从错误预测到因果影响：审计选择题问答中的 SAE 特征

View PDF HTML (experimental)

Abstract:Sparse autoencoders (SAEs) can predict internal features associated with model errors, but association alone does not establish causal influence. We study this distinction in multiple-choice question answering, where answer preference can be measured by the logit margin between the correct answer and the competing choices. Across eight datasets, we analyze Gemma~3 1B and 4B with Gemma Scope SAEs. We propose a pipeline to filter out the features that have high correlation with format and content artifacts, such as prompt and answer permutation, and evaluate the remaining ones with causal feature interventions. Our results show that SAE activations contain predictive signals about wrongness and uncertainty across models and datasets. After filtering artifact-contaminated features, selected features for causal intervention at question and option spans produce larger absolute margin shifts and a higher wrong-to-correct rate than randomly selected features, indicating greater causal leverage. However, answer flips are rare, and we do not observe stable effect directions across meaning-preserving changes in prompt format and answer order. An adapted HalluSAE baseline produces larger final-token margin changes, but relies on substantially contaminated features and does not significantly improve correction rates. Therefore, the identified features reflect context-dependent answer competition rather than the mechanism of incorrect answer errors.

Submission history

From: Aria Nourbakhsh [view email]
[v1] Tue, 27 Oct 2026 23:34:33 UTC (250 KB)