How to use ArrayBox with torch?

Hello! If applicable, put your complete code example down below. Make sure that your code:

  • is 100% self-contained — someone can copy-paste exactly what is here and run it to
    reproduce the behaviour you are observing
  • includes comments
def info_nce_loss_torch(features, n_views, batch_size):
    features = torch.from_numpy(features)
    labels =[torch.arange(batch_size) for i in range(n_views)], dim=0)
    labels = (labels.unsqueeze(0) == labels.unsqueeze(1)).float()
    # labels =

If you want help with diagnosing an error, please put the full error message below:

Cell In[3], line 52, in info_nce_loss_torch(features, n_views, batch_size)
     51 def info_nce_loss_torch(features, n_views, batch_size):
---> 52     features = torch.from_numpy(features)
     53     labels =[torch.arange(batch_size) for i in range(n_views)], dim=0)
     54     labels = (labels.unsqueeze(0) == labels.unsqueeze(1)).float()

TypeError: expected np.ndarray (got ArrayBox)

And, finally, make sure to include the versions of your packages. Specifically, show us the output of qml.about().

Hey @Pride-Huang!

Your code example isn’t complete and I can’t replicate your error. However, it seems like features is an ArrayBox. An ArrayBox is a type autograd uses to calculate derivatives of native Python / Numpy code.

Most likely all you need to do is change line 52 to features = torch.from_numpy(features._value), but I can’t verify if this will work unless you post a full and complete working example.

Hope this helps!