Natural Questions
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Natural Questions (NQ)
49.8Exact Match (EM)
10
Natural Questions
39.6QA Metric Score
10
Natural Questions (NQ) open
14.99NQ Open Score
10
Natural Questions (NQ) (test)
53.1EM
10
NQ (Natural Questions) (official)
50.2EM
9
Natural Questions (NQ)
46.9Accuracy
9
Natural Questions (NQ)
0.77AUROC
9
Natural Questions
63.41LM Score
9
Natural Questions
1.272PPL
9
Natural Questions (NQ)
64Avg@4
9
Natural Questions (dev)
54.4Accuracy
9
Natural Questions Open (dev)
36.8EM
9
Natural Questions (NQ) 2048 tokens
34Accuracy
8
Natural Questions (NQ) (test)
1Poisoned RAG Score
8
Natural Questions (NQ) (test)
97.8Accuracy
8
Natural Questions (NQ)
2.19Speedup
8
Natural Questions
4.9Helpfulness
8
Natural Questions (NQ) (dev)
83.3Top-10 Accuracy
8
Natural Questions Wikipedia-18 retriever context (test)
39.2EM
7
Natural Questions
26.7Accuracy
7
Natural Questions (NQ)
71.7Precision
7
Natural Questions 50% coverage
84.53F1
7
Natural Questions (20% coverage)
92.18F1 Score
7
Natural Questions (NQ) few-shot (test)
45.5EM
7
NQ (Natural Questions)
47.94Answer in Context
7