Filip

I think that you can pass the categorical variables as a target and define the loss function to be cross entropy.

have you tried like this? Somebody posted this in another issue: ``` for idx in range(180,200): best_model.plot_prediction(x, raw_predictions, idx=idx, add_loss_to_title=True, quantiles_kwargs={'use_metric': False}, prediction_kwargs={'use_metric': False}) ```

I have the same problem on my data. And when changing the learning rate, the outputs are constantly the same, no matter the input data.

https://github.com/muandet-lab/ExplainingGaussianProcess