avatar

Kangye Ji (季康烨)

Ph.D. Student
Tsinghua University
kangyejics@gmail.com


About Me

I am a Ph.D. student at Tsinghua University, advised by Prof. Wang Zhi, and a member of the Efficient Deep Learning and Embodiment Group. My long-term research goal is to build action-native foundation models that act, adapt, and continually learn from the physical world.

Education

Sep. 2021 - June. 2025, B.E in Computer Science and Technology, Xidian University
From Sep. 2025, Ph.D. in Computer Science and Technology, Tsinghua University

Research Interests

If this vision resonates with you, please feel free to reach out!

Publications

  1. ICML
    Sparse ActionGen: Accelerating Diffusion Policy with Real-time Pruning
    Kangye Ji, Jianbo Zhou, Yuan Meng, Ye Li, Hanyun Cui, Zhi Wang
    International Conference on Machine Learning (ICML), 2026.
    PDF Code Project Page BibTex

  2. CVPR
    Test-time Sparsity for Extreme Fast Action Diffusion
    Kangye Ji, Yuan Meng, Jianbo Zhou, Ye Li, Chen Tang, Zhi Wang
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026.
    PDF Code BibTex

  3. ICLR
    Block-wise Adaptive Caching for Accelerating Diffusion Policy
    Kangye Ji, Yuan Meng, Hanyun Cui, Ye Li, Jianbo Zhou, Shengjia Hua, Lei Chen, Zhi Wang
    International Conference on Learning Representations (ICLR), 2026.
    PDF Code Project Page BibTex

  4. AAAI
    Jump-teaching: Combating Sample Selection Bias via Temporal Disagreement
    Kangye Ji, Fei Cheng, Zeqing Wang, Qichang Zhang, Bohu Huang
    The Association for the Advancement of Artificial Intelligence (AAAI), 2026.
    PDF Code BibTex

  5. ICLR
    SP-VLA: A Joint Model Scheduling and Token Pruning Approach for VLA Model Acceleration
    Ye Li, Yuan Meng, Zewen Sun, Kangye Ji, Chen Tang, Jiajun Fan, Xinzhu Ma, Shutao Xia, Zhi Wang, Wenwu Zhu
    International Conference on Learning Representations (ICLR), 2026.
    PDF Code BibTex

  6. AAAI
    SAMCL: Empowering SAM to Continually Learn from Dynamic Domains with Extreme Storage Efficiency
    Zeqing Wang, Kangye Ji, Di Wang, Haibin Zhang, Fei Cheng
    The Association for the Advancement of Artificial Intelligence (AAAI), 2026.
    PDF Code BibTex Oral

  7. ECCV
    RoboStream: Weaving Spatio-Temporal Reasoning with Memory in Vision-Language Models for Robotics
    Yuzhi Huang, Jie Wu, Weijue Bu, Ziyi Xiong, Gaoyang Jiang, Ye Li, Kangye Ji, Shuzhao Xie, Yue Huang, Chenglei Wu, Jingyan Jiang, Zhi Wang
    European Conference on Computer Vision (ECCV), 2026.
    PDF Code Project Page BibTex

  8. PRCV
    Evolving Cache Schedules for Fast Diffusion Policy Inference
    Siying Wang, Kangye Ji, Di Wang, Fei Cheng
    Chinese Conference on Pattern Recognition and Computer Vision (PRCV), 2026.
    PDF Code BibTex

  9. ASP-DAC
    Efficient Continual Learning with Low Memory Footprint For Edge Device
    Zeqing Wang, Fei Cheng, Kangye Ji, Bohu Huang
    Proceedings of the Asia and South Pacific Design Automation Conference (ASP-DAC), 2025.
    PDF Code BibTex

Preprints

  1. Real-Time VLAs with World Model Foresight
    Kangye Ji, Lei Bai, Huanan Liu, Ye Li, Hanyun Cui, Liuyang Song, Chenglei Wu, Hongbo Zhang, Yuan Meng, Zhi Wang
    PDF Code Project Page

  2. ProprioVLA: Proprioceptive Self-Grounding for Vision-Language-Action Policies
    Kangye Ji, Lei Bai, Zhibo Wang, Boxue Yang, Ye Li, Chenglei Wu, Yuan Meng, Zhi Wang
    PDF Code Project Page

Projects

  1. VLA-Lab
    A toolkit for tracking and visualizing the real-world deployment process of VLA models. Debug, visualize, and analyze your VLA deployments with unified logging and interactive visualization.
    GitHub

Selected Awards