KANLab Interactive Visualization for Kolmogorov-Arnold Networks
October 13, 2026
KANLab is an open-source, browser-based tool for visualizing Kolmogorov-Arnold Network (KAN) training. It allows users to interactively modify B-spline activation curves, observe gradient descent, and explore pruning in a 2D setting.
HOW THIS AFFECTS YOU
●
researcherYou can use this to intuitively understand how trainable control points affect KAN activation functions.
●
designerYou can use this tool to create more effective educational visualizations for neural architecture concepts.