Pytorch implementation of Soft Actor-Critic Algorithm
Soft actor-critic (SAC) is an off-policy actor-critic deep RL algorithm that optimizes stochastic continuous policies defined in the maximum entropy framework.
Paper: Haarnoja, T., Zhou, A., Abbeel, P., Levine, S. (2018), Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
- Clone the repository
git clone https://github.com/domingoesteban/pytorch_sac
- Install required python packages
cd pytorch_sac
pip install -r requirements.txt
- Run (train) the algorithm with a continuous observation-action space OpenAI-gym environment (E.g. Pendulum-v0)
python train.py -e Pendulum-v0 -i 30
- Plot the results from the previous training process
python eval.py -p PATH_TO_LOG_DIRECTORY
- Evaluate the resulted policy
python eval.py PATH_TO_LOG_DIRECTORY