raju4179/cat-brain-service
🐱 Create a living cat AI that exhibits emotions, reactions, and realistic behavior for an engaging and interactive experience.
What's novel
🐱 Create a living cat AI that exhibits emotions, reactions, and realistic behavior for an engaging and interactive experience.
Code Analysis
18 files read · 4 roundsA FastAPI microservice that uses PPO reinforcement learning (Stable Baselines3) to predict cat actions in a game, layered with an emotion/reaction system, personality drift, and per-cat model fine-tuning, serving animation and sound hints to a Unity client.
Strengths
The layered behavioral architecture (RL prediction → emotion engine → reaction system → stochastic noise → memory) is a thoughtful design for game AI, and the per-cat model management with fine-tuning is a practical pattern. The code is well-organized with clear separation between core logic, inference, services, and API layers.
Weaknesses
The tests are completely out of sync with the implementation (expecting 4-dim observations and 4 actions vs. the actual 11-dim and 8 actions), meaning they would all fail. The README is entirely misleading, describing a desktop application with .exe/.dmg installers when this is a Python web service, and the batch prediction path in BatchPredictor is partially dead code since predict_single bypasses the queue entirely.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
41
Contributors
2
Files
187
Active weeks
4
Repository
Language
Python
Stars
1
Forks
0
License
—