Chad Scherrer

That makes sense, thanks. The "abstract" part of AbstractGPs wasn't really clear to me - how to overload, etc :)

I think I had seen that before, but it's been a while. I should definitely have another look

Thanks @andyferris, I appreciate the care you're taking in thinking through this. I wasn't aware of the Julia 2.0 plan, but that sounds like it could work out great :)...

Just to note, I was also seeing this in the latest release. But when I cloned the repository in order to add some `@show` statements for debugging, the problem went...

I'm not sure if this is what you mean, but I tried changing the seed handling in `back!` to ```julia if (seed == 1) && (ndims(tape[z].val) > 0) error("Gradient of...

This is maybe naive, but I'd expect `∘(f) = f` would just lead to any `∘(f)` call ignoring `∘` and instead putting `f` itself on the tape. Where does this...

Sounds good, thanks for the update!

This is great progress! I'm working on a big refactoring of Tilde now. I think it's getting close. I'd say it might be another week, though that might be too...

> Do you expect many models in Tilde to be low-dimensional and thus suitable e.g. for ForwardDiff? I think the answer here is the same as for machine learning. Most...

THIS WOULD BE SO GREAT!!! Thanks @torfjelde for taking the initiative on this. BTW, a Soss model is already a measure, so this would make it easy to use Soss...