This is the repository card of kernels-community/metal-flash-sdpa that has been pushed on the Hub. It was built to be used with the kernels library. This card was automatically generated.

How to use

# make sure `kernels` is installed: `pip install -U kernels`
from kernels import get_kernel

kernel_module = get_kernel("kernels-community/metal-flash-sdpa", version=2)
flash_attention_varlen = kernel_module.flash_attention_varlen

flash_attention_varlen(...)

Available functions

  • flash_attention_varlen
  • flash_attn_varlen_func
  • ops

Benchmarks

Benchmarking script is available for this kernel. Run kernels benchmark kernels-community/metal-flash-sdpa --version 2.

Downloads last month
-
apache-2.0