Publications

* denotes equal contribution.

2026

  1. beyond_scattered.png
    Beyond Scattered Acceptance: Fast and Coherent Inference for DLMs via Longest Stable Prefixes
    Pengxiang Li , Joey Tsai , Hongwei Xue , Kunyu Shi, and Shilin Yan
    In International Conference on Learning Representations (ICLR) , 2026
    arXiv
  2. swimbird.png
    SwimBird: Eliciting Switchable Reasoning Mode in Hybrid Autoregressive MLLMs
    Jintao Tong , Shilin Yan , Hongwei Xue , Xiaojun Tang , Kunyu Shi, Guannan Zhang , Ruixuan Li , and Yixiong Zou
    arXiv preprint arXiv:2602.06040, 2026
    arXiv

2025

  1. hybrid_reward.png
    Hybrid Reward Normalization for Process-Supervised Non-Verifiable Agentic Tasks
    Peiran Xu , Zhuohao Li , Xiaoying Xing , Guannan Zhang , Debiao Li , and Kunyu Shi
    arXiv preprint arXiv:2509.25598, 2025
    arXiv

2024

  1. narvl.png
    Non-autoregressive Sequence-to-Sequence Vision-Language Models
    Kunyu Shi, Qi Dong , Luis Goncalves , Zhuowen Tu , and Stefano Soatto
    In Computer Vision and Pattern Recognition (CVPR) , 2024
    arXiv Code
  2. s4.png
    Enhancing Vision-Language Pre-training with Rich Supervisions
    Yuan Gao* , Kunyu Shi*, Pengkai Zhu , Edouard Belval , Oren Nuriel , Srikar Appalaraju , Shabnam Ghadar , Vijay Mahadevan , Zhuowen Tu , and Stefano Soatto
    In Computer Vision and Pattern Recognition (CVPR) , 2024
    arXiv

2023

  1. musketeers.png
    Musketeer (All for One, and One for All): A Generalist Vision-Language Model with Task Explanation Prompts
    Zhaoyang Zhang , Yantao Shen , Kunyu Shi, Zhaowei Cai , Jun Fang , Siqi Deng , Hao Yang , Davide Modolo , Zhuowen Tu , and Stefano Soatto
    arXiv preprint arXiv:2305.07019, 2023
    arXiv

2020

  1. ocfusion.png
    Learning Instance Occlusion for Panoptic Segmentation
    Justin Lazarow* , Kwonjoon Lee* , Kunyu Shi*, and Zhuowen Tu
    In Computer Vision and Pattern Recognition (CVPR) , Jun 2020
    arXiv Code