Reinforcement Learning on DeepMind Control Suite 500K budget
452.1Acrobot Swingup ScoreEADream
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| EADreamInteraction budget=500K2026.01 | 452.1 | 825.8 | 874.3 | 916.6 | 921.5 | 311.5 | 926.2 | 648.7 | 580.3 | 784.8 | 723.8 | 805.3 | |
| TD-MPC2Interaction budget=500K2026.01 | 241.3 | 790 | 537.3 | 820.8 | 865.6 | 267.6 | 790.3 | 283.1 | 323.5 | 671.9 | 559.2 | 604.6 | |
| DreamerV3Interaction budget=500K2026.01 | 210 | 792.9 | 728.7 | 787.7 | 810.8 | 369.6 | 900.6 | 352.3 | 352.6 | 757.8 | 606.3 | 743.3 | |
| DrQ-v2Interaction budget=500K2026.01 | 128.4 | 706.9 | 691 | 448.4 | 220 | 189.9 | 893 | 407 | 660.3 | 517.1 | 486.2 | 482.8 | |
| CURLInteraction budget=500K2026.01 | 5.1 | 236.2 | 474.3 | 338 | 215.6 | 152.5 | 786.8 | 141.5 | 123.7 | 376.2 | 285 | 225.9 |