Frenchwizz/FNAF-Night1-PPO

PPO Agent Playing Five Nights at Freddy's (Night 1)

This is a trained PPO (CnnPolicy) agent playing a custom web-based Five Nights at Freddy's Gymnasium environment built with Playwright and OpenCV.

Video Replays

Model Details

• Algorithm: PPO (Proximal Policy Optimization)
• Policy: CnnPolicy (84x84x3 RGB visual frames)
• Action Space: Discrete(9) (Wait/Conserve Power, Left Door, Left Light, Right Door, Right Light, Camera Flip, CAM 2B, CAM 4B, CAM 1A)
• Evaluation Result: Survived Night 1 to 6:00 AM (Total Reward: 174.0, 741 steps)