Mixture of Gaussians on MoG
-4.52RewardDiffOpt (high reward)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DiffOpt (high reward)Settings=βr=5, βc=20, L=500, η=5·10−42026.05 | -4.52 | 21.4 | 1.48 | |
| CFOSettings=K=20, N=3002026.05 | -4.75 | 8.5 | — | |
| DiffOpt (gentle+long)Settings=βr=2, βc=15, L=1000, η=3·10−42026.05 | -5.05 | 10.4 | 1.31 | |
| DiffOpt (default)Settings=βr=1, βc=10, L=200, η=10−32026.05 | -7.54 | 49.4 | 2.05 |