Class-conditional Image Generation on ImageNet 512x512 (val test)
1.21FIDEDM2-L + DDO
Evaluation Results
| Method | Links | |
|---|---|---|
| EDM2-L + DDOType=Ours, #Params=777M, NFE=63, Guidance=Autoguidance (AG)2025.03 | 1.21 | |
| EDM2-L + DDO + DPM-Solver-v3Type=Ours, #Params=777M, NFE=25, Guidance=Autoguidance (AG)2025.03 | 1.21 | |
| EDM2-XXL + AGType=Diffusion, #Params=1.5B, NFE=63, Guidance=Autoguidance (AG)2025.03 | 1.25 | |
| EDM2-L + DDOType=Ours, #Params=777M, NFE=63, Guidance=None2025.03 | 1.26 | |
| EDM2-L + DDO + DPM-Solver-v3Type=Ours, #Params=777M, NFE=25, Guidance=None2025.03 | 1.29 | |
| EDM2-S + AGType=Diffusion, #Params=280M, NFE=63, Guidance=Autoguidance (AG)2025.03 | 1.34 | |
| SiDType=GAN, #Params=1.5B, NFE=1, Guidance=None2025.03 | 1.37 | |
| USIT-2BType=Diffusion, #Params=2B, NFE=512, Guidance=CFG2025.03 | 1.72 | |
| MAR-LType=Masked, #Params=481M, NFE=1024, Guidance=CFG2025.03 | 1.73 | |
| EDM2-L (retested)Type=Ours/Baseline, #Params=777M, NFE=63, Guidance=CFG2025.03 | 1.77 | |
| EDM2-XXLType=Diffusion, #Params=1.5B, NFE=63, Guidance=CFG2025.03 | 1.81 | |
| EDM2-LType=Diffusion, #Params=777M, NFE=63, Guidance=CFG2025.03 | 1.88 | |
| EDM2-XXLType=Diffusion, #Params=1.5B, NFE=63, Guidance=None2025.03 | 1.91 | |
| MAGVIT-v2Type=Masked, #Params=307M, NFE=64, Guidance=CFG2025.03 | 1.91 | |
| EDM2-L (retested)Type=Ours/Baseline, #Params=777M, NFE=63, Guidance=None2025.03 | 1.96 | |
| EDM2-LType=Diffusion, #Params=777M, NFE=63, Guidance=None2025.03 | 2.06 | |
| SiT-XL + REPAType=Diffusion, #Params=675M, NFE=250, Guidance=CFG2025.03 | 2.08 | |
| EDM2-SType=Diffusion, #Params=280M, NFE=63, Guidance=CFG2025.03 | 2.23 | |
| StyleGAN-XLType=GAN, #Params=168M, NFE=1, Guidance=None2025.03 | 2.41 | |
| EDM2-SType=Diffusion, #Params=280M, NFE=63, Guidance=None2025.03 | 2.56 | |
| SiT-XLType=Diffusion, #Params=675M, NFE=250, Guidance=CFG2025.03 | 2.62 | |
| VAR-d36-sType=AR, #Params=2.3B, NFE=10, Guidance=CFG2025.03 | 2.63 | |
| VDM++Type=Diffusion, #Params=2B, NFE=512, Guidance=CFG2025.03 | 2.65 | |
| MAR-LType=Masked, #Params=481M, NFE=1024, Guidance=None2025.03 | 2.74 | |
| USIT-2BType=Diffusion, #Params=2B, NFE=512, Guidance=None2025.03 | 2.9 | |
| EDM2-XSType=Diffusion, #Params=125M, NFE=63, Guidance=CFG2025.03 | 2.91 | |
| VDM++Type=Diffusion, #Params=2B, NFE=512, Guidance=None2025.03 | 2.99 | |
| U-ViT, LType=Diffusion, #Params=2B, NFE=512, Guidance=CFG2025.03 | 3.02 | |
| DiT-XL/2Type=Diffusion, #Params=675M, NFE=250, Guidance=CFG2025.03 | 3.04 | |
| MAGVIT-v2Type=Masked, #Params=307M, NFE=64, Guidance=None2025.03 | 3.07 | |
| EDM2-XSType=Diffusion, #Params=125M, NFE=63, Guidance=None2025.03 | 3.53 | |
| U-ViT, LType=Diffusion, #Params=2B, NFE=512, Guidance=None2025.03 | 3.54 | |
| ADM-UType=Diffusion, #Params=731M, NFE=500, Guidance=Classifier Guidance2025.03 | 3.85 | |
| RINType=Diffusion, #Params=410M, NFE=1000, Guidance=None2025.03 | 3.95 | |
| MaskGITType=Masked, #Params=227M, NFE=12, Guidance=None2025.03 | 7.32 | |
| ADMType=Diffusion, #Params=559M, NFE=250, Guidance=CFG2025.03 | 7.72 | |
| BigGAN-deepType=GAN, #Params=112M, NFE=1, Guidance=None2025.03 | 8.43 | |
| ADM-UType=Diffusion, #Params=731M, NFE=500, Guidance=None2025.03 | 9.96 | |
| DiT-XL/2Type=Diffusion, #Params=675M, NFE=250, Guidance=None2025.03 | 12.03 | |
| ADMType=Diffusion, #Params=559M, NFE=250, Guidance=None2025.03 | 23.24 |