimes generates objc errors in macOs High Sierra and above 10.13+.
Added a command line argument
--disable-fork-safety
to disable fork safety which sometimes generates objc errors in macOs High Sierra and above 10.13+.