For OpenMOSS Mechanistic Interpretability Team's Sparse Autoencoder (SAE) research.

Finding Direction of arrival (DOA) of small UAVs using Sparse Denoising Autoencoders and Deep Neural Networks.

[ICLR 2025] Monet: Mixture of Monosemantic Experts for Transformers

Code for the paper: Discover-then-Name: Task-Agnostic Concept Bottlenecks via Automated Concept Discovery. ECCV 2024.

Sparse Autoencoders using FashionMNIST dataset

Hyperspectral Band Selection using Self-Representation Learning with Sparse 1D-Operational Autoencoder (SRL-SOA)

Providing the answer to "How to do patching on all available SAEs on GPT-2?". It is an official repository of the implementation of the paper "Evaluating Open-Source Sparse Autoencoders on Disentangling Factual Knowledge in GPT-2 Small"

Use evolution with sparse autoencoders

Implementation and analysis of Sparse Autoencoders for neural network interpretability research. Features interactive visualization dashboard and W&B integration.

performing mechanistic interpretability on inceptionV1, from linear prob and sparse direction maximization to adversarial and ciruict patching & ablation

studying (self-)supervised representations of Euclid galaxy imaging via SAEs

A framework for conducting interpretability research and for developing an LLM from a synthetic dataset.

Official code release for the paper: "Mammo-SAE: Interpreting Breast Cancer Concept Learning with Sparse Autonencoders"

Unified SAE and Transcoder training using EleutherAI/sparsify library for neural network interpretability research