The current inference code expects to perform inference on both injected waveforms and background timeshifts.
It is often the case that one wants to analyze much more background livetime than injection livetime. We should consider either
decoupling these into different inference scripts
making performing inference on injections optional in our current inference script
The benefit of 2) is that we only have to load segments once when analyzing the background and injections (I'm not sure how much of a bottleneck this actually is).
The current inference code expects to perform inference on both injected waveforms and background timeshifts.
It is often the case that one wants to analyze much more background livetime than injection livetime. We should consider either
The benefit of 2) is that we only have to load segments once when analyzing the background and injections (I'm not sure how much of a bottleneck this actually is).