drbh's picture
drbh HF Staff
Upload folder using huggingface_hub
c415961 verified
raw
history blame
781 Bytes
# /// script
# requires-python = ">=3.10"
# dependencies = [
# "numpy",
# "torch==2.8.0",
# "kernels-benchmark-tools",
# "xformers",
# ]
#
# [tool.uv.sources]
# kernels-benchmark-tools = { path = "../../../../../tools", editable = true }
# ///
import torch
import sys
from kernels_benchmark_tools import KernelTypeEnum, run_benchmark
import xformers.ops as xops
def xformers_attention(q, k, v):
"""xFormers memory efficient attention"""
# xFormers expects [batch, seq_len, heads, head_dim]
return xops.memory_efficient_attention(q, k, v)
run_benchmark(
kernel_type=KernelTypeEnum.ATTENTION,
impl_name="xformers_meff",
impl_tags={"family": "xformers", "backend": "memory_efficient", "compile": "none"},
impl_func=xformers_attention,
)