TypeError: cross_entropy_loss(): argument 'input' (position 1) must be Tensor, not tuple

Question

I am getting this error>> TypeError: cross_entropy_loss(): argument 'input' (position 1) must be Tensor, not tuple

I tried different ways to convert tuple into tensor, however I could not manage to solve the problem. could anyone help me please.

this is the config code that includes the main args_setting.

enter image description here

this is the training code

enter image description here

and this is the model of NN

enter image description here

finally this is the error

enter image description here

thank you in advance.

Answer 1

If you want that those numbers of class_weights be a tensor, you should simply use a

class_weights = torch.tensor([0.02, 1.02])

I recommend you to see this link to know about creating tensors