Xiaoyu Shen

StreamingThinker: Large Language Models Can Think While Reading

HiDrop: Hierarchical Vision Token Reduction in MLLMs via Late Injection, Concave Pyramid Pruning, and Early Exit

Rethinking Reasoning in Document Ranking: Why Chain-of-Thought Falls Short

Bring Future Vision: Dynamic Computation Allocation Guided by Lightweight Feature Forecaster

Tools are under-documented: Simple Document Expansion Boosts Tool Retrieval

The Few Govern the Many:Unveiling Few-Layer Dominance for Time Series Models

LLaSO: A Reproducible Foundation for Large Speech-Language Models

From LLMs to LRMs: Rethinking Pruning for Reasoning-Centric Models

CHiP: Cross-modal Hierarchical Direct Preference Optimization for Multimodal LLMs

Beyond Content Relevance: Evaluating Instruction Following in Retrieval Models

SkipGPT: Each Token is One of a Kind