List your company

ResearchFormerly WizWand

RMB

Benchmarks

TaskDatasetSOTA resultTrendResults

Reward Modeling

RMB

89.3Accuracy

120

Reward Modeling

RMB (test)

89.3Score

22

Preference Evaluation

RMB Best-of-N

86.2Helpfulness Score (BoN)

16

Reward Modeling

RMB

88.6Help Accuracy

13

Best-of-N evaluation

RMB

59.69Accuracy

2