Will Dean
Will Dean
Need to - [x] cover the {var_name}_dim{d} case still #6335. Unless the naming should changed? - [ ] Fix the previous tests The 6335 comes up with this example: ```python...
I did just catch this bug: It comes from the `make_compute_graph` which causes a self loop ```python from pymc.model_graph import ModelGraph coords = { "obs": range(5), } with pm.Model(coords=coords) as...
> I think they should be in the same plate, because in the absense of dims, the shape is used to cluster RVs? How should the {var_name}_dim{d} be handled then...
> Just the length? how does a plate without any dims look like? > > I imagine the mix would be 50 x trial(30) or however the trial dim is...
> How does this differ from https://github.com/pymc-devs/pymc/pull/7302 ? 7302 is styling of the nodes and this is name of the labels for each plate (label in bottom of each rectangle)...
> I feel that may be going too far? Can it be done after a canonical graph was created by PyMC from the outside? > Or anyway that a user...
> The networkx path may not be too crazy? Will give it a try and report back with findings. I trying to avoid parsing and reconstruction of the strings that...
I explore this a bit and I wasn't able to get the result that I wanted. The cons of it: - `networkx` converts to `pygraphviz` instead of `graphviz` which requires...
> Ideally we would have a very bare bones utility that gives users the model tree structure and meta info and let them do whatever they want with it. >...
What if it works like a Deterministic upon the first run (pm.sample) is called