How to properly use covariates?

[Riofd]

Thank you for your great work. I have a problem about the zero-shot forecast with covariates. My task is electricity forecasting with weather forecasting data. So I create dataset with feat_dynamic_real like ds = PandasDataset(dataframes=df, target="power", timestamp='time', freq="15min", feat_dynamic_real=['temperature', 'humidity', 'rainfall']). However, I get worse performance when adding these covariates. (The first image is without covariates, and the second is with them.) I wonder know if there are some special settings when forecasting with covariates?