Paper

SALSA: Speech Aware LLM Adaptation via Learned Steering Activation Vectors

Unreadunread

AI summary

arXiv:2606. 00460v1 Announce Type: new Abstract: Speech-aware large language models often generalize poorly to out-of-domain settings.

Reading status

Priority

About this paper

Authors: Yekaterina Yegorova, Argyrios Gerogiannis, Haolong Zheng, Julia Hockenmaier, Chang D. Yoo, Mark A. Hasegawa-Johnson
Released: 2026-06-02 (1mo ago)
Topic: Related to current project
Relation to my work: Read next because SALSA: Speech Aware LLM Adaptation via Learned Steering Activation Vectors 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: code, text, rect, line, rate, trained, language, model. Source: arxiv cs.CL (NLP).
arXiv: 2606.00460

Ask Claude about this paper

Posts a comment with @claude so the runner spins up a Claude Code session scoped to SALSA: Speech Aware LLM Adaptation via Learned Steering Activation Vectors. Replies appear in the conversation below.

Comments

No comments yet.

Related records

No links yet.