Extrapolative Generalization on Walker-walk Unseen Dist.
231.73Mean RewardDreamSAC
Evaluation results
| Method | Links | |
|---|---|---|
| DreamSACunsupervised pre-training steps=2M, task adaptation steps=500K2026.03 | 231.73 | |
| DreamerV3+Policyunsupervised pre-training steps=2M, task adaptation steps=500K2026.03 | 125.44 | |
| DreamerV3+RNDunsupervised pre-training steps=2M, task adaptation steps=500K2026.03 | 113.4 | |
| DreamSAC0-shot=true, unsupervised pre-training steps=2M2026.03 | 67.22 | |
| DreamerV3+RND0-shot=true, unsupervised pre-training steps=2M2026.03 | 34.28 |