PPO Agent playing MineSweeper Mania

This is a trained PPO agent playing a custom web-based Minesweeper environment built using Playwright and Gymnasium.

Video Replay

Model Details

• Algorithm: PPO (Proximal Policy Optimization)
• Policy: MlpPolicy (9x9 numerical grid state)
• Observation Space: 9x9 Box (-1 = covered, 0-8 = revealed counts)
• Action Space: Discrete(81)