These files are generated as expected: arduino.ino, library.h, config.h,predict.hand model.h
The predict.cpp file contains errors: The library functions arguments are mispositioned. For example the sparse matrix multiplication function is declared in the library.h as SparseMatMul(const MYINT *Aidx, const MYINT *Aval, MYINT *C, MYINT K, MYINT shrA, MYINT shrB, MYINT shrC ). It is called in the predict.cpp as follows: SparseMatMul(&Widx[0], 128, &Wval[0], 256, 256, &tmp5[0][0], 128);
While running the command, there same issues are encountered with the code present at /tmp/ folder. Here is a screen-shot of the SeeDot.py output
Hello,
While trying to generate the arduino sketch for ProtoNN trained on ups10 dataset as explained in the given example here: https://github.com/microsoft/EdgeML/blob/f04237c664eea1e821733928101c34d3730573c3/tools/SeeDot/README.md
The results of
python SeeDot.py command
are:arduino.ino
,library.h
,config.h,
predict.h
andmodel.h
SparseMatMul(const MYINT *Aidx, const MYINT *Aval, MYINT *C, MYINT K, MYINT shrA, MYINT shrB, MYINT shrC )
. It is called in thepredict.cpp
as follows:SparseMatMul(&Widx[0], 128, &Wval[0], 256, 256, &tmp5[0][0], 128);
Please need help with this.