

Elixir meets machine learning
May 26, 2021
José Valim, the creator of Elixir and developer of projects like Phoenix and Nx, joins the discussion on integrating Elixir into machine learning. He shares insights on Numerical Elixir and the launch of NX, designed for enhanced machine learning capabilities. The conversation explores the advantages of Axon for building neural networks, the benefits of a collaborative interactive notebook with Livebook, and the vibrant community driving these innovations. Valim also discusses the exciting possibilities for AI development outside the Python ecosystem.
AI Snips
Chapters
Books
Transcript
Episode notes
NX and Machine Learning in Elixir
- José Valim's newest project, Numerical Elixir (NX), brings Elixir into machine learning.
- The project involves a layered approach, collaborations, and a collaborative notebook built on Phoenix LiveView.
NX Core Components
- NX provides multidimensional tensors and numerical definitions for efficient machine learning.
- Bindings for Google XLA and PyTorch (LibTorch) enable GPU-accelerated computations.
Origins of NX
- José Valim's interest in AI was rekindled by Sean Moriarity's book, "Genetic Algorithms in Elixir."
- This led to a collaboration and the development of NX for machine learning in Elixir.