Text-to-image generation on CC3M
6.06FIDMuse (base + super-res)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Muse (base + super-res)Model Type=Non-autoregressive, Params=632M + 268M2023.01 | 6.06 | 26 | |
| Muse (base model)Model Type=Non-autoregressive, Params=632M2023.01 | 6.8 | 25 | |
| Draft-and-reviseModel Type=Non-autoregressive, Params=654M2023.01 | 9.65 | 26 | |
| RQ-TransformerModel Type=Autoregressive, Params=654M2023.01 | 12.33 | 26 | |
| LDM-4Model Type=Diffusion, Params=645M2023.01 | 17.01 | 24 | |
| ImageBARTModel Type=Diffusion+Autoregressive, Params=2.8B2023.01 | 22.61 | 23 | |
| VQGANModel Type=Autoregressive, Params=600M2023.01 | 28.86 | 20 |