hey there, for almost a year ive been using legacy torchtext methods to load custom datasets, preprocess them add a custom pretrained embedding and build the dataloader.
in the recent progress in torchtext, can someone please help me out with the latest torchtext preprocessing loading script that work generically for almost any NLP tasks
that contain all of these functionalities, adding custom datasets, adding custom pretrainined embeddings, adding an iterator that sorts sequences based on their length that reduces to minimum padding.