Skip to content

GitLab

Unverified Commit a4730c1b authored by Chaojun Zhang's avatar Chaojun Zhang Committed by GitHub
Browse files

[XPU]Fix crash due to removed VLLM_USE_V1 attribute (#28520) 


Signed-off-by: default avatarchaojun-zhang <chaojun.zhang@intel.com>
parent d3ade61e
Hide whitespace changes
Inline Side-by-side
Showing with 3 additions and 2 deletions
vllm/platforms/xpu.py
View file @ a4730c1b
......@@ -65,7 +65,6 @@ class XPUPlatform(Platform):
if use_sparse:
raise NotImplementedError("Sparse Attention is not supported on XPU.")
use_v1 = envs.VLLM_USE_V1
if not use_v1:
raise ValueError("XPU backend only supports V1.")
if selected_backend == AttentionBackendEnum.TRITON_ATTN:
......@@ -115,7 +114,9 @@ class XPUPlatform(Platform):
@classmethod
def get_vit_attn_backend(
cls, head_size: int, dtype: torch.dtype
) -> AttentionBackendEnum:
) -> "AttentionBackendEnum":
from vllm.attention.backends.registry import AttentionBackendEnum
return AttentionBackendEnum.FLASH_ATTN
@classmethod
......
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