Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution
This paper introduces turn-averaged sparse autoencoders (SAEs), which represent each human or assistant turn with a fixed number of features by reconstructing the average model activation across the turn. This addresses the issue of features scaling linearly with context length in standard SAEs. The turn-averaged features describe high-level characteristics more completely and simplify downstream uses like attribution graphs.
[2606.28548] Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution
[Submitted on 26 Jun 2026]
Title:Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution
View a PDF of the paper titled Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution, by Kevin Der and 2 other authors
View PDF HTML (experimental)
Abstract: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.
Subjects:
Computation and Language (cs.CL); Machine Learning (cs.LG)
Cite as: arXiv:2606.28548 [cs.CL]
(or arXiv:2606.28548v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.28548
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Kevin Der [view email] [v1] Fri, 26 Jun 2026 19:07:34 UTC (1,684 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution, by Kevin Der and 2 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.CL
new | recent | 2026-06
Change to browse by:
cs cs.LG
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
Loading...
Data provided by:
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)