cuequivariance-torch

Execute equivariant tensor-product computations on GPUs from PyTorch using cuEquivariance.

420|39|Updated Oct 22, 2024

Search Tags:#gpu #cuda #polynomials #torch #tensor-product #equivariant #segmented_polynomial

Ready to run

npx skills add https://github.com/NVIDIA/cuEquivariance --skill cuequivariance-torch

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: cuequivariance-torch
Source: https://github.com/NVIDIA/cuEquivariance/tree/main/cuequivariance_torch/cuequivariance_torch
Command: npx skills add https://github.com/NVIDIA/cuEquivariance --skill cuequivariance-torch

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps researchers and engineers implement and run equivariant neural network components in PyTorch using cuEquivariance's SegmentedPolynomial backends and CUDA-accelerated primitives.

Core Features & Use Cases

High-performance primitives: SegmentedPolynomial, ChannelWiseTensorProduct, FullyConnectedTensorProduct, Linear, SphericalHarmonics, Rotation, Inversion, SymmetricContraction
Layers: BatchNorm, FullyConnectedTensorProductConv
Backend options: naive, uniform_1d, fused_tp, indexed_linear with CUDA acceleration when available
PyTorch integration: ready-to-use components for model building and training

Quick Start

Import cuequivariance_torch as cuet and instantiate a SegmentedPolynomial-based module to plug into your PyTorch model.

Frequently Asked Questions about cuequivariance-torch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I run equivariant tensor-product computations on GPUs from PyTorch?▼

You can run equivariant tensor-product computations on GPUs by importing cuequivariance_torch and instantiating SegmentedPolynomial-based modules with CUDA acceleration, integrating them directly into your PyTorch model workflows.

What equivariant neural network components are available for PyTorch?▼

Available equivariant neural network components include ChannelWiseTensorProduct, FullyConnectedTensorProduct, Linear, SphericalHarmonics, Rotation, Inversion, and SymmetricContraction, plus layers like BatchNorm and FullyConnectedTensorProductConv.

Which SegmentedPolynomial backends can I use for equivariant operations?▼

SegmentedPolynomial backends include naive, uniform_1d, fused_tp, and indexed_linear, with CUDA acceleration applied automatically when available to optimize equivariant operations performance.

Do I need to install cuequivariance to use these PyTorch equivariance modules?▼

Yes, you need both cuequivariance and cuequivariance_torch installed to execute equivariant tensor-product computations and build neural network components within your PyTorch environment.

Can I use these equivariant layers for building and training standard PyTorch models?▼

Yes, components like FullyConnectedTensorProductConv and BatchNorm are ready-to-use layers designed for model building and training, plugging directly into standard PyTorch workflows with GPU acceleration.

When should I use the fused_tp backend over naive for equivariant tensor products?▼

The fused_tp backend is typically preferred for performance, utilizing CUDA acceleration for equivariant tensor products, while the naive backend serves as a reference or fallback when specialized GPU kernels are unavailable.