List your company

ResearchFormerly WizWand

MM-RewardBench

Benchmarks

TaskDatasetSOTA resultTrendResults

Human Preference Agreement

MM-RewardBench2 T2I

78.9Accuracy

13

Text-to-image preference evaluation

MM-RewardBench T2I 2

78.9Accuracy

11

Human Preference Agreement

MM-RewardBench2 Edit

79.2Accuracy

7

Image editing preference evaluation

MM-RewardBench2 Edit

79.2Accuracy

7