Is there any known contraindication to using the "revin" scaler in combination with the Temporal Fusion Transformer (TFT)?

Hi @Manuel. No that we are aware of, we have been using other types of similar scaling like robust or standard and the performance is great

@Cristian (Nixtla) Ok, thank you! I currently use the standard scaler and now that revin has been added to neuralforecast I wanted to try to replace the standard scaler with it.

Yes, makes sense. Let us know how it works for you 🙂

Hey @Manuel Chabier Escolá Pérez, I am working on a minor update to the numerical stability of the revin scaler I hope to have it soon in the main branch: https://github.com/Nixtla/neuralforecast/pull/781