xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-07-09 22:57:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] Add missing VIT backend dispatch on CPU (#24623)

Browse Source 
Signed-off-by: jiang1.li <jiang1.li@intel.com>
This commit is contained in:
Li, Jiang 2025-09-11 13:28:41 +08:00 committed by GitHub
parent f17a6aa4ec
commit 29799ddacc
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
vllm/attention/layer.py
View File

@ -399,7 +399,8 @@ class MultiHeadAttention(nn.Module):
key, key,
value, value,
scale=self.scale) scale=self.scale)
elif self.attn_backend == _Backend.TORCH_SDPA: elif (self.attn_backend == _Backend.TORCH_SDPA
or self.attn_backend == _Backend.TORCH_SDPA_VLLM_V1):
query, key, value = (x.transpose(1, 2) query, key, value = (x.transpose(1, 2)
for x in (query, key, value)) for x in (query, key, value))
out = F.scaled_dot_product_attention(query, out = F.scaled_dot_product_attention(query,