Multilabel support for classification task

[andibuwono]

hi, just want to ask, whether FLAML will support multi label classification task?

for now flaml will fail for multi label dataset (y is 2 dim) because it flatten the y then check the length of the data ref

FLAML/flaml/automl/task/generic_task.py

Line 128 in 7f42bec

y_train_all = y_train_all.flatten()

some estimator already support the multi label natively.

many thanks

[thinkall]

Hi @andibuwono , thanks for the feedback. We'll support that, but I don't have an ETA for it yet. PRs are super welcome!