OBQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
OBQA
2.01Accuracy Improvement
12
OBQA
34.6Normalized Accuracy
11
OBQA (test)
92.7Accuracy
10
OBQA in-distribution (test)
81.6Accuracy
9
OBQA (val)
39.6Accuracy
9
OBQA In-Distribution 64
82.73Accuracy
9
OBQA Corrupt Setting
34.4Accuracy
8
OBQA Clean Setting
46Accuracy
8
OBQA
0.884Accuracy
8
OBQA
77.74Accuracy
8
OBQA
25.2Accuracy
7
OBQA
51Accuracy
6
OBQA
42.8Accuracy
4
OBQA
17.8Accuracy (GPT-2-Small)
4
OBQA
12.8Normalized PLL Score
4
OBQA
88.8Accuracy
3
OBQA (dev)
66.7Accuracy
3