A complete end-to-end pipeline for LLM interpretability with sparse autoencoders (SAEs) using Llama 3.2, written in pure PyTorch and fully reproducible.
Implementation of the stacked denoising autoencoder in Tensorflow
Pivotal Token Search
Official Code for Paper: Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation
Pytorch implementations of various types of autoencoders
SANSA - sparse EASE for millions of items
Sparse Embedding Compression for Scalable Retrieval in Recommender Systems
Multi-Layer Sparse Autoencoders (ICLR 2025)
Tensorflow Examples
Official Triton kernels for TopK and HierarchicalTopK Sparse Autoencoder decoders.
Sparse Autoencoders (SAE) vs CLIP fine-tuning fun.
[NeurIPS 2025] This is the official repository for VL-SAE: Interpreting and Enhancing Vision-Language Alignment with a Unified Concept Set
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"
Mechanistic experiments on how LLMs represent and compute arithmetic internally, with strict no-parser controls, reproducible audits, and an interactive article.
[JAMIA] Official repository of Deep Propensity Network - Sparse Autoencoder(DPN-SA)
Automated Interpretability and Feature Discovery in Language Models with AI Agents
Repository for "From What to How: Attributing CLIP's Latent Components Reveals Unexpected Semantic Reliance"
Interpret and control dense embedding via sparse autoencoder.
Specialization after Generalization
Evaluating Sparse Autoencoders for Monosemantic Representation (EACL)
To associate your repository with the sparse-autoencoder topic, visit your repo's landing page and select "manage topics."