Create models by quantization operator

[Jsy0220]

Hi,
I have a trained quantization model in customized format and I want to convert to vino model like model_creation_sample in openvino samples to manually add ops and load weights, but for the quantization ops I can't find related quantization opsets.

So how to deal with this case ?

Thanks

[zulkifli-halim]

Hi @Jsy0220
You can try the openvino.runtime.opset1.fake_quantize. The FakeQuantize operation is a key element in representing quantized models within the OpenVINO Intermediate Representation (IR).