Skip to content

#36 - Fix to_device() incorrectly casting integer tensors #49

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

#36 - Fix to_device() incorrectly casting integer tensors #49

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
11 changes: 6 additions & 5 deletions src/alpamayo_r1/helper.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,11 +86,12 @@ def to_device(
) -> Any:
"""Recursively cast data into the specified device, dtype."""
if isinstance(data, torch.Tensor):
data = data.to(
device=device,
dtype=dtype,
)
return data
# Only apply dtype conversion to floating-point tensors.
# Integer tensors (e.g., input_ids, attention_mask) must preserve their dtype
# for compatibility with Hugging Face models during mixed-precision inference.
if dtype is not None and data.is_floating_point():
return data.to(device=device, dtype=dtype)
return data.to(device=device)
elif isinstance(data, collections.abc.Mapping):
return {key: to_device(data[key], device=device, dtype=dtype) for key in data}
elif isinstance(data, collections.abc.Sequence) and not isinstance(data, (str, bytes)):
Expand Down