Skip to content

[Bugfix] Initialize attention bias on the same device as Query/Key/Value for QwenVL Series #14031

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
merged 6 commits into from
Feb 28, 2025
+4 −2
Merged

[Bugfix] Initialize attention bias on the same device as Query/Key/Value for QwenVL Series #14031

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
db48588
[Bugfix] Initialize attention bias on the same device as Query/Key/Va…
LouieYang Feb 28, 2025
dfa2cc9
Update qwen2_5_vl.py
LouieYang Feb 28, 2025
57dfd83
Initialize attention bias on the same device
LouieYang Feb 28, 2025
2cd9f3f
Initialize attention bias on the same device
LouieYang Feb 28, 2025
89225b8
Update qwen2_5_vl.py
LouieYang Feb 28, 2025
efd8280
Update qwen2_vl.py
LouieYang Feb 28, 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 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 @@ -323,7 +323,7 @@ def forward(

seqlens = (cu_seqlens[1:] - cu_seqlens[:-1]).tolist()
attn_bias = BlockDiagonalMask.from_seqlens(q_seqlen=seqlens,
kv_seqlen=None)
kv_seqlen=None, device=q.device)

context_layer = xops.memory_efficient_attention_forward(
q, k, v, attn_bias=attn_bias, p=0, scale=None)
Expand Down
2 changes: 1 addition & 1 deletion vllm/model_executor/models/qwen2_vl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -367,7 +367,7 @@ def forward(

seqlens = (cu_seqlens[1:] - cu_seqlens[:-1]).tolist()
attn_bias = BlockDiagonalMask.from_seqlens(q_seqlen=seqlens,
kv_seqlen=None)
kv_seqlen=None, device=q.device)

context_layer = xops.memory_efficient_attention_forward(
q, k, v, attn_bias=attn_bias, p=0, scale=None)
Expand Down
Loading