https://github.com/atomicarchitects/nequix

Nequix: Training a foundation model for materials on a budget.

https://github.com/atomicarchitects/nequix

Science Score: 36.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
    Found .zenodo.json file
  • DOI references
  • Academic publication links
    Links to: arxiv.org
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (9.7%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Nequix: Training a foundation model for materials on a budget.

Basic Info
Statistics
  • Stars: 21
  • Watchers: 0
  • Forks: 1
  • Open Issues: 1
  • Releases: 3
Created 10 months ago · Last pushed 10 months ago
Metadata Files
Readme License

README.md

Nequix

See more information in our preprint.

Usage

Installation

bash pip install nequix

ASE calculator

Using nequix.calculator.NequixCalculator, you can perform calculations in ASE with a pre-trained Nequix model.

```python from nequix.calculator import NequixCalculator

atoms = ... atoms.calc = NequixCalculator("nequix-mp-1") ```

Training

Models are trained with the nequix_train command using a single .yml configuration file:

bash nequix_train <config>.yml

To reproduce the training of Nequix-MP-1, first clone the repo and sync the environment:

bash git clone https://github.com/atomicarchitects/nequix.git cd nequix uv sync

Then download the MPtrj data from https://figshare.com/files/43302033 into data/ then run the following to extract the data:

bash bash data/download_mptrj.sh

Then start the training run. The first time this is run it will preprocess the data into HDF5 files:

bash nequix_train configs/nequix-mp-1.yml

This will take less than 125 hours on a single 4 x A100 node. The batch_size in the config is per-device, so you should be able to run this on any number of GPUs (although hyperparameters like learning rate are often sensitive to global batch size, so keep in mind).

Citation

bibtex @article{koker2025training, title={Training a foundation model for materials on a budget}, author={Koker, Teddy and Smidt, Tess}, journal={arXiv preprint arXiv:2508.16067}, year={2025} }

Owner

  • Name: The Atomic Architects
  • Login: atomicarchitects
  • Kind: organization
  • Location: United States of America

Research Group of Prof. Tess Smidt

GitHub Events

Total
  • Release event: 3
  • Watch event: 16
  • Delete event: 1
  • Issue comment event: 1
  • Push event: 8
  • Fork event: 3
  • Create event: 4
Last Year
  • Release event: 3
  • Watch event: 16
  • Delete event: 1
  • Issue comment event: 1
  • Push event: 8
  • Fork event: 3
  • Create event: 4

Packages

  • Total packages: 1
  • Total downloads:
    • pypi 195 last-month
  • Total dependent packages: 0
  • Total dependent repositories: 0
  • Total versions: 3
  • Total maintainers: 1
pypi.org: nequix

Nequix source code

  • Homepage: https://pypi.org/project/nequix/
  • Documentation: https://nequix.readthedocs.io/
  • License: MIT License Copyright (c) 2025 Teddy Koker Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  • Latest release: 0.2.0
    published 10 months ago
  • Versions: 3
  • Dependent Packages: 0
  • Dependent Repositories: 0
  • Downloads: 195 Last month
Rankings
Dependent packages count: 8.6%
Average: 28.7%
Dependent repos count: 48.7%
Maintainers (1)
Last synced: 10 months ago

Dependencies

.github/workflows/publish.yml actions
  • actions/checkout v4 composite
  • astral-sh/setup-uv v5 composite
.github/workflows/test.yml actions
  • actions/checkout v4 composite
  • astral-sh/setup-uv v5 composite
pyproject.toml pypi
  • ase >=3.24.0
  • cloudpickle >=3.1.1
  • e3nn-jax >=0.20.7
  • equinox >=0.11.11
  • h5py >=3.14.0
  • jax >=0.4.34; sys_platform == 'darwin'
  • jax [cuda12]>=0.4.34; sys_platform == 'linux'
  • jraph >=0.0.6.dev0
  • matscipy >=1.1.1
  • optax >=0.2.5
  • pyyaml >=6.0.2
  • tqdm >=4.67.1
  • wandb >=0.19.11
  • wandb-osh >=1.2.2
uv.lock pypi
  • 111 dependencies