Fangxiang Feng

An Information-Theoretic Parameter-Free Bayesian Framework for Probing Labeled Dependency Trees from Attention Score

TAMP: Task-aware Multimodal Pre-Interaction for fine-grained Large Language Models

SMAR: Soft Modality-Aware Routing Strategy for MoE-based Multimodal Large Language Models Preserving Language Capabilities

ChunkLLM: A Lightweight Pluggable Framework for Accelerating LLMs Inference