AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

AHAMask: Reliable Task Specification for Large Audio Language Models Without Instructions

Yiwei Guo, Bohan Li, Hankun Wang, Zhihan Li, Shuai Wang, Xie Chen, Kai Yu

AAAI 2026 · Large language models › Language, safety and interpretability · approach: Pruning and circuits in LLMs · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace instruction with AHAMaskSALMONN · gender recognitionaccuracy: 96.79 -> 98.05Improved
replace instruction with AHAMaskSALMONN · AudioCaps AACMETEOR: 20.60 -> 24.15Improved
replace instruction with AHAMaskSALMONN · ASV, S2TT, ASRaccuracy/BLEU/WER: 93.49 -> 93.24 (ASV); 34.48 -> 33.90 (S2TT)No clear effect
replace instruction with AHAMaskQwen2Audio-Base · VoxCeleb1-O ASVaccuracy: 49.24 -> 85.75Improved
replace instruction with AHAMaskQwen2Audio-Base · LibriSpeech test-clean ASRWER: 1.77 -> 3.11Hurt
replace instruction with AHAMaskQwen2Audio-Base · CoVoST2 en-zh S2TTBLEU-4: 45.00 -> 39.27Hurt
use trained AHAMask instead of random mask with same head count (ablation)SALMONN · LibriSpeech test-clean ASRWER: 21.36 -> 2.10Improved
apply AHAMask trained on the Base model to Instruct model (swap) (ablation)Qwen2Audio-Instruct · LibriSpeech test-clean ASRWER: 2.99 -> 156.87Hurt
replace instruction with AHAMask for ASR|GR orderingSALMONN · ASR|GR compositeIFR (%): 16.03 -> 97.63Improved
replace instruction with AHAMaskQwen2Audio-Instruct · JSON compositeIFR (%): 0.00 -> 58.45Improved
activate heads gradually by importanceSALMONN · ASR, AAC, S2TT, OSRtask metricsImproved
longer, more complex or alter-case instructionsSALMONN · ASR prompt variationWER: to 12%Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy