Compositional Reasoning
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
71.6Logical Question Accuracy
8
Jun 23, 2026
65.2Logical Question Accuracy
8
Jun 23, 2026
15.1Group Score
8
Apr 14, 2026
61.3Group Score
8
Apr 14, 2026
86.3Accuracy
8
Apr 14, 2026
100Replace Acc (Object)
8
Feb 26, 2026
95.3Event Probability Ratio
8
Feb 26, 2026
43.3Correction Score (C)
8
Feb 26, 2026
88.32Overall Score
7
Jun 23, 2026
51.3Strict Accuracy
6
May 27, 2026
54.3MSR Score
6
Apr 24, 2026
70.5Event Probability p(Ac) > p(Aw)
6
Feb 26, 2026
74.29Compositional Reasoning Score
5
Feb 26, 2026
32.25Text Score
4
Jun 5, 2026
89.9Stability (%)
4
May 27, 2026
83.7Relation Score
4
Mar 4, 2026
21.5Average Score
3
Jun 12, 2026
27Image Accuracy
3
Feb 26, 2026
81.7Accuracy
2
Jun 2, 2026
31.58Text Score
2
Feb 26, 2026
0.335MRR
1
Feb 26, 2026