Text-to-Image Generation on COCO 5,000 images (val)
2.2FIDG-VS
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| G-VSLDM=SD22025.09 | 2.2 | 0.332 | — | 18.5 | 0.28 | |
| G-SSigLDM=SD22025.09 | 2.3 | 0.332 | — | 19.6 | 0.22 | |
| G-VSLDM=Sana2025.09 | 4.1 | 0.346 | — | 23.5 | 0.07 | |
| G-SSigLDM=Sana2025.09 | 4.2 | 0.347 | — | 28.6 | 0.02 | |
| SanaWM=None2025.09 | 4.3 | 0.346 | — | — | — | |
| SD2WM=None2025.09 | 5 | 0.33 | — | — | — | |
| MAR-MAER+2026.04 | 6.3 | 28.7 | 25.6 | — | — | |
| Hi-MAR2026.04 | 6.8 | 27.1 | 20.3 | — | — | |
| MAR2026.04 | 7.2 | 26.9 | 19.5 | — | — | |
| Fluid2026.04 | 8 | 26.5 | 18.3 | — | — | |
| G-SSigLDM=Flux2025.09 | 9.3 | 0.271 | — | 25.4 | 0.07 | |
| G-VSLDM=Flux2025.09 | 9.3 | 0.269 | — | 26 | 0.07 | |
| FluxWM=None2025.09 | 9.5 | 0.271 | — | — | — | |
| Stable Diffusion2026.04 | 12.1 | 27.8 | 16.2 | — | — |