Commonsense Reasoning
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
73.54BoolQ Accuracy
4
Feb 26, 2026
33.5Accuracy
4
Feb 26, 2026
28.78Accuracy
4
Feb 26, 2026
52.54Zero-shot Accuracy
3
Jun 30, 2026
83.9Pass@1
3
Jun 9, 2026
77.32Pass@1
3
Jun 9, 2026
28.55Normalized Accuracy
3
Jun 8, 2026
81.9Accuracy
3
May 28, 2026
70.25Score
3
May 27, 2026
69.4Accuracy
3
May 25, 2026
87.3Average Score
3
May 25, 2026
50.87HellaSwag Accuracy
3
May 21, 2026
80.6Accuracy
3
May 19, 2026
54.4Accuracy
3
May 14, 2026
56Accuracy
3
May 14, 2026
55.5Performance
3
May 12, 2026
84.4Accuracy
3
May 12, 2026
53.1Accuracy
3
Apr 23, 2026
23.3Perplexity
3
Apr 23, 2026
48.9Accuracy
3
Apr 21, 2026
38Accuracy (HellaSwag 1k)
3
Apr 21, 2026
80.3Normalized Log Accuracy
3
Mar 18, 2026
51.66Accuracy
3
Mar 6, 2026
49.6Accuracy
3
Mar 6, 2026
39.73Accuracy
3
Mar 6, 2026