TorchConnector returns two sets of weights in the state dict instead of one

[MocanuPaulC]

Environment

• Qiskit Machine Learning version: 0.7.1
• Python version: 3.10
• Operating system: Ubuntu 22.04

What is happening?

When using TorchConnector on a SamplerQNN or EstimatorQNN, the state_dict() holds two sets of the same weights.

How can we reproduce the issue?

import torch
from qiskit_machine_learning.neural_networks import EstimatorQNN
from qiskit_machine_learning.connectors import TorchConnector
from qiskit import QuantumCircuit

qt=QuantumCircuit(1)
qt.rx(0.5,0)
qtest=TorchConnector(EstimatorQNN(circuit=qt, input_params=[], weight_params=[]))
test=torch.nn.Sequential(qtest)
print(test.state_dict())

Output:

OrderedDict([('0.weight', tensor([])), ('0._weights', tensor([]))])

What should happen?

Only one set of the weights should appear

Any suggestions?

No response