Skip to content

[BugFix][VL] Fix FA selection on Qwen2.5-VL #27790

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
yeqcharlotte merged 2 commits into from
Oct 30, 2025
Merged

[BugFix][VL] Fix FA selection on Qwen2.5-VL #27790

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
cb8da64
update test
zhewenl Oct 30, 2025
9a8de51
Merge branch 'main' into fix-amd-ci-entrypoint-mm
ywang96 Oct 30, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion .buildkite/test-amd.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -318,7 +318,7 @@ steps:

- label: V1 Test entrypoints # 35min
timeout_in_minutes: 50
mirror_hardwares: [amdexperimental]
mirror_hardwares: [amdexperimental, amdproduction]
agent_pool: mi325_1
# grade: Blocking
source_file_dependencies:
Expand Down
30 changes: 19 additions & 11 deletions vllm/model_executor/models/qwen2_5_vl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,10 +43,7 @@
)

from vllm.attention.backends.registry import _Backend
from vllm.attention.layer import (
check_upstream_fa_availability,
maybe_get_vit_flash_attn_backend,
)
from vllm.attention.layer import maybe_get_vit_flash_attn_backend
from vllm.attention.ops.vit_attn_wrappers import (
vit_flash_attn_wrapper,
vit_xformers_attn_wrapper,
Expand Down Expand Up @@ -318,6 +315,7 @@ def __init__(
use_data_parallel: bool = False,
attn_backend: _Backend = _Backend.TORCH_SDPA,
use_upstream_fa: bool = False,
attn_backend_override: _Backend | None = None,
) -> None:
super().__init__()
# Per attention head and per partition values.
Expand Down Expand Up @@ -358,8 +356,14 @@ def __init__(
maybe_get_vit_flash_attn_backend(
self.attn_backend,
self.use_upstream_fa,
attn_backend_override=attn_backend_override,
)
)
# On ROCm with FLASH_ATTN backend, upstream flash_attn is used
from vllm.platforms import current_platform

if current_platform.is_rocm() and self.attn_backend == _Backend.FLASH_ATTN:
self.use_upstream_fa = True
self.is_flash_attn_backend = self.attn_backend in {
_Backend.FLASH_ATTN,
_Backend.ROCM_AITER_FA,
Expand Down Expand Up @@ -484,6 +488,7 @@ def __init__(
use_data_parallel: bool = False,
attn_backend: _Backend = _Backend.TORCH_SDPA,
use_upstream_fa: bool = False,
attn_backend_override: _Backend | None = None,
) -> None:
super().__init__()
if norm_layer is None:
Expand All @@ -499,6 +504,7 @@ def __init__(
use_data_parallel=use_data_parallel,
attn_backend=attn_backend,
use_upstream_fa=use_upstream_fa,
attn_backend_override=attn_backend_override,
)
self.mlp = Qwen2_5_VisionMLP(
dim,
Expand Down Expand Up @@ -698,13 +704,14 @@ def __init__(
dtype=torch.get_default_dtype(),
attn_backend_override=attn_backend_override,
)
if (
self.attn_backend != _Backend.FLASH_ATTN
and self.attn_backend != _Backend.ROCM_AITER_FA
and check_upstream_fa_availability(torch.get_default_dtype())
):
self.attn_backend = _Backend.FLASH_ATTN
use_upstream_fa = True

self.attn_backend, self.flash_attn_varlen_func = (
maybe_get_vit_flash_attn_backend(
self.attn_backend,
use_upstream_fa,
attn_backend_override=attn_backend_override,
)
)

if self.attn_backend not in {
_Backend.FLASH_ATTN,
Expand All @@ -730,6 +737,7 @@ def __init__(
use_data_parallel=use_data_parallel,
attn_backend=self.attn_backend,
use_upstream_fa=use_upstream_fa,
attn_backend_override=attn_backend_override,
)
for layer_idx in range(depth)
]
Expand Down
Loading