chunked/nltk_chunking/model_doc_t5/chunk_58.txt

For encoder-decoder models, one
typically defines a max_source_length and max_target_length, which determine the maximum length of the
input and output sequences respectively (otherwise they are truncated).