Generative Modeling Evaluation on ImageNet 256x256
5.851KGEL2 ETVQ-VAE2
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| VQ-VAE2Evaluation Protocol=One-sample2023.06 | 5.851 | — | 40.55 | 2.933 | |
| VQ-VAE2Evaluation Protocol=Two-sample2023.06 | 5.723 | — | — | 2.73 | |
| BIGGAN-DEEPTruncation parameter (T)=0.2, Evaluation Protocol=Two-sample2023.06 | 1.955 | — | — | — | |
| VQGAN*k=600, t=1, p=0.92, Evaluation Protocol=Two-sample2023.06 | 1.717 | — | — | 1.495 | |
| VQGAN*k=600, t=1, p=0.92, Evaluation Protocol=One-sample2023.06 | 1.487 | 4.443 | 9.034 | 1.295 | |
| ADM-GGuidance scale=10, Evaluation Protocol=Two-sample2023.06 | 1.442 | — | — | 1.299 | |
| BIGGAN-DEEPTruncation parameter (T)=0.6, Evaluation Protocol=Two-sample2023.06 | 1.404 | — | — | 1.271 | |
| BIGGAN-DEEPTruncation parameter (T)=0.2, Evaluation Protocol=One-sample2023.06 | 1.396 | 3.965 | 6.354 | — | |
| ADM-GGuidance scale=10, Evaluation Protocol=One-sample2023.06 | 1.378 | 3.219 | 5.16 | 1.279 | |
| ADM-GGuidance scale=5, Evaluation Protocol=Two-sample2023.06 | 1.324 | — | — | 1.232 | |
| ADM-GGuidance scale=5, Evaluation Protocol=One-sample2023.06 | 1.294 | 2.492 | 3.602 | 1.232 | |
| BIGGAN-DEEPTruncation parameter (T)=0.8, Evaluation Protocol=Two-sample2023.06 | 1.288 | — | — | 1.201 | |
| BIGGAN-DEEPTruncation parameter (T)=0.6, Evaluation Protocol=One-sample2023.06 | 1.26 | 2.316 | 3.017 | 1.224 | |
| ADMEvaluation Protocol=One-sample2023.06 | 1.255 | 2.035 | 2.994 | 1.18 | |
| ADMEvaluation Protocol=Two-sample2023.06 | 1.244 | — | — | 1.208 | |
| BIGGAN-DEEPTruncation parameter (T)=1, Evaluation Protocol=Two-sample2023.06 | 1.222 | — | — | 1.166 | |
| CDMEvaluation Protocol=One-sample2023.06 | 1.21 | 1.857 | 2.467 | 1.161 | |
| CDMEvaluation Protocol=Two-sample2023.06 | 1.204 | — | — | 1.166 | |
| BIGGAN-DEEPTruncation parameter (T)=0.8, Evaluation Protocol=One-sample2023.06 | 1.203 | 1.964 | 2.443 | 1.183 | |
| ADM-GGuidance scale=1, Evaluation Protocol=Two-sample2023.06 | 1.188 | — | — | 1.151 | |
| ADM-GGuidance scale=1, Evaluation Protocol=One-sample2023.06 | 1.185 | 1.786 | 2.289 | 1.155 | |
| BIGGAN-DEEPTruncation parameter (T)=1, Evaluation Protocol=One-sample2023.06 | 1.166 | 1.735 | 2.075 | 1.15 | |
| VQGAN**k=600, a=0.05, p=1, Evaluation Protocol=Two-sample2023.06 | 1.158 | — | — | 1.202 | |
| VQGAN**k=600, a=0.05, p=1, Evaluation Protocol=One-sample2023.06 | 1.148 | 1.772 | 2.219 | 1.175 | |
| TRAINING SETEvaluation Protocol=One-sample2023.06 | 1.021 | 1.138 | 1.164 | 1.02 | |
| TRAINING SETEvaluation Protocol=Two-sample2023.06 | 1.017 | — | — | 1.018 | |
| THEOR. OPT2023.06 | 1 | 1 | 1 | 1 |