11 papers
Steering Vision-Language Models with Joint Sparse Autoencoders
Huizhen Shu, Xuying Li, Hongxu Lin +2
Sparse Autoencoders (SAEs) have shown promise for analyzing language models, but applying them to vision-language models (VLMs) often yields representations that are difficult to u…
Stop Early, Spend Less: Hidden-State Probes as a Practical Recipe for Streaming Moderation of LLM Outputs
Huizhen Shu, Xuying Li, Piao Xue
Deploying large language models in user-facing systems requires efficient output safety filtering. Existing approaches typically rely on a separate moderation model applied after g…
Controllable Mathematical Reasoning via Self-Optimizing Thought Vectors
Xuying LI
We present a novel approach for controllable mathematical reasoning that leverages self-optimizing thought vectors with entropy minimization. Our method introduces learnable though…
RECAP: Reproducing Copyrighted Data from LLMs Training with an Agentic Pipeline
André V. Duarte, Xuying li, Bin Zeng +3
If we cannot inspect the training data of a large language model (LLM), how can we ever know what it has seen? We believe the most compelling evidence arises when the model itself…
LatentGuard: Controllable Latent Steering for Robust Refusal of Attacks and Reliable Response Generation
Huizhen Shu, Xuying Li, Zhuo Li
Achieving robust safety alignment in large language models (LLMs) while preserving their utility remains a fundamental challenge. Existing approaches often struggle to balance comp…
The Resurgence of GCG Adversarial Attacks on Large Language Models
Yuting Tan, Xuying Li, Zhuo Li +2
Gradient-based adversarial prompting, such as the Greedy Coordinate Gradient (GCG) algorithm, has emerged as a powerful method for jailbreaking large language models (LLMs). In thi…
Layer-Wise Perturbations via Sparse Autoencoders for Adversarial Text Generation
Huizhen Shu, Xuying Li, Qirui Wang +3
With the rapid proliferation of Natural Language Processing (NLP), especially Large Language Models (LLMs), generating adversarial examples to jailbreak LLMs remains a key challeng…
Optimizing Safe and Aligned Language Generation: A Multi-Objective GRPO Approach
Xuying Li, Zhuo Li, Yuji Kosuga +1
Aligning large language models (LLMs) with human values and safety constraints is challenging, especially when objectives like helpfulness, truthfulness, and avoidance of harm conf…
Output Length Effect on DeepSeek-R1's Safety in Forced Thinking
Xuying Li, Zhuo Li, Yuji Kosuga +1
Large Language Models (LLMs) have demonstrated strong reasoning capabilities, but their safety under adversarial conditions remains a challenge. This study examines the impact of o…
Targeting the Core: A Simple and Effective Method to Attack RAG-based Agents via Direct LLM Manipulation
Xuying Li, Zhuo Li, Yuji Kosuga +2
AI agents, powered by large language models (LLMs), have transformed human-computer interactions by enabling seamless, natural, and context-aware communication. While these advance…
Precision Knowledge Editing: Enhancing Safety in Large Language Models
Xuying Li, Zhuo Li, Yuji Kosuga +2
Large language models (LLMs) have demonstrated remarkable capabilities, but they also pose risks related to the generation of toxic or harmful content. This work introduces Precisi…