AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Safe Responses Matter: Output-Aware Safety Guardrail Mitigate Over-Refusal in MLLMs

Jiayi Li, Kun Zhan

ECCV 2026 · Large language models › Instruction and fine-tuning · approach: Safety alignment after fine-tuning · 1 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 the gating module to the MICL classifier (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: aggregated metric set to 2, the worst value without itImproved
add the contrastive learning term L_con (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: aggregated metric set to 2, the worst value without itImproved
add the sparsity (entropy) regularization L_ent (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: aggregated metric set to 2, the worst value without itImproved
use top-k 0.3 instead of 0.1 (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: 0.1326 -> 0.1299No clear effect
use lambda_con 0.1 instead of 0.05 (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: 0.1441 -> 0.1299Improved
use temperature tau 0.5 instead of 0.1 (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: 0.1816 -> 0.1299Improved
use attention dimension 256 instead of 512 (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: 0.1498 -> 0.1299Improved
use attention dimension 256 instead of 128 (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: 0.1325 -> 0.1299No clear effect
use hidden dimension 256 instead of 768 (ablation)OutGuard · Test set (GQA+AdvBench+SafeBench), LLaVAURUP+ARSP: 0.1585 -> 0.1299Improved
replace LoD-PP linear per-layer classifier with MICL classifier (ablation)OutGuard · Test set, LLaVAF1-Score: 0.8943 -> 0.9335Improved
replace input-side guardrails with output-aware OutGuardOutGuard · XSTest, LLaVAover-refusal rate: 0.7080 -> 0.5840Improved
replace input-side guardrails with output-aware OutGuardOutGuard · XSTest, Qwenover-refusal rate: 0.7760 -> 0.6320Improved

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