Forums » Reinforcement Learning »
Added by Egor Popenov about 3 hours ago
Hi Everyone,
To keep our development organized, I have outlined the proposed structure for our GitHub repository. We should follow a modular approach to separate the AI logic from the MIS interface.
Proposed Folder Structure:
/env: Custom Gymnasium environments.
/agents: RL algorithm implementations (PPO/DQN scripts).
/models: Saved pre-trained models (.zip or .pth files).
/notebooks: Jupyter notebooks for training visualization and data analysis.
Key Repository Link:
We will be referencing the CleanRL repository for high-quality, single-file implementations:
https://github.com