Baselines

These tutorials show how to configure and compare MASA baselines in small, readable experiments.

For a verified list of the algorithms currently present in the codebase, start with Algorithms Overview.

• Tabular Safe RL Baselines compares q_learning, q_learning_lambda, lcrl, sem, and recreg on one constrained tabular environment.

• Continuous Safe RL Baselines is a stub for PPO, CPO, and PPO Lagrangian continuous-control baselines.

Runnable notebooks:

• notebooks/tutorials/06_tabular_safe_rl_baselines.ipynb

• notebooks/tutorials/07_continuous_safe_rl_baselines.ipynb