Class-conditional Image Generation on ImageNet 64x64 (test)
1.01FIDEDM2*
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| EDM2*NFE=63, Auto-Guidance=true2025.09 | 1.01 | — | — | — | |
| GDD-INFE=12024.05 | 1.16 | 75 | 60 | — | |
| RINNFE=10002024.05 | 1.23 | — | — | — | |
| RINNFE=10002025.09 | 1.23 | — | — | — | |
| AYFNFE=22025.09 | 1.25 | — | — | — | |
| DMD2NFE=12025.09 | 1.28 | — | — | — | |
| CMT (w/ ECD)NFE=22025.09 | 1.32 | — | — | — | |
| EDM2-XLNFE=632024.06 | 1.33 | — | — | — | |
| EDM2NFE=632025.09 | 1.33 | — | — | — | |
| GDDNFE=12024.05 | 1.42 | 77 | 59 | — | |
| sCTNFE=22025.09 | 1.48 | — | — | — | |
| CMT (w/ ECT)NFE=22025.09 | 1.48 | — | — | — | |
| ECDNFE=22025.09 | 1.5 | — | — | — | |
| StyleGAN2-XLNFE=12024.05 | 1.51 | — | — | — | |
| SiDNFE=1, alpha=1.22024.05 | 1.52 | 74 | 63 | — | |
| SiDNFE=12025.09 | 1.52 | — | — | — | |
| EDMType=Diff/FM2024.12 | 1.55 | — | — | — | |
| Stable CTNFE=22025.09 | 1.55 | — | — | — | |
| TCMNFE=22025.09 | 1.62 | — | — | — | |
| sCDNFE=22025.09 | 1.66 | — | — | — | |
| ECM-XL*Training budget=102.4M images, NFE=22024.06 | 1.67 | — | — | — | |
| ECTNFE=22025.09 | 1.67 | — | — | — | |
| CTMNFE=22024.05 | 1.73 | 74 | 57 | — | |
| CTMNFE=22025.09 | 1.73 | — | — | — | |
| CMT (w/ ECD)NFE=12025.09 | 1.78 | — | — | — | |
| MultiStep-CDNFE=22025.09 | 1.9 | — | — | — | |
| CTMNFE=12024.05 | 1.92 | 73 | 57 | — | |
| CTMNFE=12025.09 | 1.92 | — | — | — | |
| ECM-XLTraining iterations=100k, NFE=22024.06 | 1.96 | — | — | — | |
| SiDNFE=1, alpha=12024.05 | 2.02 | 73 | 63 | — | |
| CMT (w/ ECT)NFE=12025.09 | 2.02 | — | — | — | |
| sCTNFE=12025.09 | 2.04 | — | — | — | |
| ADMNFE=2502024.05 | 2.07 | 74 | 63 | — | |
| ADMNFE=2502024.06 | 2.07 | — | — | — | |
| ADM(dropout)Type=Diff/FM2024.12 | 2.09 | — | — | — | |
| ECM-LTraining iterations=100k, NFE=22024.06 | 2.14 | — | — | — | |
| TCMNFE=12025.09 | 2.2 | — | — | — | |
| EDMNFE=792024.06 | 2.22 | — | — | — | |
| ECDNFE=12025.09 | 2.24 | — | — | — | |
| ECM-MTraining iterations=100k, NFE=22024.06 | 2.35 | — | — | — | |
| Stable CTNFE=12025.09 | 2.42 | — | — | — | |
| sCDNFE=12025.09 | 2.44 | — | — | — | |
| ECM-XL*Training budget=102.4M images, NFE=12024.06 | 2.49 | — | — | — | |
| ECTNFE=12025.09 | 2.49 | — | — | — | |
| Separate CompressionRetention Ratio=1.0x, MACs=107G, Training Overhead=400N2026.04 | 2.57 | — | — | — | |
| DMDNFE=12024.05 | 2.62 | — | — | — | |
| EDMNFE=792024.05 | 2.64 | — | — | — | |
| TARFLOWType=NF, Patch Size (P)=2, Channel Size (Ch)=1024, Flow Blocks (T)=8, Attention Layers (K)=8, Input Noise (pe)=N(0, 0.05^2)2024.12 | 2.66 | — | — | — | |
| iCT-deepNFE=22024.05 | 2.77 | 74 | 62 | — | |
| iCT-deepNFE=22024.06 | 2.77 | — | — | — | |
| iCT-deepNFE=22025.09 | 2.77 | — | — | — | |
