Diffusion Models Beat GANs on Image Synthesis

Open paper
Benchmark Model Rank Results
conditional-image-generation-on-imagenetADM-G (classifier_scale=0.5)#5FID: 2.97
conditional-image-generation-on-imagenet-2ADM-G#3FID: 4.59Inception score: 186.7
image-generation-on-imagenet-128x128ADM-G#8FID: 2.97
image-generation-on-imagenet-256x256ADM-G, ADM-U#72FID: 3.94
image-generation-on-imagenet-256x256ADM-G#79FID: 4.59
image-generation-on-imagenet-512x512ADM-G, ADM-U#43FID: 3.85Inception score: 221.72
image-generation-on-imagenet-512x512ADM-G#48FID: 7.72Inception score: 172.71
image-generation-on-imagenet-64x64ADM (dropout)#14FID: 2.07
image-generation-on-lsun-bedroom-256-x-256ADM (dropout)#4FID: 1.90
image-generation-on-lsun-bedroom-256-x-256ADM (dropout, DINOv2)#16FD: 59.64Precision: 0.85Recall: 0.75
image-generation-on-lsun-cat-256-x-256ADM (dropout)#3FID: 5.57
image-generation-on-lsun-horse-256-x-256ADM (dropout)#3FID: 2.57

Rank counts only results with a code link.