# elixir-nx/axon

Nx-powered Neural Networks

Repository: https://github.com/elixir-nx/axon
Canonical: https://ross.abutalabs.com/products/axon
Language: Elixir
License: Apache-2.0
License Family: permissive
Topics: nx, elixir, deep-learning, neural-networks, optimizers
Last push: 2026-08-23T22:35:43+00:00

## Health v2 (maintenance only)
Score: 79/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 42, longevity 100
- inputs: {"age_days": 2046, "days_push": 10, "days_rel": 175, "gap_med": 259.5, "n_releases_24m": 3}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1688, forks 129 (observed 2026-08-28T04:05:22.307097+00:00)

## What it is
Axon is a neural network library for Elixir built on top of Nx numerical definitions. It provides a functional API, a high-level model creation API, and a PyTorch Ignite-inspired training API, all composable and compilable via Nx backends.

## Use cases
- build and train neural networks in Elixir
- define deep learning models with a high-level API
- train models with custom loss functions and optimizers
- export models to ONNX or TensorFlow Lite
- compose custom numerical definitions with standard layers
- accelerate training with Nx compilers and backends

## When to avoid
- you need a mature ecosystem of pretrained models like PyTorch or TensorFlow
- you require GPU tooling and community support beyond the Nx ecosystem
- you want a framework with extensive production deployment tooling

## Facets
- artifact type: library
- maturity: active
- function: deep-learning, machine-learning, llm-training
- domain: deep-learning, machine-learning, artificial-intelligence
- platform: elixir, jvm-scripting
- tags: neural-networks, nx, elixir, optimizers, jit-compilation, onnx-export

## Member repositories
- elixir-nx/axon (main) score 79

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:22.307097+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:38:43.126952+00:00, confidence not recorded.
  - readme: https://github.com/elixir-nx/axon (fetched 2026-08-28T04:05:22.307097+00:00, sha 1804db2fcd8a)
- Data as of 2026-08-30T08:39:29.467469+00:00.