Fritz Obermeyer

@lwiklendt Thanks, I may be blamed here, as I at one point insisted that "stacks can only be mutated at the top: pushed onto the top or popped off of...

Hi @yaowang1111, after some debugging I still don't see what's causing the error in this quite old backend code. Also I'm worried your model may be too trivial, as it...

@SourabhKul you can use Pyro easily for your Bayesian network workflow, except structure learning which you would need to hand-code on top of Pyro. I'm sorry if you found Pyro...

@SourabhKul Great. If you have any intro questions, the best place to ask is https://forum.pyro.ai

Hi @mtsokol that sounds great and I'm happy to provide any review and guidance. Your sketch looks good. The only difference I'd suggest would be for us to think hard...

Addressing your earlier questions: > Also in test_mcmc_api.py additional tests should run both types of MCMC classes and compare final statistics. Correct, most existing tests should be parametrized with ```py...

@mtsokol would you want to work on this in parallel? Maybe you could implement the `StreamingMCMC` class using hand-coded statistics, I could implement a basic `pyro.ops.streaming` module, and over the...

Feel free to implement an `AbstractMCMC` interface if you like. I defer to your design judgement here.

@mtsokol answering your latest questions: 1. tl;dr keep it simple. I do not foresee a performance hit here: it is cheap to create dicts, and `StreamingMCMC` will typically be used...

Yes could you please ignore notebooks? The easiest way to do so is to run black through `make format`. Alternatively we could set up a pyproject.toml file with the black...