Unsupervised Reinforcement Learning on URL Benchmark Jaco
9Reach Bottom LeftICM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ICMmode=Zero-shot2022.10 | 9 | 7 | 21 | 10 | |
| Disagreementmode=Zero-shot2022.10 | 5 | 3 | 27 | 9 | |
| RNDmode=Zero-shot2022.10 | 4 | 3 | 17 | 4 | |
| APTmode=Zero-shot2022.10 | 1 | 1 | 10 | 4 | |
| AS-BOBmode=Zero-shot2022.10 | 1 | 0 | 1 | 0 | |
| SMMmode=Zero-shot2022.10 | 1 | 4 | 9 | 6 | |
| DIAYNmode=Zero-shot2022.10 | 1 | 2 | 3 | 4 | |
| Protomode=Zero-shot2022.10 | 0 | 1 | 1 | 0 | |
| AS-ALICEmode=Zero-shot2022.10 | 0 | 0 | 0 | 0 | |
| APSmode=Zero-shot2022.10 | 0 | 0 | 0 | 0 | |
| CICmode=Zero-shot2022.10 | 0 | 0 | 0 | 1 | |
| MOSSmode=Zero-shot2022.10 | 0 | 2 | 0 | 1 |