Cheers,
I wonder if it is possible to wrap a fully convolutional network written in Flux (e.g. a U-Net) with MLJFlux, and benefit from the high-level interface from MLJ. FCN outputs are typically multi-dimensional array (as in the U-Net), rather than a vector (one-dimension array, as in a classifier) as stated in this page:
The object returned by
chain(x)
must be anAbstractFloat
vector of lengthn_out
If MLJ is indeed not feasible for this task, please kindly advise on alternatives that could provide high-level interface (e.g. cross-validation) for such Flux models.
Thanks in advance.