Task-based evaluation on SimBench
58.4AccuracySCOPE
Evaluation results
| Method | Links | |
|---|---|---|
| SCOPEPersona Source=SCOPE (Ours), Case Description=4. Full conditioning2026.01 | 58.4 | |
| SCOPEPersona Source=SCOPE (Ours), Case Description=2. Demo. + Narratives2026.01 | 55.7 | |
| SCOPEPersona Source=SCOPE (Ours), Case Description=6a. AI profile2026.01 | 55.3 | |
| NemotronPersona Source=Nemotron, Case Description=SCOPE augmented2026.01 | 53.3 | |
| SCOPEPersona Source=SCOPE (Ours), Case Description=3. Demo. + Traits2026.01 | 53.1 | |
| SCOPEPersona Source=SCOPE (Ours), Case Description=1. Demographics only2026.01 | 51.8 | |
| NemotronPersona Source=Nemotron, Case Description=Persona text only2026.01 | 49.6 | |
| SCOPEPersona Source=SCOPE (Ours), Case Description=5d. AI summary2026.01 | 49.4 | |
| NemotronPersona Source=Nemotron, Case Description=All Data2026.01 | 44.6 | |
| SimBenchPersona Source=SimBench, Case Description=Default personas2026.01 | 43.6 | |
| BaselinePersona Source=Baseline, Case Description=No Persona2026.01 | 25.8 |