ResearchBenchmarksText-to-Image Generation on Attend-Excite promptsText-to-Image Generation on Attend-Excite promptsFollow88.17Accuracy (Animals)CO334.6148.51562.4276.325Sep 30, 2025Evaluation resultsMethodMethodLinksAccuracy (Animals)Accuracy (Animals-Objects)Accuracy (Objects)CO3Backbone=SDXLBackbone=SDXL2025.0988.1774.6682.5SDXL2025.0952.6660.570.83ToMeBackbone=SDXLBackbone=SDXL2025.0943.8360.8361.66MagnetBackbone=SDXLBackbone=SDXL2025.093756.8368.49SynGenBackbone=SDXLBackbone=SDXL2025.0936.6743.6671.83