HANS
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
HANS (test)
78.65Accuracy
54
HANS (test)
71.2Accuracy
32
HANS (val)
84.3Accuracy
28
HANS
100Accuracy
23
HANS
99.6Accuracy
14
HANS Unknown Bias (out-of-distribution)
70.7Accuracy
13
HANS Syntactic Bias (out-of-distribution)
70.7Accuracy
13
HANS evaluated subset
157Rule-bearing
9
HANS
99.4HANS Accuracy
4
HANS (OOD)
65.82Accuracy
4