Task Generalization on Meta-World ML-45 (test)
81.7Success RateMZ+Recon
Evaluation Results
| Method | Links | |
|---|---|---|
| MZ+ReconEvaluation protocol=finetune, Input source=pixels, Environment steps=10M2021.11 | 81.7 | |
| Q-LearningEvaluation protocol=finetune, Input source=pixels, Environment steps=10M2021.11 | 78.1 | |
| MuZeroEvaluation protocol=finetune, Input source=pixels, Environment steps=10M2021.11 | 76.7 | |
| MAMLEvaluation protocol=few-shot, Input source=states, Environment steps=400M2021.11 | 39.9 | |
| RL2Evaluation protocol=few-shot, Input source=states, Environment steps=400M2021.11 | 33.3 | |
| MZ+ReconEvaluation protocol=zero-shot, Input source=pixels, Environment steps=50M2021.11 | 18.5 | |
| MuZeroEvaluation protocol=zero-shot, Input source=pixels, Environment steps=50M2021.11 | 17.7 | |
| Q-LearningEvaluation protocol=zero-shot, Input source=pixels, Environment steps=50M2021.11 | 10.8 |