Open-Domain Question Answering on TriviaQA (Joint answer-and-evidence score)
34.2Joint Answer-Evidence ScoreEVE-Agent
Evaluation results
| Method | Links | |
|---|---|---|
| EVE-Agentsearch=true2026.05 | 34.2 | |
| Initial (no search)search=false2026.05 | 18 | |
| Dr. Zerosearch=true2026.05 | 9.8 | |
| Initial (search)search=true2026.05 | 9.3 |