chunked/nltk_chunking/model_doc_xlnet/chunk_5.txt

Furthermore, XLNet integrates ideas from Transformer-XL, the state-of-the-art autoregressive model, into
pretraining.