Zheng Wang

From Collapse to Control: Understanding and Extending Context Length in Emerging Hybrid Models via Universal Position Interpolation

Slow-Fast Policy Optimization: Reposition-Before-Update for LLM Reasoning

PuzzleMoE: Efficient Compression of Large Mixture-of-Experts Models via Sparse Expert Merging and Bit-packed inference

ZoomVLM: A Tuning-Free Framework for Efficient Video Understanding via Adaptive Zooming in Vision-Language Models

Model Tells You Where to Merge: Adaptive KV Cache Merging for LLMs on Long-Context Tasks