8 papers
ConsisGuard: Aligning Safety Deliberation with Policy Enforcement in LLM Guardrails
Yan Wang, Zhixuan Chu, Zihao Xue +9
Reasoning-based LLM guardrails improve safety moderation by generating explicit rationales before issuing final decisions. However, their rationales do not always lead to faithful…
Robust and Generalizable Safety Steering for Text-to-Image Diffusion Transformers
Zihao Xue, Yan Wang, Zhen Bi +7
Diffusion Transformers have become a powerful backbone for text-to-image generation, but their layered and cross-modal generation process makes safety control fundamentally differe…
Make LLM Learn to Synthesize from Streaming Experiences through Feedback
Zhenlin Hu, Yan Wang, Zhen Bi +7
Large language models (LLMs) have been widely adopted for synthetic data generation, significantly reducing annotation costs. However, most existing studies treat synthesis as a se…
Beyond Superficial Unlearning: Sharpness-Aware Robust Erasure of Hallucinations in Multimodal LLMs
Xianya Fang, Feiyang Ren, Xiang Chen +4
Multimodal LLMs are powerful but prone to object hallucinations, which describe non-existent entities and harm reliability. While recent unlearning methods attempt to mitigate this…
RooflineBench: A Benchmarking Framework for On-Device LLMs via Roofline Analysis
Zhen Bi, Xueshu Chen, Luoyang Sun +4
The transition toward localized intelligence through Small Language Models (SLMs) has intensified the need for rigorous performance characterization on resource-constrained edge ha…
Logical Structure as Knowledge: Enhancing LLM Reasoning via Structured Logical Knowledge Density Estimation
Zhen Bi, Zhenlin Hu, Xueshu Chen +7
The reasoning capabilities of Large Language Models (LLMs) are increasingly attributed to training data quality rather than mere parameter scaling. However, existing data-centric p…