Skip to content

Slightly improve performance in engines/utils.py #8747

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
benediktjohannes wants to merge 1 commit into
base: dev
Choose a base branch
from
+3 −3
Open

Slightly improve performance in engines/utils.py #8747

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
6 changes: 3 additions & 3 deletions monai/engines/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -219,8 +219,8 @@ def __call__(
`kwargs` supports other args for `Tensor.to()` API.
"""
image, label = default_prepare_batch(batchdata, device, non_blocking, **kwargs)
args_ = list()
kwargs_ = dict()
args_ = []
Copy link
Member

@ericspod ericspod Feb 27, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Here this might make a slight improvement since this method is called a lot, but I would suggest the following would be more efficient still (You'd have to test this):

def __call__(
    self,
    batchdata: dict[str, torch.Tensor],
    device: str | torch.device | None = None,
    non_blocking: bool = False,
    **kwargs: Any,
) -> tuple[torch.Tensor, torch.Tensor, tuple, dict]:
    """
    Args `batchdata`, `device`, `non_blocking` refer to the ignite API:
    https://pytorch.org/ignite/v0.4.8/generated/ignite.engine.create_supervised_trainer.html.
    `kwargs` supports other args for `Tensor.to()` API.
    """
    image, label = default_prepare_batch(batchdata, device, non_blocking, **kwargs)
    args_ = ()
    kwargs_ = {}

    def _get_data(key: str) -> torch.Tensor:
        data = batchdata[key]

        if isinstance(data, torch.Tensor):
            data = data.to(device=device, non_blocking=non_blocking, **kwargs)

        return data

    if isinstance(self.extra_keys, (str, list, tuple)):
        args_ = tuple(_get_data(k) for k in ensure_tuple(self.extra_keys))
    elif isinstance(self.extra_keys, dict):
        kwargs_ = {k: _get_data(v) for k, v in self.extra_keys.items()}

    return cast(torch.Tensor, image), cast(torch.Tensor, label), args_, kwargs_

kwargs_ = {}

def _get_data(key: str) -> torch.Tensor:
data = batchdata[key]
Expand All @@ -235,7 +235,7 @@ def _get_data(key: str) -> torch.Tensor:
args_.append(_get_data(k))
elif isinstance(self.extra_keys, dict):
for k, v in self.extra_keys.items():
kwargs_.update({k: _get_data(v)})
kwargs_[k] = _get_data(v)

return cast(torch.Tensor, image), cast(torch.Tensor, label), tuple(args_), kwargs_

Expand Down
Loading