Will Tebbutt

@sunxd3 thanks for the feedback, and apologies for taking a while to address it. > I think get_taped_globals using task_local_storage is a great idea, but maybe explain a bit how...

~~Actually, @sunxd3 , where do you think might be a good location for the docs for the refs? I'm not 100% sure where to put them, and might be easier...

@sunxd3 @mhauru this is ready for another pass when you get some time. Apologies for taking a while to address your comments.

@yebai I've linked to the current phi node explainer above -- if you think it needs a more thorough explanation please let me know. I take your point regarding the...

Okay. All comments are now addressed, and I believe that this is good to go. @yebai happy to merge if you are.

Thanks @mhauru

Hi @sefffal -- sorry for dropping this. I'm just running CI again to see if the tests start passing. I'm happy to merge if they do, and to assist debugging...

Closing as this is now outdated. Please feel free to re-open if you feel there's more to be done.

I agree with the general sentiment here. In my view we should 1. only test long-run sampling (i.e. to test the correctness of a sampler) once, using whichever AD we...

Agreed. We should think about the precise format though. For the sake of both running the tests in downstream AD packages, and for running them in the Turing.jl integration tests,...