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

ImageReward

Benchmarks

Task NameDataset NameSOTA ResultTrend
Text-to-Image GenerationImageReward
ImageReward Score1.379
56
Human Preference EvaluationImageReward
Average Score1.0533
24
Preference evaluationImageReward
Accuracy60.7
20
Human Preference EvaluationImageReward (test)
Preference Accuracy0.675
18
Text-to-Image GenerationImageReward (test)
ImageReward Score1.115
16
Text-to-Image AlignmentImageReward (test)
Image Reward1.624
10
Binary ClassificationImageReward (test)
Macro-F165.07
5
Human preference predictionImageReward 371 prompts (test)
Recall @139.62
4
Human preference predictionImageReward 466 prompts (test)
Preference Accuracy65.14
4
Showing 9 of 9 rows