Multi-hop Question Answering on HotpotQA (Acc, F1)
45.4AccuracyQREAM-FT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| QREAM-FTReader=GPT-5 mini2026.04 | 45.4 | 53.8 | |
| FaviCompReader=GPT-5 mini2026.04 | 44.1 | 47.3 | |
| GPT-5 miniReader=GPT-5 mini2026.04 | 40.8 | 48.2 |
| Method | Links | ||
|---|---|---|---|
| QREAM-FTReader=GPT-5 mini2026.04 | 45.4 | 53.8 | |
| FaviCompReader=GPT-5 mini2026.04 | 44.1 | 47.3 | |
| GPT-5 miniReader=GPT-5 mini2026.04 | 40.8 | 48.2 |