UserWarning: using a target size (torch.Size([1])) that is different to input size (torch.Size([])) about nssadnn_iqa HOT 4 CLOSED

Hello, do you mean this error occurred where you were running the training code? Could you tell me where the error occurred?

from nssadnn_iqa.

Hello, I ran the code but found no errors during the training progress.
During the training progress, the size of each output and label:
outputs_NSS.size()=[128, 36]
features.size()=[128, 36]
outputs_q.size()=[128, 1]
label.size()=[128, 1]
you can print the size of the output and label to check out that.
And I'm using PyTorch 0.4.1, or maybe the version causes the error? Can you tell me your PyTorch version and the size of outputs to check out why this error happens?

from nssadnn_iqa.

Hello sir,
Yes , I got this warning when I started the traning

< ./lib/python3.6/site-packages/torch/nn/modules/loss.py:88: UserWarning: Using a target size (torch.Size([1])) that is different to the input size (torch.Size([])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size.
return F.l1_loss(input, target, reduction=self.reduction) >
and it affects the results a lot < Epoch 0 Valid Results: loss=26.629 SROCC=0.177 PLCC=0.356 KROCC=0.131 RMSE=31.221
Epoch 0 Test Results: loss=26.473 SROCC=0.222 PLCC=0.340 KROCC=0.165 RMSE=30.960 >
Thanks for your reply .

from nssadnn_iqa.

hello sir,
I changed my Pytorch version to your version and it works very well thank you soo much .

from nssadnn_iqa.