biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[ROCm][AITER] fix wrong argument passed to AITER flash_attn_varlen_func (#31880)

Browse Source 
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
This commit is contained in:
vllmellm
2026-01-07 12:25:03 +01:00
committed by GitHub
parent d111bc53ad
commit 41cfa50632
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/v1/attention/backends/mla/aiter_triton_mla.py
View File

@@ -54,7 +54,7 @@ class AiterTritonMLAImpl(AiterMLAImpl):
k, k,
v, v,
softmax_scale=softmax_scale, softmax_scale=softmax_scale,
return_softmax_lse=return_softmax_lse, return_lse=return_softmax_lse,
**kwargs, **kwargs,
) )
# Transpose the LSE if Triton MHA is used: # Transpose the LSE if Triton MHA is used:

2
vllm/v1/attention/backends/mla/rocm_aiter_mla.py
View File

@@ -236,7 +236,7 @@ class AiterMLAImpl(MLACommonImpl[AiterMLAMetadata]):
k=k, k=k,
v=v, v=v,
softmax_scale=softmax_scale, softmax_scale=softmax_scale,
return_softmax_lse=return_softmax_lse, return_lse=return_softmax_lse,
**kwargs, **kwargs,
) )