No default loss for KerasClassifier

Why does KerasClassifier not have a default loss?

https://github.com/adriangb/scikeras/blob/d83bffaf8497289d9755f42654bdcc15584e68e6/scikeras/wrappers.py#L1229-L1231

Shouldn't there be a least a sensible default that works in most use cases?

**Reference issues/PRs**
https://github.com/dask/dask-ml/pull/794

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

No default loss for KerasClassifier #206

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

	loss: Union[
	Union[str, tf.keras.losses.Loss, Type[tf.keras.losses.Loss], Callable], None
	] = None,

No default loss for KerasClassifier #206

Description

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions