Learning to Collocate Neural Modules for Image Captioning
2019·Arxiv