Query-driven contrastive summarization on Restaurants (test)
87Contrast Win RateQ-STRUM Debate
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Q-STRUM DebateEvaluation Model=GPT-4o, Comparison Baseline=Base2025.02 | 87 | 57 | 84 | 89 | |
| Q-STRUM DebateEvaluation Model=GPT-4o, Comparison Baseline=Contrastive2025.02 | 85 | 57 | 83 | 83 |