Class-conditional posterior sampling on CIFAR-10
0.144FIDFVD
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FVDNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.144 | 0.667 | 0.283 | 0.529 | |
| DTSNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.18 | 0.744 | 0.301 | 0.54 | |
| DASNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.213 | 0.808 | 0.204 | 0.527 | |
| FKNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.241 | 0.917 | 0.055 | 0.514 | |
| DPSNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.343 | 1.224 | 0.117 | 0.528 | |
| TDSNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.411 | 2.116 | 0.035 | 0.482 |