Commonsense Reasoning
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
83.24BoolQ Accuracy
54
Jun 9, 2026
0.105BS
54
Feb 26, 2026
98.67Accuracy
54
Jun 1, 2026
86HellaSwag Score
53
Jun 1, 2026
82.06Accuracy
52
Feb 26, 2026
73.2Aggregate Score
50
May 25, 2026
35OpenBookQA Accuracy
48
Apr 3, 2026
95.1Accuracy
47
Apr 10, 2026
64.11Accuracy
46
Feb 26, 2026
78.9HS Accuracy
44
Jul 9, 2026
77.5BoolQ Accuracy
44
May 22, 2026
93.49Accuracy
44
May 12, 2026
50HS Score
43
May 18, 2026
87.49BoolQ Accuracy
43
Mar 27, 2026
73.7Accuracy (0-shot)
42
Feb 27, 2026
85.47Normalized Accuracy
41
Apr 20, 2026
87.6Accuracy
41
May 13, 2026
76.4BoolQ Accuracy
40
Jul 7, 2026
80.93Average Score
39
Apr 23, 2026
80.2Accuracy
39
Apr 28, 2026
77.5Accuracy
38
Jul 3, 2026
65.27Accuracy
36
Jul 8, 2026
60.04Zero-shot Accuracy
36
Feb 27, 2026
74.1PIQA Accuracy
35
Jun 30, 2026
55.03ARC-c Accuracy
35
May 27, 2026