Agent Performance on HELD-OUT Suite
52.1HotpotQA Score
GPT-4
Updated 4d ago
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
2024.03 | 52.1 | 36.4 | 6.28 | 86.4 | 94.2 | 55.1 | |
2024.03 | 37.4 | 21.2 | 4.56 | 84 | 92.1 | 47.8 | |
2024.03 | 28.5 | 20 | 4.68 | 66 | 89.1 | 41.7 | |
2024.03 | 26.2 | 6.8 | 0.25 | 9.3 | 40.4 | 16.6 | |
2024.03 | 25.4 | 16.8 | 2.71 | 61.8 | 84.5 | 38.2 | |
2024.03 | 22.6 | 5.9 | 1.2 | 27.4 | 78.7 | 27.2 | |
2024.03 | 22.3 | 13.7 | 0.74 | 41.4 | 80.6 | 31.7 |