On-device perception‑to‑reasoning pipeline for Apple Silicon
It links on-device vision models with language models, offering a Python toolkit to convert, compress, and run perception, plus a SwiftUI app that executes the full pipeline on Apple hardware. Developers can choose between Apple Foundation Models or Claude for reasoning, enabling private, offline AI workflows. The project targets iOS/macOS AI developers needing an integrated, provider‑agnostic solution, and it surpasses typical demos that only handle perception or chat separately. Its modular design, tests, and documentation make it a practical, ready‑to‑use tool.
View on GitHub →felixyustian/PerceiveReason_Apple_M5