Jing Zhang

MoBE: Mixture-of-Basis-Experts for Compressing MoE-based LLMs

GeoVLMath: Enhancing Geometry Reasoning in Vision-Language Models via Cross-Modal Reward for Auxiliary Line Creation

Streamlining Redundant Layers to Compress Large Language Models

FacLens: Transferable Probe for Foreseeing Non-Factuality in Large Language Models

SP-LoRA: Sparsity-Preserved Low-Rank Adaptation for Sparse Large Language Model

Are Synthetic Time-series Data Really not as Good as Real Data?