papers

  1. SPLAT
    splat.jpg
    SPLAT: A framework for optimised GPU code-generation for SParse reguLar ATtention
    Ahan Gupta, Yueming Yuan, Devansh Jain, Yuhao Ge, David Aponte, Yanqi Zhou, Charith Mendis
    We proposed a novel sparse format, ACSR, and a code-generation scheme, SPLAT, to achieve both generality and performance in diverse sparse-MHSA patterns on GPUs, resulting in significant speedups over Triton and TVM.
    Paper
  2. Tooth Segmentation
    tooth.png
    Tooth Defect Segmentation in 3D Mesh Scans Using Deep Learning
    Hao Chen, Yuhao Ge, Jiahao Wei, Huimin Xiong, Zuozhu Liu
    We proposes TripleNet, a novel graph convolutional network, to address the challenge of tooth defect segmentation in 3D mesh scans, achieving state-of-the-art performance and highlighting AI’s potential in digital dentistry.
    Paper