Misalignment Detection on Strategic deception Bloom behaviors (in-distribution)
0.817AUROC18-indicator ensemble
Evaluation Results
| Method | Links | |
|---|---|---|
| 18-indicator ensembleLayer=292026.06 | 0.817 | |
| 18-indicator ensembleLayer=272026.06 | 0.792 | |
| RepE baselineLayer=292026.06 | 0.758 | |
| RepE baselineLayer=272026.06 | 0.729 |