1 paper
Bertram Taetz, Gal Bordelius
Generating accurate and coherent image captions in a continual learning setting remains a major challenge due to catastrophic forgetting and the difficulty of aligning evolving vis…