KANLab visualizes Kolmogorov-Arnold Networks in the browser
Original title: KANLab: TensorFlow Playground for Kolmogorov-Arnold Networks
The open-source tool, forked from TensorFlow Playground, lets users edit B-spline activations and watch gradient descent.
KANLab is a browser-based visualization tool for teaching Kolmogorov-Arnold Networks, forked from TensorFlow Playground. It omits residual connections and learnable scaling parameters so displayed B-spline activation curves match their trainable control points. Users can inspect and modify activation functions and initialization schemes, disrupt training, and observe gradient descent, plus explore activation variance, interactive pruning, and retraining in a compact 2D setting. It is open source at kanlab.app.
- KANLab is forked from TensorFlow Playground and runs in the browser.
- Residual connections and learnable scaling parameters are omitted for clarity.
- Users can modify activations, initialization, pruning, and retraining.
- The project is open source and hosted at kanlab.app.
- Products
- KANLabTensorFlow Playground
From the original article
1 min read at huggingface.co
KANLab is a browser-based visualization tool designed to interactively teach the core concepts of Kolmogorov-Arnold Networks (KAN), forked from the widely successful TensorFlow Playground.
By Lukas Niehaus, Robin Rawiel, Ulf Krumnack · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.