Pytorch 转ONNX后进行静态量化后输入报内部错误，尝试两个不同模型，均为这个错误

[lv0]

内部错误：必须同时设置刻度、零和量化器

[lv0]

这里面是我测试的两个模型，望回复

[fauvarque.daniel]

Please post in English on this forum.

The tool can report this error when the quantization is not done according to what is described in the documentation.

We support Int8 per channel static quantization.

You can find more details including sample quantization code in the documentation embedded in the package

Documentation/quantization.html

Regards

Daniel