Reinforcement Learning on HalfCheetah delta=[0.2^3, 0.5^3], kappa=2.5 v5 (test)
4,329ReturnDD-SRad
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DD-SRadBackbone=TD32026.05 | 4,329 | 12.2 | 66.8 | |
| DD-SRadBackbone=SAC2026.05 | 4,290 | 15.9 | 62.6 | |
| D-TanhBackbone=SAC2026.05 | 4,205 | 5.5 | 66.2 | |
| D-TanhBackbone=TD32026.05 | 4,000 | 7.9 | 67.2 | |
| BoxPre+Backbone=SAC2026.05 | 3,796 | 1.6 | 60 | |
| BoxPre+Backbone=TD32026.05 | 3,641 | 18.7 | 60.7 | |
| SRad-QPBackbone=TD32026.05 | 3,211 | 4 | 58.2 | |
| SRad-QPBackbone=SAC2026.05 | 2,482 | 17.4 | 55.7 | |
| Post(QP)Backbone=SAC2026.05 | 1,798 | 13.7 | 76.7 | |
| SRad-StrictBackbone=TD32026.05 | 1,518 | 12.2 | 21.1 | |
| Post(QP)Backbone=TD32026.05 | 1,484 | 70.2 | 69.5 | |
| SRad-StrictBackbone=SAC2026.05 | 1,432 | 10.3 | 21.3 |