Forecast on unseen data(target and dynamic real features) using gluonts(deepar estimator) with use_feat_dynamic_real=True?

[harivilas223]

1.how can I make forecast using last context lenght in my test data set?in that case i don't have either target as well as dynamic real features. But in the given solutions dynamic real feature length should be prediction length*2. How can that possible in future forecasting when my target and dynamic real features generate at same time.

2.when use_feat_dynamic_real=False i am able forecaste in future.

can you suggest suitable solution?