Niklas Heim

As this was part of the discussion on slack: I am getting the same behaviour when using `ForwardDiff.gradient` inside ```julia f(x::AbstractVector,y::AbstractVector) = sum(x)*sum(y) df(x,y) = ForwardDiff.gradient(x->f(x,y), x)[1] x = [0.1]...

Is there any chance that this will make it into Zygote?

Thank you for the hint to scipy, I was not planning on using the example like that but I will take a look at it. Anyway, I updated the code...

I think this is related to https://github.com/FluxML/Zygote.jl/issues/692 / https://github.com/FluxML/Zygote.jl/issues/522, because this works: ```julia loss(m,S) = sum(logpdf(TuringDiagMvNormal(m,S),x)) gs = Zygote.gradient(loss, m, S) ```

Any news on this?:) I also just bumped into the same issue.

If I am using this approach to disable/enable GC I am running out of memory at some point. Would it help if I make a minimal example that reproduces this...

Hmm, but I am enabling GC again after the threaded call. I cant find a `GC.collect` function. I tried `GC.gc` (outside threaded calls) but that results in a segfault. Thank...