chunked/nltk_chunking/tasks_masked_language_modeling/chunk_2.txt

This'll give you a chance to experiment and make sure everything works before spending more time training on the full dataset.