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

Numerical Question Answering on FinQA 1.0 (test)

91.16Execution Accuracy

Human Expert Performance

19.129637.829856.5375.2302Dec 14, 2022
Updated 1mo ago

Evaluation Results

MethodLinks
91.1687.49
2022.12
71.0768.94
2022.12
70.7968
2022.12
68.79-
2022.12
68.167.3
2022.12
6865.21
2022.12
67.9965.6
2022.12
63.361.29
2022.12
62.1657.54
2022.12
61.2458.86
50.6848.17
2022.12
48.5746.76
2022.12
48.56-
2022.12
21.920.48