You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Gen is a PPL for compositional inference. Each draw from a Gen model is stored in a Gen.Trace. Values from the trace are extracted into a Gen.ChoiceMap. We could add a converter for either of these types (not sure which is best).
I'm just not certain if ArviZ will be that useful for the Gen approach. Gen models seem to be highly customized, and I imagine that in general the plots will be too. Plus, the choices can have a hierarchical structure, which wouldn't map directly to an InferenceData.
I'll leave this issue open for input.
The text was updated successfully, but these errors were encountered:
Gen is a PPL for compositional inference. Each draw from a Gen model is stored in a
Gen.Trace
. Values from the trace are extracted into aGen.ChoiceMap
. We could add a converter for either of these types (not sure which is best).I'm just not certain if ArviZ will be that useful for the Gen approach. Gen models seem to be highly customized, and I imagine that in general the plots will be too. Plus, the choices can have a hierarchical structure, which wouldn't map directly to an
InferenceData
.I'll leave this issue open for input.
The text was updated successfully, but these errors were encountered: