TUM-DAML/gemnet_pytorch
TUM-DAML/gemnet_pytorch has 224 GitHub stars and 38 forks. GitHot measured +1 stars in 24 hours and +1 in 7 days. Last checked 2026-11-22 05:54 UTC. Last push 2023-04-26.
实测星标增长
- 24小时
- +1
- 7天
- +1
- 15天
- 正在收集历史
- 1个月
- 正在收集历史
各周期内的星标净变化,周期相互重叠,并非每日新增。缺失的测量保持空白。星标代表关注度,不代表软件质量。
关于
GemNet model in PyTorch, as proposed in "GemNet: Universal Directional Graph Neural Networks for Molecules" (NeurIPS 2021)
机器学习README.md
原文GemNet: Universal Directional Graph Neural Networks for Molecules
Reference implementation in PyTorch of the geometric message passing neural network (GemNet). You can find its original TensorFlow 2 implementation in another repository. GemNet is a model for predicting the overall energy and the forces acting on the atoms of a molecule. It was proposed in the paper:
GemNet: Universal Directional Graph Neural Networks for Molecules
by Johannes Gasteiger, Florian Becker, Stephan Günnemann
Published at NeurIPS 2021
and further analyzed in
How robust are modern graph neural network potentials in long and hot molecular dynamics simulations?
by Sina Stocker*, Johannes Gasteiger*, Florian Becker, Stephan Günnemann and Johannes T. Margraf
Published in Machine Learning: Science and Technology, 2022
*Both authors contributed equally to this research. Note that the author's name has changed from Johannes Klicpera to Johannes Gasteiger.
Run the code
Adjust config.yaml (or config_seml.yaml) to your needs.
This repository contains notebooks for training the model (train.ipynb) and for generating predictions on a molecule loaded from ASE (predict.ipynb). It also contains a script for training the model on a cluster with Sacred and SEML (train_seml.py). Further, a notebook is provided to show how GemNet can be used for MD simulations (ase_example.ipynb).
Compute scaling factors
You can either use the precomputed scaling_factors (in scaling_factors.json) or compute them yourself by running fit_scaling.py. Scaling factors are used to ensure a consistent scale of activations at initialization. They are the same for all GemNet variants.
Contact
Please contact [email protected] if you have any questions.
Cite
Please cite our papers if you use the model or this code in your own work:
@inproceedings{gasteiger_gemnet_2021,
title = {GemNet: Universal Directional Graph Neural Networks for Molecules},
author = {Gasteiger, Johannes and Becker, Florian and G{\"u}nnemann, Stephan},
booktitle={Conference on Neural Information Processing Systems (NeurIPS)},
year = {2021}
}
@article{stocker_robust_2022,
title = {How robust are modern graph neural network potentials in long and hot molecular dynamics simulations?},
author = {Stocker, Sina and Gasteiger, Johannes and Becker, Florian and G{\"u}nnemann, Stephan and Margraf, Johannes T.},
volume = {3},
doi = {10.1088/2632-2153/ac9955},
number = {4},
journal = {Machine Learning: Science and Technology},
year = {2022},
pages = {045010},
}
探索相关项目
加载中…