Open-Domain Question Answering on PopQA (Joint answer-and-evidence score)
26.4Joint Answer-and-Evidence ScoreEVE-Agent
Evaluation results
| Method | Links | |
|---|---|---|
| EVE-Agentsearch=true2026.05 | 26.4 | |
| Initial (no search)search=false2026.05 | 6.9 | |
| Dr. Zerosearch=true2026.05 | 5.9 | |
| Initial (search)search=true2026.05 | 4.2 |