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

Out-of-Domain (OOD) Bias Evaluation on StereoSet

67.2Accuracy

C2PO

31.94441.09750.2559.403Dec 29, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.12
67.2
2025.12
64
2025.12
60.4
2025.12
60.1
2025.12
60.1
2025.12
60
2025.12
58.1
2025.12
53.8
2025.12
53
2025.12
44.9
2025.12
42.8
2025.12
41.8
2025.12
33.4
2025.12
33.3