Skip to content

hooks/pyramid_attention_broadcast: remove redundant iteration==0 guard and fix stale cache VRAM leak #13497

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

Open
GitGlimpse895 wants to merge 1 commit into
base: main
Choose a base branch
from
+1 −2
Open

hooks/pyramid_attention_broadcast: remove redundant iteration==0 guard and fix stale cache VRAM leak #13497

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
3 changes: 1 addition & 2 deletions src/diffusers/hooks/pyramid_attention_broadcast.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,7 +159,6 @@ def new_forward(self, module: torch.nn.Module, *args, **kwargs) -> Any:
)
should_compute_attention = (
self.state.cache is None
or self.state.iteration == 0
or not is_within_timestep_range
or self.state.iteration % self.block_skip_range == 0
)
Expand All @@ -169,7 +168,7 @@ def new_forward(self, module: torch.nn.Module, *args, **kwargs) -> Any:
else:
output = self.state.cache

self.state.cache = output
self.state.cache = output if is_within_timestep_range else None
self.state.iteration += 1
return output

Expand Down
Loading