NQ
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
NQ D_special (test)
46.63r_pub (Public Info Retention)
10
NQ
0.945Cohen's d
10
NQ (test)
91.6R@5D
10
NQ
46.08EM
10
NQ
21.2EM
10
NQ 1,000 queries (test)
51.3EM
10
NQ
49.5F1 Score
10
NQ
76.1Recall@1
10
NQ single-hop (test)
47.7Avg@16
10
Simplified NQ
70.81EM
10
NQ
42.8EM
10
NQ (test)
42Token-level F1
9
NQ
31.9Recall@5
9
NQ
46.29Recall@2
9
NQ Open
39.4Exact Match (EM)
9
NQ
43.62F1
9
NQ (test)
75.5Recall@5
9
NQ
0.767Recall@5
9
NQ open (test)
23.63Accuracy
9
NQ 1200 noisy contexts
83.13Unhelpful
9
NQ
78.4Recall@20
9
NQ
88.3Ori
9
NQ
84.8Recall@20
9
NQ (test)
56.2NDCG@10
9
NQ
39.8Truthfulness
8