Human Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
0.864Fleiss' Kappa (κ)
2
Jun 11, 2026
3.83Quality Score
2
Jun 2, 2026
4.04Quality
2
Jun 2, 2026
4.55Naturalness
2
Apr 28, 2026
3.24Overall Offensiveness Score
2
Feb 26, 2026
14Helpfulness Win Rate
1
Feb 26, 2026
0.5Preference: Much Better
1
Feb 26, 2026
88Win Rate
1
Feb 26, 2026