vipshop/ffpa-attn
Kernel Library for Large Headdim Attention (64~1024, BF16/FP8/FP4), 1.5x~15x↑ vs PyTorch SDPA.
A star delta has been computed for 61,707 of 62,640 tracked repositories in this period, and only those can be ranked in it.
A repository missing here has not been measured far enough back yet. That does not mean it did not move.
Star counts measured between 08:02 and 02:22 (BKK).
Kernel Library for Large Headdim Attention (64~1024, BF16/FP8/FP4), 1.5x~15x↑ vs PyTorch SDPA.
cuDNN Frontend is NVIDIA's modern, open-source entry point to the cuDNN library and a growing collection of high-performance open-source kernels.
Trainable fast and memory-efficient sparse attention
That's all 3 repositories in this period.