OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Bumblebee

Bumblebee is an open-source Elixir library built on Axon for downloading supported pretrained neural-network models from Hugging Face and running machine-learning tasks. It provides task-oriented pipelines that can be used with Nx Serving.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Bumblebee — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Bumblebee when you want to run a supported pretrained model from Elixir and are prepared to check model architecture and weight-format compatibility.

Who it is for

  • Elixir developers adding pretrained-model inference to an application.
  • Teams exploring model tasks in Livebook before adapting the code for an application.
  • Phoenix and LiveView developers looking for example model integrations.

Documented uses

  • Load a supported Hugging Face model and its tokenizer.
  • Run a task pipeline, such as filling a masked word in text.
  • Build an Axon-based workflow around pretrained models in an Elixir project.
Before you choose

Requirements & limits

The project is open source under the Apache-2.0 license.

  • The model architecture must be implemented in Bumblebee; a Hugging Face model repository alone does not ensure compatibility.
  • The README identifies PyTorch and Safetensors weights as supported and Flax and TensorFlow weight formats as unsupported.
  • The documented EXLA path uses Nx's EXLA backend for just-in-time compilation and CPU or GPU execution; GPU use requires XLA_TARGET configuration.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up Bumblebee with EXLA

Add Bumblebee and configure Nx to use EXLA as its default backend.

Before you start

  • An Elixir project using Mix, or a notebook or script that can use Mix.install/2.
  • Add Bumblebee version ~> 0.6.0 and EXLA as dependencies.
  • For GPU use, set XLA_TARGET according to the EXLA/XLA usage instructions.

Steps

  1. Add Bumblebee and EXLA to the dependencies in mix.exs.

  2. Configure Nx with EXLA.Backend as its default backend in config/config.exs.

  3. For notebooks or scripts, use the documented Mix.install/2 approach to install and configure both dependencies.

Expected outcome

The project has the documented Bumblebee and EXLA dependencies, with Nx configured to use EXLA.

Guide limitations

  • This covers dependency and backend setup, not loading or running a model.
  • GPU use requires the separately documented XLA_TARGET configuration.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
12 October 2026
Original website title
elixir-nx/bumblebee · Official homepage title · retained in its source language
How we review sourcesSuggest a correction