I work on CNN model for text classification, I use textclassdatabunch class for preprocessing.
Let say that my max sequence length is 19 words so I passed for that method pad_first=True and some other arguments but its not really adding padding word to samples.
So I end up with various sequence lengths and it seems like its not added any padding.
Any suggestions? I can do it manually but seems there is easier way as it should support it.