chunked/nltk_chunking/model_doc_visual_bert/chunk_2.txt

The abstract from the paper is the following:
We propose VisualBERT, a simple and flexible framework for modeling a broad range of vision-and-language tasks.