jacobian-lens
Neuronpedia's Jacobian lens tool for mechanistic interpretability analysis of neural network activations.
Model Description
This folder contains pre-fitted Jacobian Lenses for specific models. They were trained using Anthropic's Jacobian Lens library. The same library can be used to load the lens and local visualizer.
Jacobian Lens can be explored on Neuronpedia at https://neuronpedia.org/jlens.
license: mit
Get the full context.
Sign up to read complete case studies, access detailed metrics, and unlock all use cases.
Get the full context.
Sign up to read complete case studies, access detailed metrics, and unlock all use cases.