My goal is to quantize the weight of the model and not to quantize the activation function in QAT. I went through the Pytorch GitHub. However, it gives me an error
AttributeError: ‘Identity’ object has no attribute ‘calculate_qparams’
you can’t just use identity because the quantization flow uses information from the qconfig to determine a number of things and makes assumptions about the types of objects in the qconfig, you should use the actual weight only qconfig instead