qops

torch_to_nnef.op.aten.qops

dequantize

dequantize(g, node, name_to_tensor, inference_target, **kwargs)

Translate aten::dequantize to NNEF.

We will only handle the case of zero_point affine quantization for now.. which in reverse of quantization is:

(x - zero_point) * scale

quantize_per_tensor

quantize_per_tensor(g, node, name_to_tensor, inference_target, **kwargs)

Map PyTorch: 'aten:quantize_per_tensor' to NNEF.