After training all anomaly scores are low so no anomalies are detected

[eejlny]

Hello,

I am using a simple data set like inj_cora and anomalydae. After training and testing I can see that all anomaly scores are low. I have created a histogram to clarify this where the outliers in red also get a low score. Is it because the score difference between normal and notnormal is very small ? I thought that perhaps I need to remove the outliers for training but this is not clarified anywhere. Thanks,