Kyurae Kim

See, e.g., https://github.com/compintell/Tapir.jl/edit/main/.buildkite/pipeline.yml

Connect with AdvancedVI.jl

6

Hi all, Now that `[email protected]` is shaping up, shall we consider connecting it with `NormalizingFlows`? I think most of the work would be about setting up documentation and tests matching...

Slice sampling as a `Gibbs` sampler

5

Hi all, I looked into using the samplers in [`SliceSampling`](https://github.com/TuringLang/SliceSampling.jl) as a component to `Experimental.Gibbs`. After a few patches to `SliceSampling`, it seems to be feasible. Here is a snippet...

add minibatch subsampling (doubly stochastic) objective

5

This is a draft for the subsampling variational objective, which addresses #38 . Any perspectives/concerns/comments are welcome! The current plan is to only implement random reshuffling. As I recently [showed](https://arxiv.org/abs/2406.00920)...

Enzyme seems to randomly segfaults: * [Ubuntu](https://github.com/TuringLang/AdvancedVI.jl/actions/runs/11184197549/job/31094437721) * [MacOS](https://github.com/TuringLang/AdvancedVI.jl/actions/runs/11247924796/job/31272185769) This is with AdvancedVI and Julia 1.10. and happens only once in a while. Due to the seeming randomness, I have...

Is there a reason `rand(TDist(nu))` does not obey the type of `nu`?

10

Hi, it seems that ```julia nu = 3f0 rand(TDist(nu)) ``` returns a `Float64` instead of a `Float32`. Is this intended behavior? I find it unusual because the `rand` [implementation](https://github.com/JuliaStats/Distributions.jl/blob/13029c03fa885a2b4512b954e61f9d5a7dfa0612/src/univariate/continuous/tdist.jl#L85) tries...

Add a `logdensity` specialization that takes `NamedTuple`s

3

As discussed on Slack, it would be good to have an additional specialization of `logdensity` that takes `NamedTuple`-valued inputs. The Turing main repo already [implements](https://github.com/TuringLang/Turing.jl/blob/78d110afc17daaedbc492695d258c5eb8541463e/src/mcmc/mh.jl#L267) this as boilerplate for `AdvancedMH`,...