Hi,
I have 192 0/1 inputs for a neural network that are packed as 24 8-bit integers. These have to be unpacked before training, for example by using Flatten[IntegerDigits[#, 2, 8]]. When I try to use this expression within a FunctionLayer I am getting 'Compilerr' errors and I have not been able to find a workaround yet. Meanwhile I also found 'CompiledLayer' in the documentation, but when I try to use it as in:
CompiledLayer[
Function[{Typed[input,
TypeSpecifier["NumericArray"]["MachineInteger", 1]]},
Flatten[IntegerDigits[input, 2, 8]]], None]
I am getting the error 'Compile::err: TypeError. Cannot find a definition for the function IntegerDigits that takes arguments with the types NumericArray[Integer64, 1:Integer64], Integer64 and Integer64.' I have tried other TypeSpecifiers but I always get this error.
Any ideas?
Regards,
GW