David Widmann
David Widmann
Took forever but tests passed with this PR in the end: ```julia 10702.904078 seconds (22.35 G allocations: 1.929 TiB, 8.78% gc time, 87.66% compilation time: 0% of which was recompilation)...
There's no response yet in DiffTests to my issue and PR, and I'm not sure when I will have time to dig more into other unrelated test errors on the...
CI passed.
For gradient comparisons one could use or build on https://github.com/torfjelde/TuringBenchmarking.jl (there was some discussion about moving it to TuringLang). `ess_rhat` in MCMCChains already computes ESS/s if the computation time was...
> I think what can be changed/improved is passing the right compile flags to Stan. IMO that would not make the benchmark fairer. I think the correct thing to do...
Surely the models (regardless of whether it's Stan or Turing or ...) should be optimized, and any help for optimizing the Stan models is very much appreciated (BTW I don't...
This is a very old issue but I just discoverd that the cheap acceptance check in the `GammaMTSampler` seems to be incorrect: https://github.com/JuliaStats/Distributions.jl/pull/1617#discussion_r970098297 Fixing it improves performance quite significantly: ```shell...
> Can you mention “parametrized by the log of the odds `Logitp = p/(1-p)`” in the docstring Are you interested in the term "log-odds" specifically? The docstring already mentions that...
Generally this PR should be ready and well-tested.
Ah I see, I thought the main point of your comment was the term log odds. I changed the docstring, hopefully it is clearer now.