Hugging Face introduced Storage Buckets, non-versioned, S3-like object storage on the Hub designed for intermediate ML artifacts like checkpoints and logs. Backed by Xet, the chunk-based storage backend deduplicates content across files, reducing bandwidth and storage costs. A pre-warming feature partners with AWS and GCP to bring data closer to compute regions. Users can manage Buckets via the hf CLI or the huggingface_hub Python library.
No score is assigned. Sources and their independence are shown in the citation chain below.