conda env create -f environment.yml
conda activate cse579a1
python main.py --task pg/actor_critic/sac --env pendulum/ant
(Aliases: pg = policy_gradient, ac = actor_critic.)
Examples:
python main.py --task pg --env pendulum
python main.py --task actor_critic --env pendulum
python main.py --task sac --env pendulum
Append --test to evaluate a saved checkpoint instead of training.
More details in the assignment spec.
main.py(hyperparameter tuning only)policy_gradient.py— policy gradient TODOsactor_critic.py— actor-critic TODOssac.py— SAC TODOs