Xianda Sun

No, on your branch

I recreated the issue but haven't figure out what's wrong, seems like a Zygote issue. It does work with ReverseDiff.jl as a backend, you can give it another try by:...

Looks good to me, but maybe we want to wait for @sethaxen

Love this!

What about aliasing? For instance, given ```julia @model function m() x := 1 y ~ Normal(x, 1) x := y end ``` The current implementation will be tracking the value...

Looks good to me. Tiny concern on a technicality: we are using the `tilde` functions to implement the whole `values_as_in_model`. I am a little worried that this might be confusing...

Naive question, what is `if @generated ... pattern`?

I can try and help bring `DynamicPPL` up to `AbstractPPL 0.6`, what exactly break in `0.6` from `0.5` @yebai @torfjelde ?

as far as I can tell, we can introduce `fix` to AbstractPPL, and use it for `predict`. On a higher level, we can also add `predict(model, vector_of_params_and_weights)` and support some...