RoboRewardBench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
RoboRewardBench
0.665MAE
24
RoboRewardBench 1.0
0.665Overall MAE
12
RoboRewardBench
0.67MAE
8
RoboRewardBench 500 randomly sampled trajectory pairs
87.4Accuracy
5
RoboRewardBench (held-out)
0.966Spearman VOC
4