Training a Foundation Model for Materials on a Budget

TL;DR

Nequix is a compact, efficient foundation model for materials modeling that achieves competitive accuracy with significantly reduced training costs and faster inference, making advanced materials modeling more accessible.

Contribution

We developed Nequix, a lightweight E(3)-equivariant potential with modern training techniques, enabling high-performance materials modeling at a fraction of the usual computational expense.

Findings

Nequix ranks third on Matbench-Discovery and MDR Phonon benchmarks.

It requires 20 times less training compute than comparable models.

Nequix offers two orders of magnitude faster inference speed.

Abstract

Foundation models for materials modeling are advancing quickly, but their training remains expensive, often placing state-of-the-art methods out of reach for many research groups. We introduce Nequix, a compact E(3)-equivariant potential that pairs a simplified NequIP design with modern training practices, including equivariant root-mean-square layer normalization and the Muon optimizer, to retain accuracy while substantially reducing compute requirements. Nequix has 700K parameters and was trained in 100 A100 GPU-hours. On the Matbench-Discovery and MDR Phonon benchmarks, Nequix ranks third overall while requiring a 20 times lower training cost than most other methods, and it delivers two orders of magnitude faster inference speed than the current top-ranked model. We release model weights and fully reproducible codebase at https://github.com/atomicarchitects/nequix.