Practical overview
When it fits
Use it when a model needs transformer attention over spatial coordinates and feature vectors, or when exploring the repository's protein-backbone denoising example.
Who it is for
- Researchers exploring rotation-equivariant neural networks.
- Developers prototyping geometric deep-learning models that process coordinates and features.
- Practitioners interested in the included protein-backbone denoising example.
Documented uses
- Apply vector-neuron attention to coordinate and feature tensors.
- Build or prototype geometric deep-learning components using rotation-equivariant attention.
- Invoke the included protein-backbone denoising example.
Requirements & limits
The repository documents installation through pip and is licensed under MIT.
- The README's example uses PyTorch tensors and the SidechainNet dependency.
- The README notes that a nonzero bias_epsilon makes equivariance approximate; setting it to zero turns off that epsilon.
- The README reports that the denoising example performs less well than En-Transformer and Equiformer.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Run the protein-backbone denoising example
Install the example dependency and invoke the repository's denoising script.
Before you start
- Python environment with pip
- VN Transformer repository
- SidechainNet, installed as shown in the README
Steps
Install the library with the command below.
pip install VN-transformerInstall the example dependency with the command below.
pip install sidechainnetFrom the repository, invoke the command below.
python denoise.py
Expected outcome
The repository's protein-backbone denoising example is invoked.
Guide limitations
- The README says the example performs less well than En-Transformer and Equiformer.
- The README does not specify further setup or expected output details.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 17 October 2026
- Original website title
- lucidrains/VN-transformer · Official homepage title · retained in its source language
Related options
En-Transformer — The README names it as an alternative whose performance exceeds the included denoising example.
Equiformer — The README names it as an alternative whose performance exceeds the included denoising example.