ResearchBenchmarksConditional Image Generation on ImageNet 64x64 (train)Conditional Image Generation on ImageNet 64x64 (train)Follow2.16FIDLEGO-L-PG2.0762.6433.213.777Oct 10, 2023Evaluation resultsMethodMethodLinksFIDLEGO-L-PG#Params=464M, FLOPs=68...#Params=464M, FLOPs=68.8G, Spatial Refinement=Progressive Growth2023.102.16LEGO-L-PR#Params=464M, FLOPs=68...#Params=464M, FLOPs=68.8G, Spatial Refinement=Progressive Refinement2023.102.29DiT-L/2#Params=458M, FLOPs=16...#Params=458M, FLOPs=161.4G2023.102.91U-ViT-L/4#Params=287M, FLOPs=91.2G#Params=287M, FLOPs=91.2G2023.104.26