chunked/nltk_chunking/model_doc_t5/chunk_40.txt

One can use [T5ForConditionalGeneration] (or the Tensorflow/Flax variant), which includes the
language modeling head on top of the decoder.