chunked/nltk_chunking/_training/chunk_58.txt

[~datasets.Dataset.to_tf_dataset]: This method is more low-level, and is useful when you want to exactly control how
your dataset is created, by specifying exactly which columns and label_cols to include.