Personality Trait Steering on TRAIT Extraversion
0.6609Trait Score (+)SAE-based steering
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SAE-based steeringLayer=13, Feature Idx=27392, Polarity=±2026.01 | 0.6609 | 0.3971 | 98.5 | 0.972 | |
| Baseline2026.01 | 0.4463 | — | 97.7 | — | |
| CAALayer=13, Polarity=±2026.01 | 0.295 | 0.143 | 92.4 | 0.862 |