Matth21

@Matth21 · User

GitHub profile ↗ · Compare

0 followers5 repositories

Repositories

Matth21/atlas

MLX-native LLM quantizer for Apple Silicon — measured per-block sensitivity, joint (bit-width, group-size) Pareto allocation, fully on-device

★ 4PythonForks 0

Matth21/second-brain

Pre-registered falsifiable gates for agent memory: a frozen agent reaches oracle ceiling from experience alone. Kills included.

★ 0PythonForks 0