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

[Attention] FlashAttn MLA (#14258)

Browse Source 
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Signed-off-by: Matthew Bonanni <mbonanni001@gmail.com>
Co-authored-by: Matthew Bonanni <mbonanni001@gmail.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
This commit is contained in:
Lucas Wilkinson
2025-09-04 05:47:59 -04:00
committed by GitHub
parent 2c301ee2eb
commit 402759d472
22 changed files with 480 additions and 200 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
vllm/envs.py
View File

@@ -463,6 +463,7 @@ environment_variables: dict[str, Callable[[], Any]] = {
# - "ROCM_FLASH": use ROCmFlashAttention
# - "FLASHINFER": use flashinfer
# - "FLASHMLA": use FlashMLA
# - "FLASH_ATTN_MLA": use FlashAttention for MLA
"VLLM_ATTENTION_BACKEND":
lambda: os.getenv("VLLM_ATTENTION_BACKEND", None),