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

TAT-QA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Table Question AnsweringTAT-QA
Accuracy75.88
45
Multi-hop ReasoningTAT-QA
F1 Score81
21
Question AnsweringTAT-QA (held-out)
Accuracy59.33
21
Table Question AnsweringTAT-QA
Execution Match (EM)77.78
17
Table Question AnsweringTAT-QA (test)
Accuracy60.75
15
Financial Question AnsweringTAT-QA (test)
Execution Accuracy74.96
9
Question AnsweringTAT-QA
HalRate31.5
9
Question AnsweringTAT-QA
Accuracy75.7
9
Question AnsweringTAT-QA 1.0 (test)
EM84.1
6
Question AnsweringTAT-QA 1.0 (dev)
EM55.2
5
Question AnsweringTAT-QA (eval)
Risk8.2
4
Numerical ReasoningTAT-QA (dev)
Exact Match (EM)59.1
4
Question AnsweringTAT-QA (dev)
EM59.1
4
Table structure recognitionTAT-QA (held-out)
TEDS Score70
3
Showing 14 of 14 rows