Seeing is Believing: Brain-Inspired Modular Training for Mechanistic Interpretability

Publication date: 4 May 2023

Abstract: We introduce Brain-Inspired Modular Training (BIMT), a method for making neural networks more modular and interpretable. Inspired by brains, BIMT embeds neurons in a geometric space and augments the loss function with a cost proportional to the length of each neuron connection. We demonstrate that BIMT discovers useful modular neural networks for many simple tasks, revealing compositional structures in symbolic formulas, interpretable decision boundaries and features for classification, and mathematical structure in algorithmic datasets. The ability to directly see modules with the naked eye can complement current mechanistic interpretability strategies such as probes, interventions or staring at all weights.

Has companion code repository: https://github.com/KindXiaoming/BIMT

This page was built for publication: Seeing is Believing: Brain-Inspired Modular Training for Mechanistic Interpretability

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6436669)