Debate Quality Evaluation Alignment on Human-annotated debate rating dataset 1.0 (test)
0.643Pearson RInspireScore
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| InspireScore2025.06 | 0.643 | 0.581 | 0.479 | |
| DebatrixBackbone=gpt-3.5-turbo2025.06 | 0.394 | 0.412 | 0.375 | |
| GPT-4o-miniVersion=2024-07-182025.06 | 0.118 | 0.134 | 0.129 | |
| GPT-4oVersion=2024-08-062025.06 | 0.112 | 0.143 | 0.137 | |
| DAGNVariation=with CL2025.06 | 0.045 | 0.016 | 0.013 | |
| SPARK2025.06 | 0.019 | 0.037 | 0.032 |