SPUS-PDE-UNet-36M: Neural PDE Solver with 36M Parameters

This is a 36M-parameter neural network model hosted on HuggingFace, designed as a U-Net-based solver for partial differential equations. The model supports data-driven approximation of PDE solutions and can be used as a surrogate model for scientific computing applications. Model weights are available for download, with 42 downloads recorded as of September 2026.