Image Generation on ImageNet-32
4.11FIDNFDM-OT
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| NFDM-OTNFE=122024.04 | 4.11 | — | — | — | — | |
| Continuous FMModeling category=Continuous2025.05 | 5.02 | — | — | — | — | |
| Flow MatchingNFE=1222024.04 | 5.02 | — | — | — | — | |
| NFDM-OTNFE=42024.04 | 6.13 | — | — | — | — | |
| MDM-PrimeModeling category=Discrete, NFE=1,0242025.05 | 6.98 | 11.65 | — | — | — | |
| Multisample Flow MatchingNFE=122024.04 | 7.18 | — | — | — | — | |
| MDMModeling category=Discrete, NFE=1,0242025.05 | 7.91 | 11.6 | — | — | — | |
| MDM-MixtureModeling category=Discrete, NFE=1,0242025.05 | 8.08 | 11.56 | — | — | — | |
| ScoreFlow (VP)Modeling category=Continuous2025.05 | 8.31 | — | — | — | — | |
| Soft Trunc. DDPMModeling category=Continuous2025.05 | 8.42 | 11.82 | — | — | — | |
| Stochastic Interp.Modeling category=Continuous2025.05 | 8.49 | — | — | — | — | |
| ScoreFlow (subVP)Modeling category=Continuous2025.05 | 8.87 | — | — | — | — | |
| i-DODE (VP)Modeling category=Continuous2025.05 | 9.09 | — | — | — | — | |
| NFDM-OTNFE=22024.04 | 9.83 | — | — | — | — | |
| i-DODE (SP)Modeling category=Continuous2025.05 | 10.31 | — | — | — | — | |
| MSGANModeling category=Continuous2025.05 | 12.3 | — | — | — | — | |
| DDPMModeling category=Continuous2025.05 | 16.18 | — | — | — | — | |
| NDMModeling category=Continuous2025.05 | 17.02 | — | — | — | — | |
| Multisample Flow MatchingNFE=42024.04 | 17.28 | — | — | — | — | |
| QC-NCSN++Modeling category=Continuous2025.05 | 19.62 | 9.94 | — | — | — | |
| BOLT-GANTraining Epochs=302025.10 | 34.8 | — | — | — | — | |
| HINGE-GANTraining Epochs=302025.10 | 42.3 | — | — | — | — | |
| WGANTraining Epochs=302025.10 | 58 | — | — | — | — | |
| Axial Transformermodeling_paradigm=Autoregressive2021.10 | — | — | 3.76 | — | — | |
| BigGANBase Architecture=BigGAN2021.12 | — | 13.25 | — | 5.44 | 5.6 | |
| BigGAN+LCSABase Architecture=BigGAN, Module=LCSA2021.12 | — | 13.61 | — | 4.82 | 5.01 | |
| DenseFlow2021.06 | — | — | 3.63 | — | — | |
| DenseFlowmodeling_paradigm=Non-Autoregressive2021.10 | — | — | 3.63 | — | — | |
| Hourglassmodeling_paradigm=Autoregressive2021.10 | — | — | 3.74 | — | — | |
| Image Transformer2021.06 | — | — | 3.77 | — | — | |
| Image Transformermodeling_paradigm=Autoregressive2021.10 | — | — | 3.77 | — | — | |
| NPRF-Transformer with RPERPE=true2021.06 | — | — | 3.68 | — | — | |
| OmniGANBase Architecture=OmniGAN2021.12 | — | 29.55 | — | 3.74 | 4.56 | |
| OmniGAN+LCSABase Architecture=OmniGAN, Module=LCSA2021.12 | — | 30.53 | — | 3.53 | 4.37 | |
| PixelCNNmodeling_paradigm=Autoregressive2021.10 | — | — | 3.83 | — | — | |
| PRF-Transformer2021.06 | — | — | 4.04 | — | — | |
| ScoreFlow2021.06 | — | — | 3.76 | — | — | |
| VDM2021.06 | — | — | 3.72 | — | — | |
| VDMmodeling_paradigm=Non-Autoregressive2021.10 | — | — | 3.72 | — | — |