| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Continuous Control | DeepMind Control Suite visual observations | Acrobot Swingup Score24,829 | 16 | |
| Continuous Control | DeepMind Control (DMC) Suite (1M steps) | IQM87.1 | 14 | |
| World model image prediction | DeepMind Control Suite Humanoid | MSE4.2077 | 12 | |
| Physical state prediction | DeepMind Control Suite Humanoid Easy tasks (random policy) | MSE0.6535 | 12 | |
| Physical state prediction | DeepMind Control Suite Reacher Easy tasks (random policy) | MSE0.0005 | 12 | |
| Physical state prediction | DeepMind Control Suite Cheetah Easy tasks (random policy) | MSE0.1206 | 12 | |
| World model image prediction | DeepMind Control Suite Acrobot | MSE0.1806 | 10 | |
| World model image prediction | DeepMind Control Suite Cheetah | MSE0.1565 | 10 | |
| Physical state prediction | DeepMind Control Suite Acrobot Easy tasks (random policy) | MSE0.0001 | 10 | |
| Walker Walk | DeepMind Control Suite (in-distribution) | Average Return996.502 | 10 | |
| World model image prediction | DeepMind Control Suite Walker | MSE1.0044 | 9 | |
| World model image prediction | DeepMind Control Suite Hopper | MSE0.3149 | 9 | |
| Pixel-based Control | DeepMind Control Suite 500k environment steps | Cheetah Run Score803 | 9 | |
| Pixel-based Control | DeepMind Control Suite 100k steps | Cheetah/Run Score512 | 9 | |
| Continuous Control | DeepMind Control (DMC) Suite 500k steps | IQM81.2 | 8 | |
| Continuous Control | DeepMind Control (DMC) Suite 200k steps | IQM57.5 | 8 | |
| Continuous Control | DeepMind Control (DMC) Suite (100k steps) | IQM0.332 | 8 | |
| Continuous Control | DeepMind Control Suite (DMC) | Total Reward0.86 | 8 | |
| Physical state prediction | DeepMind Control Suite Go2 Easy tasks (random policy) | MSE0.041 | 8 | |
| Physical state prediction | DeepMind Control Suite Hopper Easy tasks (random policy) | MSE0.0375 | 8 | |
| Continuous Control Reinforcement Learning | DeepMind Control Suite (DMC) (test) | Cheetah Run917.4 | 8 | |
| Continuous Control | DeepMind Control Suite (DMC) 8 tasks | IQM81 | 7 | |
| Reinforcement Learning | DeepMind Control Suite Easy & Medium | Acrobot Swingup443.53 | 7 | |
| Reinforcement Learning | DeepMind Control Suite (DMC) Hard Tasks (test) | Dog Run544.86 | 7 | |
| Walker Stand | DeepMind Control Suite (in-distribution) | Average Return395.51 | 7 |