SPUS PDE-UNet-36M Navier-Stokes PWC Model

This is a pretrained PDE-UNet model with 36 million parameters designed for Navier-Stokes equation solving in computational fluid dynamics. The model architecture is based on UNet, which is commonly used in image-to-image regression tasks for fluid flow prediction. The PWC variant implements point-wise coupling strategy for improved flow field accuracy. Available on HuggingFace for download and deployment.