View source on GitHub |
PPO Agents.
Modules
ppo_actor_network
module: Sequential Actor Network for PPO.
ppo_agent
module: A PPO Agent.
ppo_clip_agent
module: A PPO Agent implementing the clipped probability ratios.
ppo_kl_penalty_agent
module: A PPO Agent implementing the KL penalty loss.
ppo_policy
module: An ActorPolicy that also returns policy_info needed for PPO training.
ppo_utils
module: Utils functions for ppo_agent.py.