| ECM-S*Training budget=102.4M images, NFE=22024.06 | 2.79 | — | — | — | |
| TARFLOWType=NF, Patch Size (P)=2, Channel Size (Ch)=768, Flow Blocks (T)=8, Attention Layers (K)=8, Input Noise (pe)=N(0, 0.05^2)2024.12 | 2.9 | — | — | — | |
| iDDPMType=Diff/FM2024.12 | 2.92 | — | — | — | |
| AYFNFE=12025.09 | 2.98 | — | — | — | |
| VCTNFE=22025.09 | 3.07 | — | — | — | |
| ECM-STraining iterations=100k, NFE=22024.06 | 3.18 | — | — | — | |
| iCTNFE=22024.05 | 3.2 | 73 | 63 | — | |
| iCTNFE=22024.06 | 3.2 | — | — | — | |
| iCTNFE=22025.09 | 3.2 | — | — | — | |
| MultiStep-CDNFE=12025.09 | 3.2 | — | — | — | |
| iCT-deepNFE=12024.05 | 3.25 | 72 | 63 | — | |
| iCT-deepType=CM2024.12 | 3.25 | — | — | — | |
| iCT-deepNFE=12025.09 | 3.25 | — | — | — | |
| OFA-Diffusion CompressionFine-tune=40K, Retention Ratio=1.0x, MACs=107G, Training Overhead=400 + 40N2026.04 | 3.65 | — | — | — | |
| TARFLOWType=NF, Patch Size (P)=4, Channel Size (Ch)=1024, Flow Blocks (T)=8, Attention Layers (K)=8, Input Noise (pe)=N(0, 0.05^2)2024.12 | 3.99 | — | — | — | |
| iCTNFE=12024.05 | 4.02 | 70 | 63 | — | |
| iCTNFE=12025.09 | 4.02 | — | — | — | |
| ECM-S*Training budget=102.4M images, NFE=12024.06 | 4.05 | — | — | — | |
| BigGAN-deepNFE=12024.05 | 4.06 | 79 | 48 | — | |
| BigGANType=GAN2024.12 | 4.06 | — | — | — | |
| CDNFE=22024.05 | 4.7 | 69 | 64 | — | |
| CD (LPIPS)NFE=22024.06 | 4.7 | — | — | — | |
| CD(LPIPS)Type=CM2024.12 | 4.7 | — | — | — | |
| CDNFE=22025.09 | 4.7 | — | — | — | |
| VCTNFE=12025.09 | 4.93 | — | — | — | |
| TRACTNFE=22024.06 | 4.97 | — | — | — | |
| Diff-InstructNFE=12024.06 | 5.57 | — | — | — | |
| PD (LPIPS)NFE=22024.06 | 5.74 | — | — | — | |
| CDNFE=12024.05 | 6.2 | 68 | 63 | — | |
| CDNFE=12025.09 | 6.2 | — | — | — | |
| IC-GANType=GAN2024.12 | 6.7 | — | — | — | |
| DFNONFE=12024.05 | 7.83 | — | 61 | — | |
| DFNO (LPIPS)NFE=12024.06 | 7.83 | — | — | — | |
| OFA-Diffusion CompressionFine-tune=40K, Retention Ratio=0.25x, MACs=32.35G, Training Overhead=400 + 40N2026.04 | 8.65 | — | — | — | |
| Separate CompressionRetention Ratio=0.25x, MACs=32.35G, Training Overhead=400N2026.04 | 9.45 | — | — | — | |
| DDPMNFE=2502024.05 | 11 | 67 | 58 | — | |
| OFA-Diffusion CompressionRetention Ratio=0.25x, MACs=32.35G, Training Overhead=4002026.04 | 11.35 | — | — | — | |
| PDNFE=12024.05 | 15.39 | — | — | — | |
| BOOTNFE=12024.05 | 16.3 | 68 | 36 | — | |
| BOOTNFE=12024.06 | 16.3 | — | — | — | |
| Naive OFATraining Overhead=4002026.04 | — | — | — | 8.36 | |
| OFA-Diffusion CompressionTraining Overhead=4002026.04 | — | — | — | 7.43 | |
| OFA-Diffusion CompressionFine-tune=40K, Training Overhead=400 + 40N2026.04 | — | — | — | 5.47 | |
| Separate CompressionTraining Overhead=400N2026.04 | — | — | — | 6.08 |