about score

[cyy-1234]

Hello author, I run the code example.py you provided, and some scores appear to be less than 0. May I ask, what is the meaning of less than 0? Does it mean that there is an error in this picture？

[xujz18]

The reward is normalized to have a mean of 0 and a standard deviation of 1. When testing stable diffusion v1.4 on our metric set, the scope of the reward is observed to have 62.4% of [-1,1] and 98.2% of [-2,2]. #5 #7 #22

[cyy-1234]

In addition, I would like to ask whether the error is more serious than the aesthetics, and whether the score is lower

[xujz18]

Just refer to our paper: https://arxiv.org/abs/2304.05977. In Appendix B (Annotation Document):

[cyy-1234]

Thank you for your time