Avoid using the onnx.mapping module

[justinchuby]

Usage in

neural-compressor/neural_compressor/adaptor/ox_utils/quantizer.py

Lines 969 to 978 in 4eaef0f

	packed_weight_np_data = np.asarray(
	weight.quantized_data, dtype=onnx.mapping.TENSOR_TYPE_TO_NP_TYPE[weight.qType]
	).reshape(weight.initializer.dims)
	packed_weight_initializer = onnx.numpy_helper.from_array(packed_weight_np_data, packed_weight_name)
	if not self.add_qdq_pair_to_weight or self.mode != "qdq":
	self.model.initializer().append(packed_weight_initializer)
	if weight.axis is not None:
	zero_scale_shape = [weight.initializer.dims[weight.axis]]
	else: # scale and zero point must be scalar

.

It is deprecated. Use helper.tensor_dtype_to_np_dtype instead.

[mengniwang95]

Thanks, I will update it