A curated list of Machine Learning (ML) and Generative AI (GenAI) packages and resources for the Elixir programming language. It targets Elixir developers exploring AI/ML and AI practitioners new to Elixir, showcasing tools that leverage Elixir's concurrency for advanced AI workloads.

How It Works

The Elixir ML/GenAI ecosystem is built on foundational libraries like Nx (tensor computations, CPU/GPU compilation), Explorer (dataframes), and Livebook (interactive notebooks). Libraries extend these for traditional ML (Scholar), deep learning (Axon, Bumblebee for pre-trained models), and computer vision (YOLO). A strong focus is on Generative AI, with numerous tools for LLM interaction, Retrieval Augmented Generation (RAG), and agent frameworks. Pythonx is a key enabler, embedding Python interpreters via NIFs for seamless integration with existing Python ML libraries.

Quick Start & Requirements

This is a curated list, not a single project. Users require an Elixir environment. Specific libraries may necessitate Python installations (for Pythonx) and GPU hardware with CUDA drivers for optimal performance with Nx and deep learning models. Links to individual library documentation are available within the repository.

Highlighted Details

• Nx: Efficient tensor operations with CPU/GPU compilation.
• Pythonx: Direct embedding of Python interpreters via NIFs for seamless Python ML library calls.
• Axon & Bumblebee: Native Elixir neural networks and pre-trained models (Hugging Face integration).
• YOLO: Real-time object detection claiming ~38ms processing.
• GenAI & Agents: Extensive libraries for LLM interaction, RAG, and agent orchestration.
• Model Context Protocol (MCP): Enables semantic search of Elixir documentation for AI assistants.

Maintenance & Community

Contributions are welcomed, with guidelines provided. Specific details on active maintainers, community channels (Discord/Slack), or a public roadmap are not present in the README.

Licensing & Compatibility

Content is licensed under the CC0 License (public domain dedication). This permits unrestricted use, sharing, and adaptation, including for commercial applications, without attribution or copyleft restrictions.

Limitations & Caveats

As a curated list, it reflects the ecosystem's state. GPU-accelerated performance depends on user hardware and configuration. Python integration via Pythonx may add environment management complexity. Some listed tools might be in early development.