Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

SearchQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Web Search Question AnsweringSearchQA held-out (test)
Score (%)87.3
59
Question AnsweringSearchQA (test)
N-gram F175.1
48
Question AnsweringSearchQA
EM78
46
Question AnsweringSearchQA
Accuracy95.1
30
Question AnsweringSearchQA (dev)
F1 (N-gram)68.5
28
Extractive Question AnsweringSearchQA MRQA
F1 Score83
22
Search-based Question AnsweringSearchQA
Hotpot Score42
18
Uncertainty EstimationSQW (SearchQA)
AUROC0.83
18
Open-domain Question AnsweringSearchQA
EM61.96
13
Retrieval-Augmented GenerationSearchQA
Accuracy94.51
9
QA retrievalSearchQA
P@10.587
8
Question AnsweringSearchQA (val)
EM49.6
7
Reading ComprehensionSearchQA (test)
EM56.8
4
Showing 13 of 13 rows