Long-form Answer Generation on LongStory
0.8442Spearman CorrelationWPA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| WPA2026.04 | 0.8442 | 0.8058 | |
| Merge2026.04 | 0.8316 | 0.7891 | |
| PCP2026.04 | 0.7986 | 0.7546 | |
| Coarse 5-level2026.04 | 0.7636 | 0.7268 | |
| Coarse 3-level2026.04 | 0.6636 | 0.6339 | |
| BLEU2026.04 | 0.5918 | 0.5442 | |
| ROUGE-L2026.04 | 0.5714 | 0.5442 | |
| Checklist2026.04 | 0.3247 | 0.2993 |