stable-baselines3

To train reinforcement learning agents with Stable Baselines3, this skill automates workflow construction for setting up environments, selecting algorithms like PPO or SAC, and implementing vectorized training and callbacks.

Yes, you can use custom Gymnasium environments for SB3 agent training. The skill provides templates to create, integrate, and run vectorized training setups specifically tailored for your custom Gymnasium environments.

The best way to evaluate and compare SB3 agents is by using the skill's integrated evaluation templates. These automate the assessment process, allowing you to monitor performance and compare different algorithms like DQN, TD3, and A2C.

Yes, the reinforcement learning workflow supports model persistence and callbacks. It provides ready-made templates to customize monitoring and control via SB3 callbacks, alongside saving and loading trained models.

Available algorithms for RL agent training in this setup include PPO, SAC, DQN, TD3, and A2C. The skill allows you to easily select and apply these algorithms across various Gymnasium environments.