Class-conditional posterior sampling on MNIST even odd
0.01FIDFVD
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 | 0.01 | 0.051 | 0.004 | 61.5 | |
| DTSNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.014 | 0.075 | 0.015 | 61.6 | |
| FKNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.017 | 0.071 | 0.005 | 61.5 | |
| DASNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.028 | 0.131 | 0.005 | 61.4 | |
| TDSNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.14 | 0.68 | 0.001 | 58.3 | |
| DPSNFEs=10^6, Number of Seeds=5, Generated Samples=5000, Final Sampling Strategy=Best performance under either reward-weighted or uniform final sampling2026.04 | 0.2 | 0.386 | 0.015 | 60.2 |