arXiv AI October 16, 2026 By Ziye Mei, Haotian Ma

Representation Redistribution in Sparse Autoencoders

The paper introduces the Redistribution Sensitivity Score (RSS) to detect consistent changes in sparse feature activations of Sparse Autoencoders (SAEs) when prompts are transformed. RSS identifies large, directionally consistent feature responses and evaluates their impact on downstream readouts across 27 pretrained SAEs from eight language models. Experiments show that combining RSS support erosion with the clean margin predicts recurrent safety readout failures with high AUROC, demonstrating RSS’s utility in quantifying representation changes without conflating them with failure.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

arXiv AI

Jun 16

Rational Sparse Autoencoder

arXiv:2606. 14990v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) are standard tools for mechanistic interpretability, but current SAE families are constrained by fixed encoder nonlinearities such as ReLU, JumpReLU, and TopK.

By Naiyu Yin, Yue Yu

llms fine-tuning efficiency safety

arXiv Machine Learning

Jun 18

From Sparse Features to Trustworthy Proxies: Certifying SAE-Based Interpretability

arXiv:2606. 18383v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) are increasingly used to extract interpretable features from language models (LMs), yet a central question remains: when can an SAE-based explanation be treated as a faithful view of an underlying frozen LM We study this through a post-hoc generalization framework that certifies the LM via a sparse proxy, obtained by replacing a native hidden activation with its pretrained SAE reconstruction.

By Dibyanayan Bandyopadhyay, Asif Ekbal

llms efficiency safety

arXiv Machine Learning

Sep 10

LLM Layers Immediately Correct Each Other

arXiv:2609.07876v1 Announce Type: cross Abstract: Recent methods in language model interpretability employ techniques such as sparse autoencoders to decompose residual stream contributions into linea...

By Arjun Patrawala, Jiahai Feng, Erik Jones, Jacob Steinhardt

llms safety

arXiv AI

Jun 3

How Quantization Changes Interpretable Features: A Sparse Autoencoder Analysis of Language Models

arXiv:2606. 03002v1 Announce Type: cross Abstract: Quantization is a standard path to deploying large language models, and a quantized model is typically judged acceptable when its perplexity or downstream accuracy stays close to the full-precision original.

By Evan Duan

llms efficiency safety

arXiv Machine Learning

Jul 23

Building Fast, Evaluating Slow: Pipeline Choices Dominate Autointerpretability Score Variance

arXiv:2607. 19386v1 Announce Type: new Abstract: Cross-paper comparison of sparse autoencoder (SAE) interpretability often relies on autointerpretability scores.

By Sinie van der Ben, Neele Roch, Anna Hedstr\"om, Mennatallah El-Assady

llms safety

arXiv AI

Jun 9

Pre-Intervention Prediction of Sparse Autoencoder Steering Side Effects

arXiv:2606. 08365v1 Announce Type: cross Abstract: Sparse autoencoder (SAE) features are increasingly used to steer language models, but feature steering is rarely clean: the same intervention can behave inconsistently across contexts and perturb unrelated features.

By Evan Duan

llms

One email a morning, machine-written

Related stories

Rational Sparse Autoencoder

From Sparse Features to Trustworthy Proxies: Certifying SAE-Based Interpretability

LLM Layers Immediately Correct Each Other

How Quantization Changes Interpretable Features: A Sparse Autoencoder Analysis of Language Models

Building Fast, Evaluating Slow: Pipeline Choices Dominate Autointerpretability Score Variance

Pre-Intervention Prediction of Sparse Autoencoder Steering Side Effects