Skip to content
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

Fix the size of int and bool type when computing module size #2411

Merged
merged 1 commit into from
Feb 2, 2024
Merged

Fix the size of int and bool type when computing module size #2411

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
4 changes: 4 additions & 0 deletions src/accelerate/utils/modeling.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -696,6 +696,10 @@ def compute_module_sizes(
size = tensor.numel() * special_dtypes_size[name]
elif dtype is None:
size = tensor.numel() * dtype_byte_size(tensor.dtype)
elif str(tensor.dtype).startswith(("torch.uint", "torch.int", "torch.bool")):
# According to the code in set_module_tensor_to_device, these types won't be converted
# so use their original size here
size = tensor.numel() * dtype_byte_size(tensor.dtype)
else:
size = tensor.numel() * min(dtype_size, dtype_byte_size(tensor.dtype))
name_parts = name.split(".")
Expand Down
Loading