Showing cs.CLShow all
2 papers · 1 filter
cs.CL2025
Where Did It Go Wrong? Attributing Undesirable LLM Behaviors via Representation Gradient Tracing
Zhe Li, Wei Zhao, Yige Li +1
Large Language Models (LLMs) have demonstrated remarkable capabilities, yet their deployment is frequently undermined by undesirable behaviors such as generating harmful content, f…
cs.CL2025
Zero-Shot Defense Against Toxic Images via Inherent Multimodal Alignment in LVLMs
Wei Zhao, Zhe Li, Yige Li +1
Large Vision-Language Models (LVLMs) have made significant strides in multimodal comprehension, thanks to extensive pre-training and fine-tuning on large-scale visual datasets. How…