Swift-huggingface: Complete Swift client for Hugging Face Hub with resumable downloads and Python cache

AI Impact Summary

swift-huggingface is a ground-up Swift client for the Hugging Face Hub that exposes a complete Hub API surface through HubClient, including models, datasets, spaces, collections, and discussions, with resume-capable downloads and progress tracking. It introduces a Python-compatible cache and a TokenProvider-based authentication flow (static tokens, Keychain, OAuth), simplifying credential sources and enabling user-facing apps to authenticate robustly. The package is designed to replace the existing HubApi in swift-transformers, enabling fast, snapshot-based downloads and cross-language cache sharing to reduce re-downloads and download failures for large models.