The stack goes: sample from BNN prior --> data generation (export metadata) --> training (export model and log) --> inference (export results) --> analysis
The analysis class should take in the BNN posterior and the test set data to output the posterior over H0.
The stack goes: sample from BNN prior --> data generation (export metadata) --> training (export model and log) --> inference (export results) --> analysis
The analysis class should take in the BNN posterior and the test set data to output the posterior over H0.