chunked/nltk_chunking/model_doc_lxmert/chunk_3.txt

The abstract from the paper is the following:
Vision-and-language reasoning requires an understanding of visual concepts, language semantics, and, most importantly,
the alignment and relationships between these two modalities.