Kyurae Kim
Kyurae Kim
Hi I also think this would be a really useful idea. Any plans for this to make it forward?
Hi, I think `DifferentiationInterface` could be perfectly useful for `AdvancedVI`. However, for that, I think it would be useful to add an interface where auxiliary information can be passed to...
> It is not supported by the default implementation in AbstractMCMC anymore (downstream packages can in principle support arbitrary keyword arguments but if they do swapping samplers and generic implementations...
Wait I'm a little bit confused here. So `AdvancedMH` and `AdvancedHMC` both seem to do a transition on the first call to `step`, but it seems like only `EllipticalSliceSampling` does...
Oh I got it. I got distracted by the name `transition`. Okay then, it seems there isn't anything preventing AdvancedHMC to do it that way. (In fact the name `step`...
Closed in favor of #38
Amortized inference would be something that is supported by AdvancedVI.jl, but unfortunately, it is not yet possible to do this without pirating the internals. Supporting amortization is on the agenda...
With the current state of things, I think this would be a simple example in the documentation once #38 is worked out.
The magic of Julia is that we don't everything to be implemented in-house in order for things to work. If DeepSets is implemented elsewhere, it could definitely be used here....
At this point, I think it would be better to have particle VI methods in their own package.