Configure quantization output op types

[kleiti]

Hi, is there a way to control what ops the quantization process is allowed to output? Let's say I don't want QLinearMul to appear in the final quantized model.

[mengniwang95]

Sorry for may late reply, you can set op_types_to_quantize of StaticQuantConfig and exclude Mul

neural-compressor/onnx_neural_compressor/quantization/config.py

Line 1536 in 3dd43d3

op_types_to_quantize=None,