Quantization Accuracy Issue After Optimizing ONNX Model.

General
1

Hi, I am trying to convert my ONNX model. I successfully parsed the model into the .har format, but I am facing an issue with quantization accuracy. After running the optimization, the model’s output differs significantly from the naive model.

image
image1364×508 50.6 KB

Here are my ONNX model and the notebook I used: Download here.

My dataset can be downloaded using the following command: ```bash
gdown ‘1ijYsPq7dtLQSl-oEsUOGH1fAy21YLc7H’


Could you please help me understand why this is happening or suggest how to resolve it?

Thank you!
1 Like