GStars
    backblaze-labs

    backblaze-labs/hf-cache-sync

    Shared Hugging Face cache for teams. Sync local HF cache to Backblaze B2 or S3, eliminate repeated downloads, and enable fast, consistent model access across machines and CI.

    cli
    nlp
    machine-learning
    ai-infrastructure
    b2-labs
    cloudstorage
    hugging-face
    huggingface
    Python
    MIT
    4 stars
    2 forks
    4 GitHub watchers
    Updated Nov 5, 2026
    View on GitHub

    Build with Backblaze B2

    SDKs, agent skills, IDE extensions, and reference pipelines from Backblaze Labs. All open source.

    Explore Backblaze Labs

    Loading star history...

    Where it stands

    Rank, growth pace and release cadence relative to everything GStars tracks.

    The project lists 2 contributors on GitHub.

    Use Cases & Benefits

    • Sync local Hugging Face model caches to Backblaze B2 or any S3-compatible storage for shared access across machines and CI environments.
    • Eliminates redundant downloads by enabling a two-tier cache with remote storage, reducing bandwidth and disk usage for ML teams and pipelines.
    • Use for sharing large Hugging Face model weights efficiently across multiple engineers' workstations in a team environment.
    • Use for speeding up CI/CD pipelines by caching and reusing Hugging Face models instead of downloading them on every run.
    • Use for managing GPU cluster nodes by preloading and synchronizing model caches to ensure consistent and fast access during training or inference.

    Should you depend on it?

    License terms, project hygiene and risks, from GitHub metadata and GStars' snapshots.

    License

    MIT

    Permissive: commercial use, modification and redistribution are allowed with attribution.

    Community files

    • README: present
    • Contributing guide: present

    Watch-outs

    • No GitHub releases; pin a commit or tag instead of a release.

    General guidance from repository metadata, not legal advice. How we score

    From the README

    Sync your Hugging Face model cache (~/.cache/huggingface/hub) to Backblaze B2 or any S3-compatible object storage, share it across machines and CI runners, and cap local disk usage with LRU eviction.

    hf-cache-sync is a CLI that turns the local Hugging Face Hub cache into a two-tier cache: local disk plus a shared remote bucket on Backblaze B2 (or any S3-compatible store such as AWS S3 or Cloudflare R2). One machine downloads a model from the Hugging Face Hub, pushes the content-addressed blobs and revision manifests to the bucket, and every other machine — workstation, GPU cluster node, or CI runner — hydrates from the bucket instead of re-downloading from the Hub.

    Built for ML teams and CI/CD pipelines that waste bandwidth and disk re-fetching the same 5–30 GB transformer, diffusion, or LLM weights on every machine.

    Hugging Face models and datasets are cached locally. Disks fill up fast, teams re-download identical artifacts on every machine, CI runners fetch the same weights every run, and Windows symlink limitations cause blob duplication. There is no native way to back the cache with shared object storage.

    Excerpt from the backblaze-labs/hf-cache-sync README on GitHub (MIT license). © its authors.Read the full README on GitHub

    Daily snapshots

    Stars gained per day, and how forks and open issues moved, from GStars' daily snapshots

    Jun 25: +0Jun 26: +0Jun 27: +0Jun 28: +0Jun 29: +0Jun 30: +0Jul 1: +0Jul 2: +0Jul 3: +0Jul 4: +0Jul 5: +0Jul 6: +0Jul 7: +0Jul 8: +0Jul 9: +0Jul 10: +0Jul 11: +0Jul 12: +0Jul 13: +0Jul 14: +0Jul 15: +0Jul 16: +0Jul 17: +0Jul 18: +0Jul 19: +0Jul 20: +0Jul 21: +0Jul 22: +0Jul 23: +0Jul 24: +0Jul 25: +0Jul 26: +0Jul 27: +0Jul 28: +0Jul 29: +0Jul 30: +0Jul 31: +0Aug 1: +0Aug 2: +0Aug 3: +0Aug 4: +0Aug 5: +0Aug 6: +0Aug 7: +0Aug 8: +0Aug 9: +0Aug 10: +0Aug 11: +0Aug 12: +0Aug 13: +0Aug 14: +0Aug 15: +0Aug 16: +0Aug 17: +0Aug 18: +0Aug 19: +0Aug 20: +0Aug 27: +0Aug 28: +0Aug 29: +0Aug 30: +0Aug 31: +0Sep 1: +0Sep 2: +0Sep 3: +0Sep 4: +0Sep 5: +0Sep 6: +0Sep 7: +0Sep 8: +0Sep 9: +0Sep 10: +0Sep 11: +0Sep 12: +0Sep 13: +0Sep 14: +0Sep 15: +0Sep 16: +0Sep 17: +0Sep 18: +0Sep 19: +0Sep 20: +0Sep 21: +0Sep 22: +0

    +0 stars over 90 days across 86 snapshot days.

    Forks

    2+1

    Open issues

    0+0

    Compared with peers in CLI

    Fastest-growing repositories GStars tracks in this cohort for the week of Nov 1, by 7-day star growth

    View category →
    backblaze-labs/hf-cache-sync compared with CLI repositories by stars, 7-day star growth, health score and last release
    RepositoryStars+7dHealthLast release
    vercel-labs/agent-desktop17.0K+13.3K33today
    mattpocock/skills315.4K+8.7K357 days ago
    stablyai/orca116.8K+6.6K40today
    obra/superpowers313.1K+4.5K532 days ago
    Panniantong/Agent-Reach109.3K+3.0K604 days ago
    backblaze-labs/hf-cache-sync4—11—
    Category median116.8K+6.6K40—
    Compare hf-cache-sync with agent-desktop, skills, orca →

    Also by backblaze-labs

    Other repositories from this maintainer that GStars tracks

    Discover Repositories

    Search across tracked repositories by name or description