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

HybridQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Question AnsweringHybridQA (test)
EM (Total)88.2
23
Question AnsweringHybridQA (dev)
EM (Total)68.4
17
Open-ended Question AnsweringHybridQA (test)
Accuracy91
11
Question Answering over Tables and TextHybridQA
Accuracy91
11
Question AnsweringHybridQA 1.0 (test)
In-Table EM56.2
6
Question AnsweringHybridQA 1.0 (dev)
In-Table EM54.3
6
Table-Text Question AnsweringHybridQA (dev)
Accuracy63.4
5
Multi-hop Question AnsweringHybridQA (dev)
Answer Correctness71
4
Question AnsweringHybridQA
Inference Time (s)10.8
4
Table Question AnsweringHybridQA (test)
Accuracy60.86
4
Cell SelectionHybridQA
Top 1 Accuracy0.761
3
Showing 11 of 11 rows