TorchANI documentation#
Date: Aug 05, 2025 Version: 2.6 (dev)
Useful links: GitHub Repo | Issue Tracker | Roitberg Lab
TorchANI is an open-source library that supports training, development, and research of ANI-style neural network interatomic potentials. It was originally developed and is currently maintained by the Roitberg group.
If you were using TorchANI before version 2, your code may need updating to be compatible with its new features. Please consult the migration guide for more information. If you still have problems, please open an issue on the GitHub issue tracker.
Installing
Want to install the TorchANI library? Read this section for details on how to
install using conda
or pip
. If you want to build TorchANI from source
instead, please consult the README
in the GitHub repo.
User guide
Start here! This sections describes how to use the TorchANI models in your research or application. It also teaches you about the main classes in in the library, and how to extend them.
API reference
A detailed description of the TorchANI API. It shows all its public functions,
classes, and their methods and properties. Use this for reference. It assumes a
basic understanding of Python and torch
.
Publications
Articles regarding TorchANI itself, and also with specific methods and models it implements. Please consult and cite the corresponding articles if you use TorchANI in a scientific publication.