Collusion Detection on NARCBENCH STEGO
1AUROCActivation Alignment
Evaluation Results
| Method | Links | |
|---|---|---|
| Activation AlignmentTraining Split=CORE, Layers=26–30, Evaluation Protocol=Zero-shot2026.04 | 1 | |
| Sorted ConcatenationTraining Split=CORE, Layers=26–30, Evaluation Protocol=Zero-shot2026.04 | 0.99 | |
| Peak SuspicionTraining Split=CORE, Layers=26–30, Evaluation Protocol=Zero-shot2026.04 | 0.99 | |
| Asymmetry ProbeTraining Split=CORE, Layers=26–30, Evaluation Protocol=Zero-shot2026.04 | 0.98 | |
| Deception SplitTraining Split=CORE, Layers=26–30, Evaluation Protocol=Zero-shot2026.04 | 0.9 |