14 papers
Outlier Smoothing with Closed-Form Rotations for W4A4 Large Language Model Quantization
Jinying Xiao, Bin Ji, Shasha Li +8
Large Language Models (LLMs) quantization facilitates deploying LLMs in resource-limited settings, but existing methods that combine incompatible gradient optimization and quantiza…
EMSEdit: Efficient Multi-Step Meta-Learning-based Model Editing
Xiaopeng Li, Shasha Li, Xi Wang +7
Large Language Models (LLMs) power numerous AI applications, yet updating their knowledge remains costly. Model editing provides a lightweight alternative through targeted paramete…
Rethinking Residual Distribution in Locate-then-Edit Model Editing
Xiaopeng Li, Shanwen Wang, Shasha Li +4
Model editing enables targeted updates to the knowledge of large language models (LLMs) with minimal retraining. Among existing approaches, locate-then-edit methods constitute a pr…
Where Does Vision Meet Language? Understanding and Refining Visual Fusion in MLLMs via Contrastive Attention
Shezheng Song, Shasha Li, Jie Yu +6
Multimodal Large Language Models (MLLMs) have achieved remarkable progress in vision-language understanding, yet how they internally integrate visual and textual information remain…
Seeing Right but Saying Wrong: Inter- and Intra-Layer Refinement in MLLMs without Training
Shezheng Song, Shasha Li, Jie Yu
Multimodal Large Language Models (MLLMs) have demonstrated strong capabilities across a variety of vision-language tasks. However, their internal reasoning often exhibits a critica…
JPU: Bridging Jailbreak Defense and Unlearning via On-Policy Path Rectification
Xi Wang, Songlei Jian, Shasha Li +5
Despite extensive safety alignment, Large Language Models (LLMs) often fail against jailbreak attacks. While machine unlearning has emerged as a promising defense by erasing specif…