Seth Axen

In addition to a trait to indicate that tree viewer should be used, it would be helpful to be able to customize how the tree is rendered. e.g. I might...

Okay, I'm happy to tackle this once I understand some confusing behavior of complex array rules (see https://github.com/EnzymeAD/Enzyme.jl/issues/744 and https://github.com/EnzymeAD/Enzyme.jl/pull/739#discussion_r1173527477). Unlike the ChainRules support, I would restrict Enzyme rules to...

A few observations after starting work on this: - unlike ChainRules, we should not define rules for `fft`, `ifft`, `bfft` etc, since this will cover up the primal functions or...

> unlike ChainRules, we should not define rules for `fft`, `ifft`, `bfft` etc, since this will cover up the primal functions or any custom methods and do more work than...

@mhauru with #178 and #198 I made it so we use ADTypes internally and LogDensityProblemsAD for LogDensityProblems inputs. Then support for Turing models only requires 1) converting the Turing model...

No, we haven't implemented a solution for this (mostly because no one else reported this being an issue). But now that `LKJCholesky` draws will appear in `Chains` with a syntax...

Okay, I think we can support this by removing support for the old `.` syntax for separating indices. Only Stan sample/CmdStan ever used that, and now they have native InferenceData...

@SamuelBrand1, with the just-released ArviZ v0.11.0, you shouldn't have any issues with your example model, but let me know if you run into any. (e.g. instead you would get the...

No problem!

Any of the Manifolds.jl devs could also review this text.