Springenberg, et al.. Image-to-text Transduction with Spatial Self-attention. 2018.