Deploying Livebook Notebooks as Apps to Hugging Face Spaces

Hugging Face has integrated with Livebook to allow users to deploy interactive Elixir notebooks as functional applications on Hugging Face Spaces. This integration enables the Elixir community to build, experiment with, and share machine learning (ML) applications—such as audio-to-text chat apps powered by Whisper—within minutes.

Livebook Application Deployment on Hugging Face Spaces

Users can now deploy Livebook notebooks as apps to Hugging Face Spaces, transforming interactive code environments into shareable applications. A primary example of this capability is a Whisper-based chat app where audio messages are automatically converted to text using the Whisper ML model.

This deployment workflow highlights several key technical capabilities of the Elixir ML ecosystem:

  • Hugging Face Model Integration: Direct use of models hosted on the Hugging Face platform.
  • Multiplayer ML Apps: The ability to create collaborative, multi-user machine learning applications.
  • Concurrent Model Serving: The capacity to serve ML models concurrently, with the option to distribute model serving across a cluster.

The Elixir and Hugging Face Ecosystem

The Elixir machine learning landscape relies heavily on Hugging Face's open-source projects and platform infrastructure to provide numerical computing and data science tools.

Bumblebee

Bumblebee is a library that brings pre-trained neural network models from Hugging Face to the Elixir community. Inspired by Hugging Face Transformers, Bumblebee utilizes the Hugging Face Hub to download model parameters.

Tokenizers

The tokenizers library serves as an Elixir binding for the Hugging Face Tokenizers library, providing essential text processing capabilities for ML models.

Livebook Integration

Livebook is an open-source tool for writing interactive code notebooks in Elixir. Beyond deploying apps, Livebook can be run directly within Hugging Face Spaces using a Space Docker template, allowing users to run and experiment with notebooks for free.

Getting Started with Livebook on Hugging Face

To begin building and deploying ML apps, users can install Livebook on Hugging Face Spaces via the provided Docker templates and follow the official video tutorials to transition from a notebook to a deployed application.

Sources