UnboundLocalError: cannot access local variable 'outputs' where it is not associated with a value

[Siegward233]

I have a problem with the test step. The environment is Google Colab. I tried changing num_workers to 0, but it didn't solve the problem. How should I solve this problem? Thank you！
!python /content/University1652-Baseline/test.py --name ft_ResNet50 --test_dir /content/University-Release/test --batchsize 32 --which_epoch 119
We use the scale: 1 -------test----------- /usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead. warnings.warn( /usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or Nonefor 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passingweights=ResNet50_Weights.IMAGENET1K_V1. You can also use weights=ResNet50_Weights.DEFAULTto get the most up-to-date weights. warnings.warn(msg) Load the model from ./model/ft_ResNet50/net_119.pth 1 -> 3: 32 64 96 128 160 192 224 256 288 320 352 384 416 448 480 512 544 576 608 640 672 701 32 Traceback (most recent call last): File "/content/University1652-Baseline/test.py", line 249, in <module> gallery_feature = extract_feature(model,dataloaders[gallery_name], which_gallery) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/content/University1652-Baseline/test.py", line 181, in extract_feature ff += outputs ^^ UnboundLocalError: cannot access local variable 'outputs' where it is not associated with a value

[layumi]

Hi @Siegward233
Sorry for the late reply. The model did not ouput outputs.
Could you check the --views ? Usually --views 2 or --views 3.