Text-to-Image Generation on MS-COCO 50k descriptions randomly selected (train)
7.27FIDImagen
Evaluation results
| Method | Links | |
|---|---|---|
| Imagen#Param=3.0B, Zero-Shot=true, Components=Diffusion part only, excluding language encoder2022.11 | 7.27 | |
| DALLE2#Param=5.5B, Zero-Shot=true, Components=Diffusion part only, excluding language encoder2022.11 | 10.39 | |
| GLIDE#Param=5.0B, Zero-Shot=true, Components=Diffusion part only, excluding language encoder2022.11 | 12.24 | |
| LDM#Param=1.45B, Zero-Shot=true, Components=Diffusion part only, excluding language encoder2022.11 | 12.63 | |
| Spectral Diffusion#Param=77.6M, Zero-Shot=true, Components=Diffusion part only, excluding language encoder2022.11 | 18.43 |