As it stands, a significant portion of the SVI tutorial code is written in markdown code blocks, rather than standalone Jupyter cells. When formatted this way, the tutorial's notebook cannot be excecuted and experimented with by the reader. I suggest that the markdown code blocks should be refactored into Jupyter code ce

Summary:

It'd be nice to have a builder pattern for var contexts to make them easy to construct for testing. Something that could be used like this:

MatrixXd m(3, 2); 
...
var_context vc
  = var_context::builder()
    .matrix("a", m)
    .real("f", 2.3)
    .build();

Current Version:

v2.23.0

trace_to_dataframe() in PyMC3 to save traces is currently implemented in Rethinking_2 notebooks (e.g. Chp_04). But the function is planned for deprecation, with Arviz being the intended package to save traces. As per this comment by @AlexAndorra, Arviz's InferenceData format is a superior replacement to this function as it

Following my post here, https://forum.pyro.ai/t/dirichlet-process-gaussian-mixture-model-typeerror-for-small-concentration-parameter/2685, when one tries creating a Dirichlet Process Mixture Model with 0 mixture components, the error message returned is

TypeError: iota shape must have every element be nonnegative, got (-1,).

This isn't helpful in identifying the source of the error. Could

Hi @JavierAntoran @stratisMarkou,

First of all, thanks for making all of this code available - it's been great to look through!

Im currently spending some time trying to work through the Weight Uncertainty in Neural Networks in order to implement Bayes-by-Backprop. I was struggling to understand the difference between your implementation of `Bayes-by-Bac

Hi,

is there any plan to implement the Generalized Pareto Distribution in brms (paul-buerkner/brms#110 (comment))? I am playing around with an extreme values analysis and it looks like extremes collected as Peak Over Threshold are better represented by the GPD instead of the generalized extreme value distribution, which I am so happy to see already in `b

There are a variety of interesting optimisations that can be performed on kernels of the form

k(x, z) = w_1 * k_1(x, z) + w_2 * k_2(x, z) + ... + w_L k_L(x, z)

A naive recursive implementation in terms of the current Sum and Scaled kernels hides opportunities for parallelism in the computation of each term, and the summation over terms.

Notable examples of kernels with th