Xet has arrived on the Hugging Face Hub
Summary
The Hub activates Xet as the storage layer for model and data set artefacts. Users keep Git-like workflows, while transfer and deduplication happen at block level. Xet is gradually replacing Git LFS as the storage path for large Hub files.
Ideas
- Infrastructure changes are accepted more easily when existing workflows are preserved.
- Shared blocks reduce redundant copies of similar model revisions.
Insights
- ML storage has to move large, similar artefacts efficiently without giving up reproducibility.
Facts
- The hf_xet extension speeds up uploads and downloads for compatible clients.
Critique
- Performance and availability now also depend on a more complex storage layer close to being proprietary.
Recommendations
- Document client requirements and keep a tested fallback ready for automated builds.
References
Read the original article on Hugging Face
Links to the original source and the Web Archive open in a new tab.