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

QAMPARI

Benchmarks

Task NameDataset NameSOTA ResultTrend
RetrievalQAMPARI N=1000 (test)
MRecall@10033.7
27
End-to-end generationQAMPARI
Precision23.3
26
Grounded GenerationQAMPARI (test)
Correctness Precision30.28
20
Attributed Text GenerationQAMPARI
Correctness Recall-545.2
19
Information RetrievalQAMPARI N=1000 (test)
Number of Calls16.42
18
Question AnsweringQAMPARI (test)
Correctness Rec@518.86
17
AttributionQAMPARI
Precision37.3
15
List-based Question Answering with CitationsQAMPARI
Correctness24.7
8
Multiple-Answer Question AnsweringQAMPARI
ECE0.0816
4
Showing 9 of 9 rows