Open wcwitt opened 1 year ago
Could we go even further - all linear solvers after whatever they process to pick some hyperparameters in the end can be interpreted as linear regression with a prior. Can we have a generic function that will compute the posterior distribution and produce a committee? @tinatorabi will also need this in the next PR she is working on.
I need to add the option for all solver variants to return a committee. Currently only some do (maybe one?).