How to Use the TFT Model for Binary Classification on Time Series Data

[Hspix]

Hello,

I would like to know about how to configure the Temporal Fusion Transformer (TFT) model for training on a binary classification task over time series data. It's important to note that in my dataset, the classification label Y is not part of the input features X.

Based on the example provided in the official documentation, would simply replacing QuantileLoss with CrossEntropy as the loss function suffice for this purpose? If there are any other adjustments needed, could you please advise?

Thank you very much for your assistance!