AttributeError: 'list' object has no attribute 'float'

[sleepingcat4]

I was running the c2q_transfer_learning_ants_bees.ipynb notebook and received an error in the qnet function about an attribute error. I am assuming it is happenning as float point is being applied on a list rather than tensors.

Code that is causing the error:

@qml.qnode(dev, interface='torch')
def q_net(q_in, q_weights_flat):
        
        # Reshape weights
        q_weights = q_weights_flat.reshape(max_layers, n_qubits)
        
        # Start from state |+> , unbiased w.r.t. |0> and |1>
        H_layer(n_qubits)   
        
        # Embed features in the quantum node
        RY_layer(q_in)      
       
        # Sequence of trainable variational layers
        for k in range(q_depth):
            entangling_layer(n_qubits)
            RY_layer(q_weights[k + 1])

        # Expectation values in the Z basis
        return [qml.expval(qml.PauliZ(j)) for j in range(n_qubits)]

class Quantumnet(nn.Module):
        def __init__(self):
            super().__init__()
            self.pre_net = nn.Linear(512, n_qubits)
            self.q_params = nn.Parameter(q_delta * torch.randn(max_layers * n_qubits))
            self.post_net = nn.Linear(n_qubits, 2)

        def forward(self, input_features):
            pre_out = self.pre_net(input_features) 
            q_in = torch.tanh(pre_out) * np.pi / 2.0   
            
            # Apply the quantum circuit to each element of the batch and append to q_out
            q_out = torch.Tensor(0, n_qubits)
            q_out = q_out.to(device)
            for elem in q_in:
                q_out_elem = q_net(elem,self.q_params).float().unsqueeze(0)
                q_out = torch.cat((q_out, q_out_elem))
            return self.post_net(q_out)

Error Message:

Training started:
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-39-efc42279baeb> in <cell line: 1>()
----> 1 model_hybrid = train_model(model_hybrid, criterion, optimizer_hybrid,exp_lr_scheduler, num_epochs=num_epochs)

5 frames
<ipython-input-33-2b46b36c83e0> in forward(self, input_features)
     14             q_out = q_out.to(device)
     15             for elem in q_in:
---> 16                 q_out_elem = q_net(elem,self.q_params).float().unsqueeze(0)
     17                 q_out = torch.cat((q_out, q_out_elem))
     18             return self.post_net(q_out)

AttributeError: 'list' object has no attribute 'float'

Can someone help me fixing this error?

[isaacdevlugt]

Hey @sleepingcat4, could you tell us what version of PennyLane you're running (you can just print qml.about())? This repo runs on a very old version of PennyLane (0.7.0 — we are currently on 0.31.1).

[sleepingcat4]

I am using PennyLane version: 0.31.1

[isaacdevlugt]

Yep that explains it! With v0.30, we introduced new return types from QNodes — I think the error you're getting is most likely due to that particular change in PennyLane, but no guarantees because 0.7.0 is quite old.

We have a nice documentation page that shows how the new return types impact legacy code:

https://docs.pennylane.ai/en/stable/introduction/returns.html

Let me know if that helps!

[sleepingcat4]

thanks for reference, it worked and my issue is resolved!

[sleepingcat4]

I though I had fixed it but unfortunately I am geetting the error: AttributeError: 'list' object has no attribute 'float' again. It turns out while I was running it on colab, though I had restarted my runtime, it used version 0.7.0 to execute the training process.

I had earlier installed version v0.7.0 to test but installed the latest immediately afterwards as v0.7.0 didn't have device plug-in pre-packed. Can you please help me in resolving this error?

@isaacdevlugt

[isaacdevlugt]

Hey @sleepingcat4, I'm having some issues getting this notebook to run on my machine due to how old it is 😅. I'll need to investigate this further. Sit tight and I or one of my colleagues will respond this week.

[KetpuntoG]

Hi @sleepingcat4 😄

The problem here is in q_net when returning this list:
return [qml.expval(qml.PauliZ(j)) for j in range(n_qubits)]

In this case the Qnode is returning a list instead of a Tensor. To solve this, I would put:

q_out_elem = torch.Tensor(q_net(elem,self.q_params)).float().unsqueeze(0)

within the forward method of the Quantumnet class.

From my point of view this is not expected behavior so I will let the team know so that they can keep it in mind and change it if they think it is appropriate 😊