Hey, it’s mentioned in this paper that N-BEATS could do zero-shot forecasting. Is this implemented in neural forecast?

Hi @Cyril de Catheu. Yes! We have a tutorial on Transfer Learning here: https://nixtla.github.io/neuralforecast/examples/transfer_learning.html. You can pretrain a model and then use it on any other dataset without further changes.

awesome thanks!