Wait so what happens if I use a model with torch backend now and call .compile()? Does it just return and then do normal torch jit when .fit() (or whatever the

Wait so what happens if I use a model with torch backend now and call .compile()? Does it just return and then do normal torch jit when .fit() (or whatever the keras notation is, i have forgotten most of it) is called?