COMIC: Towards A Compact Image Captioning Model with Attention
2019·Arxiv