paper-with-me

Papers

Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution

2026-06-26 · Kevin Der, Harish Kamath, Ben Thompson arxiv

Sparse autoencoders (SAEs) have become a useful tool for extracting interpretable features in language models. However, standard SAE architectures operate on individual token activations, meaning that the number of active features scales linearly with context length, and studying long model transcripts becomes difficult. We introduce turn-averaged SAEs, which represent a single Human or Assistant turn with a fixed number of features by learning to reconstruct the average model activation across the turn. We find that turn-averaged features describe a single turn's high-level characteristics more completely than per-token features when judged by an LLM. We also demonstrate that turn-averaged SAEs greatly simplify common downstream uses of SAEs like attribution graphs. Broadly, turn-averaged SAEs make interpretability techniques practical at long context lengths.

📄 PDF Abstract BibTeX arXiv:2606.28548

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Llama Scope: Extracting Millions of Features from Llama-3.1-8B with Sparse Autoencoders

2024-10-27 · Zhengfu He, Wentao Shu, Xuyang Ge, Lingjie Chen 외

Sparse Autoencoders (SAEs) have emerged as a powerful unsupervised method for extracting sparse representations from language models, yet scalable training remains a significant challenge. We introduce a suite of 256 SAE…

Language ModelingLanguage Modelling

Multimodal Model Diffing for Feature Discovery and Control

2026-08-10 · Hunar Batra, Lachin Naghashyar, Ashkan Khakzar, Philip Torr 외 hf

Multimodal Large Language Models (MLLMs) exhibit strong visual understanding, yet the internal features that cause these behaviors remain difficult to identify, audit, or control. While applicable to post-hoc inspection,…

Improving Dictionary Learning with Gated Sparse Autoencoders

2024-04-24 · Senthooran Rajamanoharan, Arthur Conmy, Lewis Smith, Tom Lieberum 외

Recent work has found that sparse autoencoders (SAEs) are an effective technique for unsupervised discovery of interpretable features in language models' (LMs) activations, by finding sparse, linear reconstructions of LM…

Dictionary Learning

Features that Make a Difference: Leveraging Gradients for Improved Dictionary Learning

2024-11-15 · Jeffrey Olmo, Jared Wilson, Max Forsey, Bryce Hepner 외

Sparse Autoencoders (SAEs) are a promising approach for extracting neural network representations by learning a sparse and overcomplete decomposition of the network's internal activations. However, SAEs are traditionally…

Dictionary Learning

Which Sparse Autoencoder Features Are Real? Model-X Knockoffs for False Discovery Rate Control

2025-11-12 · Tsogt-Ochir Enkhbayar arxiv

Although sparse autoencoders (SAEs) are crucial for identifying interpretable features in neural networks, it is still challenging to distinguish between real computational patterns and erroneous correlations. We introdu…