Text-to-Image Generation on CC12M (val)
6.27FIDmmDiT-XL/2 (ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| mmDiT-XL/2 (ours)Resolution=5122024.11 | 6.27 | |
| mmDiT-XL/2 (ours)Resolution=2562024.11 | 6.79 | |
| mmDiT-XL/2 (SD3)Implementation=Our re-implementation, Resolution=2562024.11 | 7.54 | |
| UNet (SDXL)Implementation=Our re-implementation, Resolution=2562024.11 | 8.53 | |
| mmDiT-XL/2 (SD3)Implementation=Our re-implementation, Resolution=5122024.11 | 11.24 | |
| UNet (SDXL)Implementation=Our re-implementation, Resolution=5122024.11 | 12.56 | |
| UNet (SD/LDM-G4)Resolution=2562024.11 | 17.01 |