Decomposed feature vis - simple - Colaboratory

Decomposed feature visualization is a techhnique for understanding the mechanics of how neural network features operate. This is a very early experiment with them.

Google Colab Sign in

related reading

• Feature Visualizationdistill.pub
• Feature Visualizationdistill.pub
• Toy Models of Superpositiontransformer-circuits.pub
• Neural Networks, Manifolds, and Topology -- colah's blogcolah.github.io
• Towards Monosemanticity: Decomposing Language Models With Dictionary Learningtransformer-circuits.pub
• The Building Blocks of Interpretabilitydistill.pub
• Home - colah's blogcolah.github.io
• Neural Networks, Types, and Functional Programming -- colah's blogcolah.github.io
• Tensorflow — Neural Network Playgroundplayground.tensorflow.org
• Interpreting Language Model Parametersgoodfire.ai
• Towards Monosemanticity: Decomposing Language Models With Dictionary Learningtransformer-circuits.pub
• Interpretability in Parameter Space: Minimizing Mechanistic Description Length with Attribution-based Parameter Decompositionarxiv.org