Paper

Beyond Importance: Interchange-Sobol Sensitivity Reveals Task-Specific Content Channels in Transformer Components

Unreadunread

AI summary

arXiv:2606. 20678v1 Announce Type: new Abstract: Mechanistic interpretability methods summarize a transformer component by a single importance score, conflating two distinct roles: a component may matter because it transports task-relevant content, or because the forward computation degrades when its contribution is removed.

Reading status

Priority

About this paper

Authors: Yifeng Guo, Jin-Hong Du, Xiang Chen
Released: 2026-06-23 (16d ago)
Topic: Related to current project
Relation to my work: Read next because Beyond Importance: Interchange-Sobol Sensitivity Reveals Task-Specific Content Channels in Transformer Components overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: donor, latin, under, eval, rate, compare, control, position. Source: arxiv stat.ML (Machine Learning).
arXiv: 2606.20678

Ask Claude about this paper

Posts a comment with @claude so the runner spins up a Claude Code session scoped to Beyond Importance: Interchange-Sobol Sensitivity Reveals Task-Specific Content Channels in Transformer Components. Replies appear in the conversation below.

Comments

No comments yet.

Related records

No links yet.