Models

This section documents the core model implementations used across the ObjectRL library. Each model encapsulates a reinforcement learning agent and its underlying architecture, including actor, critic, loss functions, and update logic.

We divide the models into two main categories:

• Basic Building Blocks: General-purpose modules such as actors, critics, ensembles, and customizable loss functions.

• Algorithm-Specific Models: Full algorithm implementations.

Note

This structure in ObjectRL allows users to flexibly mix and match model components for custom algorithm development.

Attention

If you’re implementing a new algorithm or extending an existing one, refer to the UML diagrams and configuration templates provided in each algorithm’s documentation for structural guidance.