spkan 0.0.2a4

A drop-in replacement for sparse convolutions using Kolmogorov-Arnold Networks

spkan - Sparse Convolutions with Kolmogorov-Arnold Network

Introducing Sparse Convolutional KANs

This project extends the idea of the innovative architecture of Kolmogorov-Arnold Networks (KAN) to sparse convolutions.

view the PyPI here

Authors

This repository was made by:

• Mellon Zhang | [e-mail] | LinkedIn

Credits

This repository builds upon an implementation of Convolutional-KANs which is available here. This repository uses an efficient implementation of KAN which is available here. The original implementation of KAN is available here. The original paper of the KAN is available here.

Installation

To use as package: currently compatible with: Python: 3.7/3.8/3.9 CUDA: 11.3/11.8

pip install spkan

To edit: Use python==3.9 torch>=2.3.0 and spconv-cu118

git clone https://github.com/meilongzhang/spkan.git