Is there an example of a simple neural network from start to finish using the MKL dnn primitives that are included in the MKL library? I'm trying to directly use the MKL primitives methods like dnnLayoutCreate_F64, dnnExecute_F64, etc. but even after reading the developer manual its difficult to figure out how to use these correctly without some examples. The developer manual has some examples on FFT, is there anything like that for DNN? For instance there is an dnnPrimitiveAttributes_t object but no information in the developer manual on what attributes are in there, I use dnnPrimitiveAttributesCreate_F64 but no information on what it is doing or why I am doing it.
Adam, you may have a look at the examples show how to call create the simplest net by using these function.
you may find out these examples for training and inference ( scoring ) into the folder MKLROOT\examples\dnn.
we hope that helps. with regards.