When running the following line from the DeepDiagnostics example notebook:
! diagnose --model_path ../resources/savedmodels/sbi/sbi_linear_from_data.pkl --data_path ../resources/saveddata/data_validation.h5 --plots CoverageFraction
the output reads: "Simulator not loaded. Can only run non-generative metrics.”
When it comes to generative vs. non-generative metrics, do these categories of metrics overlap with the metrics we might know/be familiar from traditional machine learning (ML)? Or, are these metrics more aligned with astrophysics data/domain which could mean they are different than what an individual already knows with respect to metrics in ML?