protomotions.agents.ppo package#
Proximal Policy Optimization (PPO) implementation.
This package implements the PPO algorithm for reinforcement learning, including the actor-critic model architecture and training logic.
- Key Components:
PPO: Main PPO agent
PPOModel: Actor-critic model
PPOActor: Policy network
PPOAgentConfig: Configuration
Example
>>> from protomotions.agents.ppo.agent import PPO
>>> agent = PPO(fabric, env, config)
>>> agent.train()
Submodules#
- protomotions.agents.ppo.agent module
PPOconfigmotion_index__init__()has_criticcreate_model()add_agent_info_to_obs()create_optimizers()actor_modulecritic_moduleload_parameters()get_state_dict()save()register_algorithm_experience_buffer_keys()record_rollout_step()perform_optimization_step()actor_step()calculate_extra_actor_loss()critic_step()optimize_model()
- protomotions.agents.ppo.config module
- protomotions.agents.ppo.model module
- protomotions.agents.ppo.utils package