Input Feature Labels

The MaxHistoryTrackerFeaturizer featurizer for instance creates a feature vector (num_unique_turns, max_history, num_input_features)

  1. How to know what each value of num_input_features dimension represents? eg: X[0,0,1] represents certain entity label? Y[0] represents certain action label?

  2. Does num_unique_turns mean number of examples in training dataset?