Debiasing
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
96.89ID Accuracy
9
May 28, 2026
83.7Ground-truth Answer Rate
8
May 11, 2026
0.006KL Divergence
8
Feb 26, 2026
0.069KL Divergence
8
Feb 26, 2026
5.07Mean Log-Likelihood Difference
6
Mar 25, 2026
90.4Accuracy
3
Mar 9, 2026
92Accuracy
3
Mar 9, 2026
95.8Accuracy
3
Mar 9, 2026
0.878Content Preservation
2
Mar 20, 2026