StoryDALL-E: Adapting Pretrained Text-to-Image Transformers for Story Continuation

Benchmark Model Rank Results
story-continuation-on-flintstonessvStoryDALL-E#3FID: 28.37Char-F1: 74.28F-Acc: 52.35
story-continuation-on-flintstonessvStoryDALL-E (Story Embeddings)#4FID: 29.21Char-F1: 72.18F-Acc: 53.28
story-continuation-on-flintstonessvStoryDALL-E (Cross-Attention)#5FID: 35.04Char-F1: 73.94F-Acc: 52.72
story-continuation-on-flintstonessvStoryDALL-E (Story Embeddings + Cross-Attention)#6FID: 36.28Char-F1: 72.44F-Acc: 51.32
story-continuation-on-pororosvStoryDALL-E#3FID: 21.64Char-F1: 40.28F-Acc: 20.94
story-continuation-on-pororosvStoryDALL-E (Cross-Attention)#4FID: 23.27Char-F1: 40.25F-Acc: 18.16
story-continuation-on-pororosvStoryDALL-E (Story Embeddings)#5FID: 30.45Char-F1: 39.32F-Acc: 34.65
story-continuation-on-pororosvStoryDALL-E (Story Embeddings + Cross-Attention)#6FID: 31.68Char-F1: 35.29F-Acc: 16.73