Skip to content

handle weight sharing with init_on_device #2737

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

handle weight sharing with init_on_device #2737

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
9 changes: 8 additions & 1 deletion src/accelerate/big_modeling.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -128,7 +128,14 @@ def register_empty_parameter(module, name, param):
param_cls = type(module._parameters[name])
kwargs = module._parameters[name].__dict__
kwargs["requires_grad"] = param.requires_grad
module._parameters[name] = param_cls(module._parameters[name].to(device), **kwargs)
# When we have a case of tensor2 = tensor1, it would call the set_attr
# of param, which in turn would call the register_parameter API.
# In this case, the new param is already on meta-device, since it was moved
# previously when it was initialized. Hence, when resetting, you can
# directly assign that tensor instead of re-init. If you re-init you would
# lose the relationship.
module._parameters[name] = param if param.device == device else \
param_cls(module._parameters[name].to(device), **kwargs)

def register_empty_buffer(module, name, buffer, persistent=True):
old_register_buffer(module, name, buffer, persistent=persistent)
Expand Down