juliaflux.jl

What about Flux.jl enables it to be differentiable?


Reading through some of the Flux.jl docs, I saw this quote:

Differentiable Programming: Existing Julia libraries are differentiable and can be incorporated directly into Flux models.

In the context of Flux and Julia, what does this mean? I understand that Julia and Flux make use of automatic differentiation but how this enables other Julia Libraries to integrate with Flux.jl is not clear to me.


Solution

  • I might be off here, but isn't this just the usual composability story? I.e. if I have some package MyPackage which defines some function myfun then I can plug that function into a Flux model and Flux will be able to differentiate through it, assuming that the types of myfun aren't too constrained.