Causal model results on Valentini

[VenkateshParvathala]

Hi,
I have a question on how the causal model is evaluated on Valentini. Is it true streaming evaluation or just the evaluation (not true streaming) on the model trained without Bidirectional LSTM ? As the causal pretrained model on Valentini was not provided, I could not try it by myself and see it. From the code, I understood that the convolutions were not changed to causal(By appropriately prepending the zeros to the input) when training the causal model. I am just wondering if this model is a true causal model if the convolutions are still not causal.