Research

What we find inside.

Open research on mechanistic interpretability, internal model signals, and the foundations behind NeuronLens products.

Training Sparse Autoencoders That Produce Useful Features: A Practical Guide

A practical guide to training sparse autoencoders across model families, covering frameworks, failure modes, evaluation metrics, and case studies.

From feature discovery to label fidelity in large language models (Nemotron + Llama case study)