Preference Alignment
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
61.24Accuracy
11
Mar 23, 2026
58.7OOD Win Rate
10
May 25, 2026
68.4Win Rate
10
May 25, 2026
-5.89Score
10
May 12, 2026
42.7Human Win Rate
8
May 25, 2026
39.1Human Win Rate
8
May 25, 2026
46.11Win Rate
8
Feb 26, 2026
0.658MAE
8
Feb 26, 2026
15.236HPSv3
7
May 21, 2026
19.1Score
6
Jun 30, 2026
78.6Choice Accuracy
6
Jun 30, 2026
77.7LLM-Evaluated Score
6
Jun 30, 2026
4.66Win-rate Delta
6
Apr 21, 2026
61.25Pairwise Win Rate (excl. ties)
5
May 25, 2026
62.35Pairwise Win Rate (excl. ties)
5
May 25, 2026
60.91Pairwise Win Rate (Excl. Ties)
5
May 25, 2026
67.78Pairwise Win Rate (excl. ties)
5
May 25, 2026
15.373HPSv3 Score
5
May 21, 2026
21.63K-Sort Eval Score
5
May 8, 2026
78.2Preference Alignment
5
Feb 26, 2026
87.3Preference Alignment
5
Feb 26, 2026
80.1Preference Alignment
5
Feb 26, 2026
77.4Preference Alignment
5
Feb 26, 2026
9.23LC Win Rate
5
Feb 26, 2026
2.393Borda Average
5
Feb 26, 2026