File tree Expand file tree Collapse file tree 1 file changed +0
-8
lines changed
vllm/model_executor/models Expand file tree Collapse file tree 1 file changed +0
-8
lines changed Original file line number Diff line number Diff line change 4343)
4444
4545from vllm .attention .backends .registry import _MHA_Backend
46- from vllm .attention .layer import maybe_get_vit_flash_attn_backend
4746from vllm .attention .ops .vit_attn_wrappers import (
4847 vit_flash_attn_wrapper ,
4948 vit_torch_sdpa_wrapper ,
@@ -351,9 +350,6 @@ def __init__(
351350 disable_tp = use_data_parallel ,
352351 )
353352 self .attn_backend = attn_backend
354- self .flash_attn_varlen_func = maybe_get_vit_flash_attn_backend (
355- self .attn_backend ,
356- )
357353
358354 self .is_flash_attn_backend = self .attn_backend in {
359355 _MHA_Backend .FLASH_ATTN ,
@@ -681,10 +677,6 @@ def __init__(
681677 attn_backend_override = attn_backend_override ,
682678 )
683679
684- self .flash_attn_varlen_func = maybe_get_vit_flash_attn_backend (
685- self .attn_backend ,
686- )
687-
688680 if self .attn_backend not in {
689681 _MHA_Backend .FLASH_ATTN ,
690682 _MHA_Backend .VLLM_FLASH_ATTN ,
You can’t perform that action at this time.
0 commit comments