Hi,
I’ve started testing/playing with TVM recently and I’d appreciate feedback on my attempt on running ResNet50 ONNX model based on Compile ONNX Models example.
I am using pre-trained model from ONNX zoo (v1.3 if that matters). To be able to run it I have created gist which is slightly modified file from original example. Most changes are regarding image loading.
For some yet unknown reason I am getting completely wrong result after executing - label 818, which according to what I found is: ‘spotlight, spot’,".
Could someone please check if I didn’t messed up anything in the process? I would like to understand why I might be getting so wrong label.
Thanks!