Pairwise LLM Evaluation on Restaurants
83Win Rate (Contrast)Q-STRUM Debate
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Q-STRUM DebateComparison Baseline=Q-STRUM Contrastive, Judge Model=Deepseek-v32025.02 | 83 | 52 | 77 | 78 | |
| Q-STRUM DebateComparison Baseline=Q-STRUM Base, Judge Model=Deepseek-v32025.02 | 82 | 53 | 73 | 75 |