Dialogic Deference Evaluation on 10-Domain Aggregate (Test Set)
69.6Average Accuracy C1GPT-4o
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GPT-4o2026.01 | 69.6 | 67.6 | 2.5 | 1.5 | -1.1 | |
| GPT-4o-mini2026.01 | 63.6 | 62.6 | 3.5 | 5.5 | 9 | |
| Gemma-3-12B2026.01 | 62.4 | 61.4 | 13.7 | 15.7 | 29.5 | |
| Qwen-2.5-7B2026.01 | 60.4 | 59.2 | 15.8 | 18 | 33.8 |