Long-form Question Answering on HQ2A
100ComprehensivenessError-Informed Refinement (EIR)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Error-Informed Refinement (EIR)Preference Category=Refined answers2024.07 | 100 | 92.16 | |
| BaselinePreference Category=Baseline answers from dataset2024.07 | 0 | 7.84 | |
| TiePreference Category=Tie (no preference)2024.07 | 0 | 0 |