Error when combining MultiOutputRegressor(AutoML()) and specifying X_val, y_val parameters in multi-task (regression) scenario using FLAML framework

[BabelTower]

I am using the FLAML framework for multi-task regression and attempting to combine MultiOutputRegressor(AutoML()) with specified X_val and y_val parameters for model training.

[sonichi]

Specifying X_val and y_val won't work when using MultiOutputRegressor. Workaround: combine training and validation data as X_train and y_train, and then use a custom splitter to split the data as you wanted. See https://microsoft.github.io/FLAML/docs/Use-Cases/Task-Oriented-AutoML#data-split-method