Did you know that you can use
@elixirlang for ML or building AI apps?
👇 Here are 5 key parts of the "Elixir for AI" initiative:
1. Nx: Standing for "Numerical Elixir". Imagine
#Numpy for Elixir! Multi-dimensional tensors, CPU/GPU compilation, and inspired by Google's JAX.
🔗 Link →
github.com/elixir-nx/nx
2. Livebook: Local-first, open-source notebooks in Elixir. Interactive, collaborative, and fully reproducible. Perfect for Elixir newbies and pros alike! 📈
#InteractiveCoding
🔗 Link →
livebook.dev/ (it also features a growing ecosystem of integrations for databases, messaging, visualization, and more - check them out here:
livebook.dev/integrations/)
3. Explorer: Brings series (one-dimensional) and dataframes (two-dimensional) for fast data exploration to Elixir. Built on top of Rust's Polars library. Think of it as Elixir's answer to dplyr from R. Fast and elegant API. 📊
#DataExploration
🔗 Link →
github.com/elixir-explorer/e…
4. Axon: The neural network library powered by Nx, Axon provides 3 essentials for working with neural nets: Functional API, Model Creation, and Training API. Inspired by PyTorch Ignite.
#NeuralNetworks
🔗 Link →
github.com/elixir-nx/axon (there are also pre-trained models available through Bumblebee
github.com/elixir-nx/bumbleb…, which is built on top of Axon and also can be used to load
@huggingface models!)
5. Scholar: Traditional ML, new approach. Scholar tackles classic machine learning techniques with a modern twist. Fully built on Nx, it's multi-GPU-ready and scalable.
#MachineLearning
🔗 Link →
github.com/elixir-nx/scholar
Are you considering leveraging Elixir and Phoenix for building AI SaaS apps?