chunked/nltk_chunking/_training/chunk_57.txt

Because it is a method
on your model, it can inspect the model to automatically figure out which columns are usable as model inputs, and
discard the others to make a simpler, more performant dataset.