Skip to content

Fix(backend/torch): Resolved MPS broadcast crash in binary_crossentropy #21816

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

Merged
merged 1 commit into from
Nov 3, 2025
Merged

Fix(backend/torch): Resolved MPS broadcast crash in binary_crossentropy #21816

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
14 changes: 14 additions & 0 deletions keras/src/backend/torch/nn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -755,12 +755,26 @@ def binary_crossentropy(target, output, from_logits=False):
target = convert_to_tensor(target)
output = convert_to_tensor(output)

# We only apply the squeeze fix if we are on an MPS device,
# as this change breaks tests on other platforms that
# expect the original tensor shape to be preserved.
if (
torch.backends.mps.is_available()
and target.ndim > 1
and output.ndim == target.ndim
and target.shape[-1] == 1
and output.shape[-1] == 1
):
target = torch.squeeze(target, -1).contiguous()
output = torch.squeeze(output, -1).contiguous()

if target.shape != output.shape:
raise ValueError(
"Arguments `target` and `output` must have the same shape. "
"Received: "
f"target.shape={target.shape}, output.shape={output.shape}"
)

# By default, PyTorch, does reduction of `sum` over all rows,
# change reduction to `none` to keep dim
if from_logits:
Expand Down