Failed to run a quantized module on in8-version of Vanilla using UMA

[SamiraAhm]

Hello, I am trying to run a quantized tflite model on int8 version of Vanilla using UMA. I have added a Canonicalize pass to support the quantized operators. I think compilation and generating the output c code work well but running fails at the end. here is the main source code:

and here is the errors that I met

Does anyone know how to solve this problem?

[MJKlaiber]

@SamiraAhm, if you are trying to run a quantized model, see here:

UMA qnn.conv2d replacement

Hi @jwouters, yes we are still missing these. I propose to start a thread on that and add it to the RFC? @cgerum @r.stahl @paulpb , any thoughts?

[SamiraAhm]

The issue that I had here was that the output of the original model and runtime model on the Vanilla did not match. So I modified the calculation in Vanilla to make it right and then it worked well.