Qinsi Wang

Vision-Zero: Scalable VLM Self-Evolution via Multi-Agent Self-Play

Voice Evaluation of Reasoning Ability: Diagnosing the Modality-Induced Performance Gap

Accelerating Discrete Diffusion Decoding with Parallel Scan

Accelerating Denoising Generative Models is as Easy as Predicting Second-Order Difference

Angles Don’t Lie: Unlocking Training‑Efficient RL Through the Model’s Own Signals

KVCOMM: Online Cross-context KV-cache Communication for Efficient LLM-based Multi-agent Systems

CoreMatching: A Co-adaptive Sparse Inference Framework with Token and Neuron Pruning for Comprehensive Acceleration of Vision-Language Models