Skip to content

Welcome to Pulp Hugging Face!

The pulp_hugging_face plugin extends pulpcore to support hosting and caching Hugging Face Hub content.

If you just got here, you should take our Getting Started with Hugging Face tutorial to set up your first pull-through cache for Hugging Face content. We also recommend that you read the Basic Concepts section before diving into the workflows and reference material.

Features

  • Pull-through caching:
    • Automatically fetch and cache content from Hugging Face Hub on first access
    • Support for on-demand content fetching to reduce disk space
    • Cache models, datasets, and spaces from Hugging Face Hub
  • Authentication support:
    • Use Hugging Face tokens for accessing private repositories
    • Seamless integration with HF Hub authentication
  • Full Hugging Face Hub compatibility:
    • Support for all Hugging Face URL patterns
    • API proxying for metadata operations
    • Compatible with huggingface-cli and transformers library
  • Content Types:
    • Models: PyTorch, TensorFlow, ONNX, and other model formats
    • Datasets: Training data, evaluation data, data descriptions
    • Spaces: Gradio apps, Streamlit apps, static sites
  • Versioned Repositories so every operation is a restorable snapshot
  • De-duplication of all saved content
  • Host content either locally or on S3