Xuming Hu

PMark: Towards Robust and Distortion-free Semantic-level Watermarking with Channel Constraints

You only need 4 extra tokens: Synergistic Test-time Adaptation for LLMs

CATMark: A Context-Aware Thresholding Framework for Robust Cross-Task Watermarking in Large Language Models

SaFeR-VLM: Toward Safety-aware Fine-grained Reasoning in Multimodal Models

Investigating Redundancy in Multimodal Large Language Models with Multiple Vision Encoders

KnowMT-Bench: Benchmarking Knowledge-Intensive Long-Form Question Answering In Multi-Turn Dialogues

DiffAdapt: Difficulty-Adaptive Reasoning for Token-Efficient LLM Inference

Rethinking LLM Evaluation: Can We Evaluate LLMs with 200× Less Data?

Zero in on Faithful Anchors: High-Fidelity Visual Token Condensation for Multimodal Large Language Models

Rethinking GCNs for the Traveling Salesman Problem: Are We Encoding Effectively?

Accelerate Diffusion Transformers with Feature Momentum

DocPruner: A Storage-Efficient Framework for Multi-Vector Visual Document Retrieval via Adaptive Patch-Level Embedding Pruning

Understanding-in-Generation: Reinforcing Generative Capability of Unified Model via Infusing Understanding into Generation

Distilling the Thought, Watermarking the Answer: A Principle Semantic Guided Watermark for Reasoning Large Language Models

Consistent3DGen: Bridging Stochastic Generation and Deterministic Reconstruction for Image-to-3D Diffusion Models

Learning Robust Anymodal Segmentor with Unimodal and Cross-modal Distillation

Spot the Critical Words: Text-Guided Visual Token Pruning for Efficient Large Vision-Language Model Inference

FlowKV: Enhancing Multi-Turn Conversational Coherence in LLMs via Isolated Key-Value Cache Management

Can LLMs Maintain Fundamental Abilities under KV Cache Compression?

MOSS-ChatV: Reinforcement Learning with Process Reasoning Reward for Video Temporal Reasoning

Can Watermarked LLMs be Identified by Users via Crafted Prompts?

Don't Just Chase “Highlighted Tokens” in MLLMs: Revisiting Visual Holistic Context Retention

OneForecast: A Universal Framework for Global and Regional Weather Forecasting

LoTA-QAF: Lossless Ternary Adaptation for Quantization-Aware Fine-Tuning

ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference

Look Twice Before You Answer: Memory-Space Visual Retracing for Hallucination Mitigation in Multimodal Large Language Models

Gnothi Seauton: Empowering Faithful Self-Interpretability in Black-Box Transformers

ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection

Mitigating Modality Prior-Induced Hallucinations in Multimodal Large Language Models via Deciphering Attention Causality

RealRAG: Retrieval-augmented Realistic Image Generation via Self-reflective Contrastive Learning

ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference

EXPLORING RESPONSE UNCERTAINTY IN MLLMS: AN EMPIRICAL EVALUATION UNDER MISLEADING SCENARIOS

LAIA-SQL: Enhancing Natural Language to SQL Generation in Multi-Table QA via Task Decomposition and Keyword Extraction

MINER: Mining the Underlying Pattern of Modality-Specific Neurons in Multimodal Large Language Models

Unlocking Speech Instruction Data Potential with Query Rewriting

Look Twice Before You Answer: Memory-Space Visual Retracing for Hallucination Mitigation in Multimodal Large Language Models

MAD: Multi-Alignment MEG-to-Text Decoding

Interpretable Contrastive Monte Carlo Tree Search Reasoning

DRUPI: Dataset Reduction Using Privileged Information