HydraGNN
Project record
Maintained by Oak Ridge National Laboratory12
HydraGNN is a PyTorch implementation of multi-headed graph neural networks from Oak Ridge National Laboratory, with separate output heads for graph-level and node-level properties. Its README lists distributed training with DDP, FSDP, and DeepSpeed; equivariant layers such as EGNN, PaiNN, MACE, and DimeNet; heterogeneous graph learning; global attention through GPS; and training of machine-learned interatomic potentials with energy-conserving forces.1
- Repository: Repository (ORNL/HydraGNN) (external site: github.com)
- Documentation: User manual (external site: github.com)
- License: LICENSE (BSD 3-Clause) (external site: github.com)
- Release notes: Releases (external site: github.com)
Availability and license
Overall availability
Source code is public on GitHub and installed from source with pip after installing the listed dependencies.1
Availability is separate from permission: read the license before using or redistributing.
Component reuse rights
- weights
- Unknown — no complete fact-level rights review
- code
- Reviewed qualifying license recorded — check scope and conditions
- data
- Unknown — no complete fact-level rights review
- documentation
- Unknown — no complete fact-level rights review
No complete system-rights review is recorded for this release.
Public materials checklist
| Item | Status | Notes and evidence |
|---|---|---|
| Source codeIs the source code publicly readable? | Public | Public repository in ORNL's GitHub organization under the BSD 3-Clause License.12 |
| DocumentationIs user documentation published? | Public | The README covers capabilities, dependencies, installation, a quick start, and configuration settings, and links to a user manual, feature guides, and a project wiki.1 |
| InstallationAre installation instructions or packages publicly available? | Public | Dependencies are installed with an included script or modular pip requirement files (core, PyTorch, PyTorch Geometric, optional extras); HydraGNN itself is installed with pip from the cloned repository.1 |
| Supported platformsAre supported operating systems or hardware documented? | Partial | The README states that HydraGNN is tested on Python 3.11 through 3.14, accepts PyTorch 2.13 or 2.14, and ships installation assets for HPC facilities. No operating-system or accelerator support matrix was found.1 |
| Release statusAre versioned releases published? | Public | Versioned GitHub releases include v4.0 (August 15, 2025) and v5.0 (April 14, 2026).3 |
What it is useful for
Training graph neural network surrogate models on atomistic and materials data, such as density functional theory outputs, including distributed training at supercomputing scale. The README also documents constraint-aware training for optimal power flow problems and gives example scripts for each workflow.1
Run and use notes
- The README's baseline dependency script uses NumPy 2.4.6, PyTorch 2.13.0, torchvision 0.28.0, and PyTorch Geometric 2.8.0. The pyg-lib dependency is built from source, which requires a working C++ compiler.1
Organization context
U.S. eligibility
Eligible · basis: U.S.-governed project
HydraGNN is published in Oak Ridge National Laboratory's ORNL GitHub organization, and its BSD 3-Clause license names Oak Ridge National Laboratory as copyright holder. ORNL is a U.S. Department of Energy national laboratory in Oak Ridge, Tennessee, managed by UT-Battelle LLC for DOE.124
Sources
This listing is not an endorsement, a safety assessment, or a federal approval.