Part Number: TDA4VM
Hi,
I'm using both SDK 7.1 and SDK 8.1 to test model on TDA4 device.
When import onnx model, I encounter some post-quantization issue. The output .bin model has large scale in one of output layer.
I'm using post-quantization, and I can adjust calibration dataset to bypass this issue. But it would be great If I can manually setup the output layer range.
Thanks,
Naitian