Task Generalization on Meta-World ML-45 (train)
77.2Success RateMuZero
Evaluation results
| Method | Links | |
|---|---|---|
| MuZeroInput source=pixels, Environment steps=50M2021.11 | 77.2 | |
| MZ+ReconInput source=pixels, Environment steps=50M2021.11 | 74.9 | |
| RL2Input source=states, Environment steps=400M2021.11 | 70 | |
| Q-LearningInput source=pixels, Environment steps=50M2021.11 | 55.9 | |
| MAMLInput source=states, Environment steps=400M2021.11 | 40.7 |