Commit 1a493a24 authored by zhuwenwen's avatar zhuwenwen
Browse files

remove xops

parent fc7db442
......@@ -4,14 +4,16 @@ import time
import pytest
import torch
from xformers import ops as xops
from xformers.ops.fmha.attn_bias import BlockDiagonalCausalFromBottomRightMask
from vllm.attention.backends.xformers import _make_alibi_bias
from vllm.attention.ops.prefix_prefill import context_attention_fwd
from vllm.utils import is_hip
from vllm.utils import STR_DTYPE_TO_TORCH_DTYPE, seed_everything
if not is_hip():
from xformers import ops as xops
from xformers.ops.fmha.attn_bias import BlockDiagonalCausalFromBottomRightMask
from vllm.attention.backends.xformers import _make_alibi_bias
NUM_HEADS = [64]
NUM_QUERIES_PER_KV = [1, 8, 64]
HEAD_SIZES = [128, 96, 24]
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment