Why future values of covariates are not used during training?

[doitslow]

For covariates informed tasks, why not include future values of covariates for training?

[abdulfatir]

The model is trained with future_covariates, if the keys are provided. The actual values inside future_covariates cannot be used because the corresponding target values are not present for loss computation.