AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

LASER: A Corrective Lens for LVLMs via Visual Attention Preservation and Sink Suppression

Bowen Yuan, Zijian Wang, Yadan Luo, Shijie Wang, Zi Helen Huang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Attention-based visual token pruning · 0 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
add GRPO training (ablation)Qwen2.5-VL-7B · average over 8 benchmarksaverage accuracy: 48.4 -> 51.1Improved
add visual grounding reward R_vis (ablation)GRPO on Qwen2.5-VL-7B · average over 8 benchmarksaverage accuracy: 51.1 -> 52.2Improved
add sink suppression reward R_supp (ablation)GRPO + R_vis on Qwen2.5-VL-7B · General reasoningaccuracy: 55.1 -> 58.6Improved
add sink suppression reward R_supp (ablation)GRPO + R_vis on Qwen2.5-VL-7B · average over 8 benchmarksaverage accuracy: 52.2 -> 53.7Improved
add LASER rewards on SFT cold-start modelRevisual-R1 · average over benchmarksaverage accuracy: 52.2 -> 58.0Improved
add LASER rewards on top of GRPO for Revisual-R1 (ablation)GRPO on Revisual-R1 · average over benchmarksaverage accuracy: 56.5 -> 58.0Improved
one-time suppression of visual value states at 1/8 decoding stage (full masking) (ablation)Qwen2.5-VL-7B inference · MMStaraccuracy: 62.8 -> 48.9Hurt
one-time full masking of visual states at 3/4 decoding stage (ablation)Qwen2.5-VL-7B inference · MMStaraccuracy: 2.0% decreaseHurt
moderate suppression (kappa 0.5) at early stage versus later stage (ablation)Qwen2.5-VL-7B inference · MMStaraccuracy: 3.7% drop vs 1.3%Hurt
redistribute attention from sink to non-sink tokens instead of uniform scaling (ablation)Qwen2.5-VL-7B inference · MMStaraccuracyImproved
early-weighted R_vis instead of uniform R_vis (ablation)LASER · 6-benchmark averageaverage accuracy: 52.2 vs 52.3No clear effect
vary lambda in R_vis (ablation)LASER · average accuracy over benchmarksaverage accuracy: 52.4 / 53.3 / 54.6 / 53.9 / 53.2 for lambda 1 / 2.5 / 5 / 7.5 / 10No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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