COMIC: Towards A Compact Image Captioning Model with Attention | Read Paper on Bytez