Distribution Alignment on PopAlign-Bench Average
31.9Behavioral Alignment ScoreICL
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ICLLabeller=Claude Sonnet 42026.06 | 31.9 | 23.1 | 24.8 | 8.5 | |
| InterlocutorLabeller=Claude Sonnet 42026.06 | 25.8 | 18.3 | 18.6 | 8.1 | |
| PersonaLensLabeller=Claude Sonnet 42026.06 | 25.3 | 17.6 | 16.8 | 6.6 | |
| AutoPALLabeller=Claude Sonnet 42026.06 | 24 | 17 | 16.7 | 7.3 | |
| DiaSynthLabeller=Claude Sonnet 42026.06 | 23.9 | 17.6 | 15.5 | 7 | |
| ConceptPersonaLabeller=Claude Sonnet 42026.06 | 23.5 | 17.7 | 15.4 | 7 | |
| FaithfulPersonaLabeller=Claude Sonnet 42026.06 | 23.4 | 16.9 | 14.5 | 7.2 | |
| GroupPersonaLabeller=Claude Sonnet 42026.06 | 17.7 | 16.7 | 11.2 | 6.8 | |
| ReferenceLabeller=Claude Sonnet 42026.06 | 2.6 | 8.5 | — | — |