flash-attn2-rs / README.md
drbh's picture
drbh HF Staff
Uploaded using `kernel-builder`.
3042848 verified
|
Raw History Blame Contribute Delete
1.01 kB
metadata
library_name: kernels
license: apache-2.0

This is the repository card of kernels-community-rs/flash-attn2-rs 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-rs/flash-attn2-rs", version=1)
fwd_kvcache = kernel_module.fwd_kvcache

fwd_kvcache(...)

Available functions

  • fwd_kvcache
  • flash_attn_with_kvcache
  • fwd
  • bwd
  • varlen_fwd
  • varlen_bwd
  • fwd_explicit
  • reserve_rng_state
  • flash_attn_func
  • flash_attn_qkvpacked_func
  • flash_attn_kvpacked_func
  • flash_attn_varlen_func
  • flash_attn_varlen_qkvpacked_func
  • flash_attn_varlen_kvpacked_func

Benchmarks

Benchmarking script is available for this kernel. Run kernels benchmark kernels-community-rs/flash-attn2-rs --version 1.