CSQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CSQA
2.21Expected Calibration Error (ECE)
16
CSQA (CommonsenseQA)
84.7AUROC (128 steps)
16
CSQA
13.4ASR@3
16
CSQA (dev)
85.42Accuracy
16
CSQA
81.3Accuracy
15
CSQA
91.75Accuracy
15
csqa
80.6Pass Rate
14
CSQA (leave-one-out setup)
83.8Accuracy
12
CSQA
91.5Accuracy
12
CSQA (OOD)
63.8Accuracy
10
CSQA (dev)
71.1Accuracy
10
CSQA
0.83Kendall Correlation
10
CSQA2
72Accuracy
9
mCSQA
36.8Accuracy (ZH)
9
CSQA
79.2FRS
9
CSQA
79.85Accuracy
9
CSQA
84.98PIQA
9
CSQA (test)
74.8AUROC
8
CSQA
0.9µbias
8
CSQA (test)
82.2Accuracy
8
CSQA (test)
87.2Accuracy
7
CSQA
92.6ASRr
7
CSQA
0.026MAE
7
CSQA
0.769AUROC
7
CSQA2 (dev)
74Accuracy
7