chunked/nltk_chunking/model_doc_beit/chunk_1.txt

Inspired by BERT, BEiT is the first paper that makes self-supervised pre-training of
Vision Transformers (ViTs) outperform supervised pre-training.