Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Question Answering on Natural Questions (NQ) (test) (Robust Accuracy)

68Robust Accuracy

Sampling + MIS

36.844.95361.1Sep 27, 2025
Updated 4d ago

Evaluation Results

MethodLinks
2025.09
68
2025.09
67.6
2025.09
63.6
2025.09
63.6
2025.09
62.8
2025.09
62.4
2025.09
62.2
2025.09
61.3
2025.09
60.8
2025.09
60.4
2025.09
60.4
2025.09
60
2025.09
59.8
2025.09
59
2025.09
58.8
2025.09
58.2
2025.09
57.6
2025.09
57
2025.09
57
2025.09
55.8
2025.09
55.2
2025.09
55
2025.09
54.6
2025.09
54.2
2025.09
53.8
2025.09
53.8
2025.09
53.8
2025.09
53.2
2025.09
52.8
2025.09
52.8
2025.09
52.2
2025.09
52
2025.09
51.4
2025.09
50.6
2025.09
50.2
2025.09
49.8
2025.09
48.2
2025.09
47.6
2025.09
47.2
2025.09
46.4
2025.09
46
2025.09
43.4
2025.09
43.4
2025.09
42.2
2025.09
38