Skip to content

[Bugfix] Initialize attention bias on the same device as Query/Key/Value #13468

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 1 commit into from
Feb 25, 2025
Merged

[Bugfix] Initialize attention bias on the same device as Query/Key/Value #13468

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
10 changes: 6 additions & 4 deletions vllm/attention/backends/xformers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -673,7 +673,9 @@ def _run_memory_efficient_xformers_forward(

# Cross-attention mask is non-causal
attn_bias = BlockDiagonalMask.from_seqlens(
attn_metadata.seq_lens, attn_metadata.encoder_seq_lens)
attn_metadata.seq_lens,
attn_metadata.encoder_seq_lens,
device=query.device)

# Encoder branch of encoder-decoder model uses
# attn_metadata.encoder_seq_lens
Expand All @@ -683,7 +685,7 @@ def _run_memory_efficient_xformers_forward(

# Encoder self-attention mask is non-causal
attn_bias = BlockDiagonalMask.from_seqlens(
attn_metadata.encoder_seq_lens)
attn_metadata.encoder_seq_lens, device=query.device)

# Self-attention block of encoder-only model just
# uses the seq_lens directly.
Expand All @@ -692,7 +694,7 @@ def _run_memory_efficient_xformers_forward(

# Encoder self-attention mask is non-causal
attn_bias = BlockDiagonalMask.from_seqlens(
attn_metadata.seq_lens)
attn_metadata.seq_lens, device=query.device)

# Self-attention block of decoder branch just
# uses the seq_lens directly
Expand All @@ -701,7 +703,7 @@ def _run_memory_efficient_xformers_forward(

# Decoder self-attention mask is causal
attn_bias = BlockDiagonalCausalMask.from_seqlens(
attn_metadata.seq_lens)
attn_metadata.seq_lens, device=query.device)
else:
raise ValueError("Unknown AttentionType: %s", attn_type)

Expand Down