11 papers
Learning-Based Automated Adversarial Red-Teaming for Robustness Evaluation of Large Language Models
Zhang Wei, Hanxuan Chen, Peilu Hu +19
Red-teaming is becoming a central part of large language model (LLM) safety evaluation, yet current practice still relies heavily on expert-written prompts or fixed benchmark suite…
Dispersion Loss Counteracts Embedding Condensation and Improves Generalization in Small Language Models
Chen Liu, Xingzhi Sun, Xi Xiao +8
Large language models (LLMs) achieve remarkable performance through ever-increasing parameter counts, but scaling incurs steep computational costs. To better understand LLM scaling…
V-Retrver: Evidence-Driven Agentic Reasoning for Universal Multimodal Retrieval
Dongyang Chen, Chaoyang Wang, Dezhao Su +6
Multimodal Large Language Models (MLLMs) have recently been applied to universal multimodal retrieval, where Chain-of-Thought (CoT) reasoning improves candidate reranking. However,…
Fast-Slow Efficient Training for Multimodal Large Language Models via Visual Token Pruning
Dingkun Zhang, Shuhan Qi, Yulin Wu +3
Multimodal Large Language Models (MLLMs) suffer from severe training inefficiency issue, which is associated with their massive model sizes and visual token numbers. Existing effor…
FMVP: Masked Flow Matching for Adversarial Video Purification
Duoxun Tang, Xueyi Zhang, Chak Hin Wang +6
Video recognition models remain vulnerable to adversarial attacks, while existing diffusion-based purification methods suffer from inefficient sampling and curved trajectories. Dir…
Transferability of Adversarial Attacks in Video-based MLLMs: A Cross-modal Image-to-Video Approach
Linhao Huang, Xue Jiang, Zhiqiang Wang +5
Video-based multimodal large language models (V-MLLMs) have shown vulnerability to adversarial examples in video-text multimodal tasks. However, the transferability of adversarial…