List your company

ResearchFormerly WizWand

Preference Learning

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

Toy dataset 0% label noise (test)

99.6Accuracy

76

Feb 26, 2026

Toy dataset 50% label noise (test)

75.7Accuracy

24

Feb 26, 2026

Driving Simulated

0.948Alignment

15

Mar 11, 2026

Toy dataset Noise 30% (test)

0.739Accuracy

12

Feb 26, 2026

Toy dataset Noise 10% (test)

93.1Accuracy

12

Feb 26, 2026

Simulated Matchmaking Environment row-norm sampling, p_flip=0.2, T=6,400

80Like Rate

6

Apr 14, 2026

0.58Test AUC

4

Feb 26, 2026

Pokémon (test)

86Test AUC

4

Feb 26, 2026

Chameleon (test)

92Test AUC

4

Feb 26, 2026

Synthetic (test)

98Test AUC

4

Feb 26, 2026

Robot Voice Design Simulated

0.852Alignment

3

Mar 11, 2026

Robot Face Design Simulated

96Alignment

3

Mar 11, 2026

Lunar Lander Simulated

93.3Alignment

3

Mar 11, 2026

Anthropic HH-RLHF+VI Preference (test)

64Overall Accuracy

3

Feb 26, 2026

Bayesian algorithm

69.5AUC

2

Feb 26, 2026

Bayesian algorithm

56.5AUC

2

Feb 26, 2026

3 Grades (test)

Bayesian algorithm

53.22AUC

2

Feb 26, 2026

Spectral algorithm

0.707AUC

2

Feb 26, 2026

0.66Test AUC

2

Feb 26, 2026