Description

A high-performance C and Metal implementation designed to run MiniMax H3 video models locally on Apple Silicon. It introduces distilled 5-6 step sampling via 'folded Turbo LoRA' techniques, cutting video generation times nearly in half on M5 Max hardware compared to standard implementations.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.