Junfeng Fang

AlphaSteer: Learning Refusal Steering with Principled Null-Space Constraint

AudioTrust: Benchmarking The Multifaceted Trustworthiness of Audio Large Language Models

DualEdit: Mitigating Safety Fallback in LLM Backdoor Editing via Affirmation-Refusal Regulation

GuardAlign: Test-time Safety Alignment in Multimodal Large Language Models

On Predictability of Reinforcement Learning Dynamics for Large Language Models

Goal-Aware Identification and Rectification of Misinformation in Multi-Agent Systems

Contrastive Weak-to-Strong Generalization

Enhancing Multi-Modal LLMs Reasoning via Difficulty-Aware Group Normalization

Parameters vs. Context: Fine-Grained Control of Knowledge Reliance in Language Models

RefineX: Learning to Refine Pre-training Data at Scale from Expert-Guided Programs

UniErase: Towards Balanced and Precise Unlearning in Language Models

Multi-agent Architecture Search via Agentic Supernet

AnyEdit: Edit Any Knowledge Encoded in Language Models

AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models

DAMA: Data- and Model-aware Alignment of Multi-modal LLMs

GuardReasoner-VL: Safeguarding VLMs via Reinforced Reasoning

Search and Refine During Think: Facilitating Knowledge Refinement for Improved Retrieval-Augmented Reasoning

G-Designer: Architecting Multi-agent Communication Topologies via Graph Neural Networks

Reinforced Lifelong Editing for Language Models

On the Role of Attention Heads in Large Language Model Safety

3D-GSRD: 3D Molecular Graph Auto-Encoder with Selective Re-mask Decoding

Is Factuality Enhancement a Free Lunch For LLMs? Better Factuality Can Lead to Worse Context-Faithfulness

Iterative Substructure Extraction for Molecular Relational Learning with Interactive Graph Information Bottleneck

DiffGAD: A Diffusion-based Unsupervised Graph Anomaly Detector

Capturing substructure interactions by invariant Information Bottle Theory for Generalizable Property Prediction

NExT-Mol: 3D Diffusion Meets 1D Language Modeling for 3D Molecule Generation

Neuron-Level Sequential Editing for Large Language Models

Decoding by Contrasting Knowledge: Enhancing LLMs' Confidence on Edited Facts

MIND SCRAMBLE: UNVEILING LARGE LANGUAGE MODEL PSYCHOLOGY VIA TYPOGLYCEMIA