Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Human Preference

Benchmarks

Task NameDataset NameSOTA ResultTrend
Human Preference PredictionHuman Preference 22-joint
Accuracy86.09
6
Text-to-Image GenerationHuman Preference Evaluation Set
DEQA4.39
6
Human Preference PredictionHuman Preference 300 motion descriptions (test)
Accuracy77.4
4
Showing 3 of 3 rows