Junhao Hu

Junhao Hu

Fourth-year Ph.D. student at CS@PKU, interested in applications and system optimizations of Large Language Models (LLMs).

Published Conference/Journal Papers:

  1. [SC 2026] FluidGPU: Fine-Grained Kernel Disaggregation for Large Model Inference on Heterogeneous GPUs (Findings)
    Tiancheng Hu, Jin Qin, Lei Chen, Junhao Hu, Yuzheng Wang, Mingxing Zhang, Chunwei Xia, Yizhou Shan, Huimin Cui, Ting Cao, Zheng Wang, Xie Tao, Chenxi Wang
    In: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, 2026
    [paper] [code]

  2. [ACL 2026] Lil: Less is Less When Applying Post-Training Sparse-Attention Algorithms in Long-Decode Stage (Findings)
    Junhao Hu, Fangze Li, Mingtao Xu, Feifan Meng, Shiju Zhao, Tiancheng Hu, Ting Peng, Anmin Liu, Wenrui Huang, Chenxu Liu, Ziyue Hua, Tao Xie
    In: Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics, 2026
    [paper] [code]

  3. [FAST 2026] CacheSlide: Unlocking Cross Position-Aware KV Cache Reuse for Accelerating LLM Serving (CCF-A)
    Yang Liu, Yunfei Gu, Liqiang Zhang, Chentao Wu, Guangtao Xue, Jie Li, Minyi Guo, Junhao Hu, Jie Meng
    In: Proceedings of the 24th USENIX Conference on File and Storage Technologies, 2026
    [paper] [code]

  4. [ATC 2025] DeepServe: Serverless Large Language Model Serving at Scales (CCF-A)
    Junhao Hu, Jiang Xu, Zhixia Liu, Yulong He, Yuetao Chen, Hao Xu, Jiang Liu, Jie Meng, Baoquan Zhang, Shining Wan, Gengyuan Dan, Zhiyu Dong, Zhihao Ren, Changhong Liu, Tao Xie, Dayun Lin, Qin Zhang, Yue Yu, Hao Feng, Xusheng Chen, Yizhou Shan
    In: Proceedings of the 2025 USENIX Annual Technical Conference, 2025
    [paper] [code]

  5. [ACL 2025] RaaS: Reasoning-Aware Attention Sparsity for Efficient Long-Decoding Inference (Findings)
    Junhao Hu, Wenrui Huang, Weidong Wang, Zhenwen Li, Tiancheng Hu, Zhixia Liu, Xusheng Chen, Tao Xie, Yizhou Shan
    In: Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics, 2025
    [paper] [code]

  6. [ICML 2025] EPIC: Efficient Position-Independent Caching for Serving Large Language Models (CCF-A)
    Junhao Hu, Wenrui Huang, Weidong Wang, Haoyi Wang, Tiancheng Hu, Qin Zhang, Hao Feng, Xusheng Chen, Yizhou Shan, Tao Xie
    In: Proceedings of the 42nd International Conference on Machine Learning, 2025
    [paper] [code]

  7. [TSE 2025] Directional Diffusion-Style Code Editing Pre-training (CCF-A)
    Qingyuan Liang, Zeyu Sun, Qihao Zhu, Junhao Hu, Yifan Zhao, Yizhou Chen, Mingxuan Zhu, Guoqing Wang, Lu Zhang
    In: Transactions of Software Engineering, 2025
    [paper] [code]

  8. [SCIS 2025] Cupcleaner: A data cleaning approach for comment updating (CCF-A)
    Qingyuan Liang, Zeyu Sun, Qihao Zhu, Junhao Hu, Yifan Zhao, Lu Zhang
    In: Transactions of SCIENCE CHINA Information Sciences, 2025
    [paper] [code]

  9. [ASE 2023] Predicting Compilation Resources for Adaptive Build in an Industrial Setting (CCF-A)
    Junhao Hu, Chaozheng Wang, Hailiang Huang, Huang Luo, Yu Jin, Yuetang Deng, Tao Xie
    In: Proceedings of the 38th IEEE/ACM International Conference on Automated Software Engineering, pages 1808-1813, 2023.
    [paper] [code]

  10. [ESEC/FSE 2023] How Practitioners Expect Code Completion? (CCF-A)
    Chaozheng Wang, Junhao Hu, Cuiyun Gao, Yu Jin, Tao Xie, Hailiang Huang, Zhenyu Lei, Yuetang Deng
    In: Proceedings of the 31th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, pages 1294-1306, 2023.
    [paper] [code]


Technical Reports:

  1. [preprint] MiMo-V2-Flash Technical Report
    Junhao Hu (Core contributor in author list), LLM-Core @ Xiaomi
    In: GitHub
    [paper] [code]

  2. [preprint] MiMo-Audio: Audio Language Models are Few-Shot Learners
    Junhao Hu (Contributor in author list), LLM-Core @ Xiaomi
    In: GitHub
    [paper] [code]

  3. [preprint] xDeepServe: Model-as-a-Service on Huawei CloudMatrix384
    Junhao Hu (Core contributor in author list), xDeepServe Team @ Huawei
    In: arXiv preprint arXiv:2508.02520
    [paper] [code]


Unpublished High-Impact Papers:

  1. [preprint] MemServe: Context Caching for Disaggregated LLM Serving with Elastic Memory Pool
    Cunchen Hu, Heyang Huang, Junhao Hu, Jiang Xu, Xusheng Chen, Tao Xie, Chenxi Wang, Sa Wang, Yungang Bao, Ninghui Sun, Yizhou Shan
    In: arXiv preprint arXiv:2406.17565
    [paper] [code]

  2. [preprint] RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention
    Yang Liu, Zhaokai Luo, Huayi Jin, Zhiyong Wang, Ruozhou He, Boyu Wang, Guanjie Chen, Junhao Hu
    In: arXiv preprint arXiv:2606.06256
    [paper] [code]

  3. [preprint] HYPIC: Accelerating Hybrid-Attention LLM Serving with Position-Independent Caching
    Yifei Liu, Juntong Wu, Yang Liu, Junhao Hu (Corresponding Author), Minghao Li, Xiaoxu Chen, Weihang Chen
    In: arXiv preprint arXiv:2607.01299
    [paper] [code]

  4. [preprint] Akashic: A Low-Overhead LLM Inference Service with MemAttention
    Yang Liu, Zhaokai Luo, Huayi Jin, Ruozhou He, Chenchen Hong, Zhiyong Wang, Yifei Liu, Yunfei Gu, Chentao Wu, Junhao Hu
    In: arXiv preprint arXiv:2607.05708
    [paper] [code]

  5. [preprint] You Need an Encoder for Native Position-Independent Caching
    Shiju Zhao, Junhao Hu, Jiaqi Zheng, Guihai Chen
    In: arXiv preprint arXiv:2602.01519
    [paper] [code]

  6. [preprint] MPIC: Position-Independent Multimodal Context Caching System for Efficient MLLM Serving
    Shiju Zhao, Junhao Hu, Rongxiao Huang, Jiaqi Zheng, Guihai Chen
    In: arXiv preprint arXiv:2502.01960
    [paper] [code]

  7. [preprint] LLMigrate: Transforming “Lazy” Large Language Models into Efficient Source Code Migrators
    Yuchen Liu, Junhao Hu, Yingdi Shan, Ge Li, Yanzhen Zou, Yihong Dong, Tao Xie
    In: arXiv preprint arXiv:2503.23791
    [paper] [code]