Skip to content

fix(tuners/lora/bnb): normalize output device for CPU-offloaded BnB layers #3181

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
Anai-Guo wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix(tuners/lora/bnb): normalize output device for CPU-offloaded BnB layers #3181

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
2 changes: 2 additions & 0 deletions src/peft/tuners/lora/bnb.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -247,6 +247,7 @@ def forward(self, x: torch.Tensor, *args, **kwargs) -> torch.Tensor:
result = self.base_layer(x, *args, **kwargs)
else:
result = self.base_layer(x, *args, **kwargs)
result = result.to(x.device) # normalize device for CPU-offloaded layers
for active_adapter in self.active_adapters:
if active_adapter not in self.lora_A.keys():
continue
Expand Down Expand Up @@ -521,6 +522,7 @@ def forward(self, x: torch.Tensor, *args, **kwargs) -> torch.Tensor:
result = self.base_layer(x, *args, **kwargs)
else:
result = self.base_layer(x, *args, **kwargs)
result = result.to(x.device) # normalize device for CPU-offloaded layers
# As per Tim Dettmers, for 4bit, we need to defensively clone here.
# The reason is that in some cases, an error can occur that backprop
# does not work on a manipulated view. This issue may be solved with
Expand Down