We need to communicate about recent progress in array API support in scikit-learn.
@lucyleeow started a draft.
I plan to work on a case study for multiclass classification of images from a pre-trained feature extractor from the timm library and RidgeClassifierCV or LogisticRegression as the final layer.
I was thinking about using a document classification problems from screenshots. I think there are public datasets that are well suited for this task.