NQ
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
NQ-Open
28.8Exact Match (EM)
8
NQ
52.2F1 Score
8
NQ320K D_new
64.9Hits@1
8
NQ320K D (train)
0.483Hits@1
8
NQ (Held-out)
52.5Accuracy
8
NQ
33Recall@1 (Base)
8
NQ
44EM
8
NQ Normal
40.17F1 Score
8
NQ (dev)
517.6Words
8
NQ 2019 (test)
61.4F1 Score
8
NQ
0.416ECE
8
NQ-open
22.25Accuracy
8
NQ (test)
36.2METEOR
8
NQ (val)
30.6EM
8
NQ
950QPS
8
NQ
55.7P@1
8
NQ multi-dataset training (test)
84.2Accuracy@20
8
NQ
54.8EM (Exact Match)
7
NQ
34.3EM
7
NQ
56Exact Match (EM)
7
NQ (initial documents)
69.6R@1
7
NQ Extreme
22.5F1 Score
7
NQ
32,601.3Avg Answer Token Count
7
NQ Noisy (test)
37.71F1 Score
7
NQ
98.97True Positive Rate (TPR)
6