Luciano Paz

It could also be useful for situations in which posterior samples are streamed into the `Dataset` in no particular order. What I mean by this is that the points of...

If we all agree that the do operation works as an intervention that removes the causal links that feed into the intervened node, then we can call this kwarg `do`....

@drbenvincent pointed me to Pearl’s “Causality” section 1.3.1, that is quite clear in saying that `do` removes the incoming causal links of the intervened node. With that, I agree with...

I'm confused with the [PEP](https://peps.python.org/pep-0561/#packaging-type-information). It says that we need to include the py.typed file as a package data in pyproject.toml, but maybe this is not necessary thanks to [this](https://github.com/pypa/setuptools/pull/4021)?

I'll first address the stuff that @zaxtax said. I agree with @ricardoV94. The idea of having a `Model` instance is for it to work as a sort of book-keeping helper....

@ricardoV94, in version 3.x it used to be necessary to supply the component distribution shape. This could be used to do weird things like making multivariate mixture components out of...

Could you also add a test where the model has a name? To me, it looks like the variable autoname will still have the model name prepended to it, but...

Just to link this, as it could be useful, Torch does something like this with [named tensors](https://pytorch.org/docs/stable/named_tensor.html). Numpy also has its [structured arrays](https://numpy.org/doc/stable/user/basics.rec.html), but those are much more flexible than...

Thanks @EliasRas, I haven't been able to reproduce this yet but that's just because I'm in the middle of switching workstations and haven't gotten everything setup yet. Your fix looks...

> I guess the underlying reason for the failure is that pickling of `DensityDist` doesn't work out of the box? I don’t think the problem is about pickling. The `DensityDist`...