作者位次 #3/9ACL 2026Poster· Natural Language Processing
C u B ridge: An LLM -Based Framework for Understanding and Reconstructing High-Performance Attention Kernels
Xing Ma · Yangjie Zhou · Wu Sun · Zihan Liu · Jingwen Leng · Yun Lin · Shixuan Sun · Minyi Guo · Jin Song Dong
摘要
Efficient CUDA implementations of attention mechanisms are critical to modern deep learning systems, yet supporting diverse and evolving attention variants remains challenging. Existing frameworks and compilers trade performance for flexibility, while expert-written kernels achieve high efficienc...