13 papers
Introspecting Alignment Shifts Beyond Behaviors Implanted Through Fine-Tuning
Kotaro Yoshida, Laura Gomezjurado Gonzalez, Yukinori Yamamoto +3
Fine-tuning enables a source model to acquire desired capabilities and behaviors in a target domain while retaining much of its general-purpose competence. However, this adaptation…
Learning More from Less: Exploiting Counterfactuals for Data-Efficient Chart Understanding
Jianzhu Bao, Haozhen Zhang, Kuicai Dong +5
Vision-Language Models (VLMs) have demonstrated remarkable progress in chart understanding, largely driven by supervised fine-tuning (SFT) on increasingly large synthetic datasets.…
Vocabulary Hijacking in LVLMs: Unveiling Critical Attention Heads by Excluding Inert Tokens to Mitigate Hallucination
Yangneng Chen, Junlin Li, Weijun Yao +4
Large Vision-Language Models (LVLMs) have achieved remarkable progress in multimodal tasks, yet their reliability is persistently undermined by hallucinations-generating text that…
Personalizing LLMs with Binary Feedback: A Preference-Corrected Optimization Framework
Xilai Ma, Liye Zhao, Weijun Yao +3
Large Language Model (LLM) personalization aims to align model behaviors with individual user preferences. Existing methods often focus on isolated user histories, neglecting the e…
Task-Aware LLM Routing with Multi-Level Task-Profile-Guided Data Synthesis for Cold-Start Scenarios
Hui Liu, Bin Zou, Kecheng Chen +3
Large language models (LLMs) exhibit substantial variability in performance and computational cost across tasks and queries, motivating routing systems that select models to meet u…
Unveiling Modality Bias: Automated Sample-Specific Analysis for Multimodal Misinformation Benchmarks
Hehai Lin, Hui Liu, Shilei Cao +3
Numerous multimodal misinformation benchmarks exhibit bias toward specific modalities, allowing detectors to make predictions based solely on one modality. While previous research…
STARE at the Structure: Steering ICL Exemplar Selection with Structural Alignment
Jiaqian Li, Qisheng Hu, Jing Li +1
In-Context Learning (ICL) has become a powerful paradigm that enables LLMs to perform a wide range of tasks without task-specific fine-tuning. However, the effectiveness of ICL hea…
Jailbreak-R1: Exploring the Jailbreak Capabilities of LLMs via Reinforcement Learning
Weiyang Guo, Zesheng Shi, Zhuo Li +6
As large language models (LLMs) grow in power and influence, ensuring their safety and preventing harmful output becomes critical. Automated red teaming serves as a tool to detect…
Adaptive Detoxification: Safeguarding General Capabilities of LLMs through Toxicity-Aware Knowledge Editing
Yifan Lu, Jing Li, Yigeng Zhou +7
Large language models (LLMs) exhibit impressive language capabilities but remain vulnerable to malicious prompts and jailbreaking attacks. Existing knowledge editing methods for LL…
Multi-objective Large Language Model Alignment with Hierarchical Experts
Zhuo Li, Guodong Du, Weiyang Guo +8
Aligning large language models (LLMs) to simultaneously satisfy multiple objectives remains a significant challenge, especially given the diverse and often conflicting nature of hu…
MTSA: Multi-turn Safety Alignment for LLMs through Multi-round Red-teaming
Weiyang Guo, Jing Li, Wenya Wang +4
The proliferation of jailbreak attacks against large language models (LLMs) highlights the need for robust security measures. However, in multi-round dialogues, malicious intention…
Multi-Modality Expansion and Retention for LLMs through Parameter Merging and Decoupling
Junlin Li, Guodong DU, Jing Li +8
Fine-tuning Large Language Models (LLMs) with multimodal encoders on modality-specific data expands the modalities that LLMs can handle, leading to the formation of Multimodal LLMs…
Knowledge Fusion of Large Language Models Via Modular SkillPacks
Guodong Du, Zhuo Li, Xuanning Zhou +9
Cross-capability transfer is a key challenge in large language model (LLM) research, with applications in multi-task integration, model compression, and continual learning. Recent…