Interactive Attention Model Explorer for Natural Language Processing Tasks with Unbalanced Data Sizes

Conventional attention visualization tools compromise either the readability or the information conveyed when documents are lengthy, especially when these documents have imbalanced sizes. Our work strives toward a more intuitive visualization for a subset of Natural Language Processing tasks, where attention is mapped between documents wi...More