Autoxlstm produces NaN

[supriamir]

I would like to ask whether autoxLSTM is currently supported for xLSTM in Nixtla.

I tried running autoxLSTM, but the evaluation results always return NaN values, even though the same dataset works correctly with standard xLSTM training.

Could you please clarify if autoxLSTM is officially supported for xLSTM, and if there are any known limitations or required settings?

Thank you for your assistance.