RuntimeError: The size of tensor a (1600) must match the size of tensor b (196) at non-singleton dimension 1

[SeyedAliRezMousavi]

I tried to train model but got error:

UCTransNet
transformer head num: 4
transformer layers num: 4
transformer expand ratio: 4
Let's use 2 GPUs!
log dir:

========= Epoch [1/2001] =========
Test_session_09.13_13h31
Training with batch size : 4
Traceback (most recent call last):
  File ".\train.py", line 187, in <module>
    model = main_loop(model_type=config.model_name, tensorboard=True)
  File ".\train.py", line 135, in main_loop
    train_one_epoch(train_loader, model, criterion, optimizer, writer, epoch, None, model_type, logger)
  File "D:\Projects\Crack\UCTransNet\Train_one_epoch.py", line 70, in train_one_epoch
    preds = model(images)
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\modules\module.py", line 1051, in _call_impl
    return forward_call(*input, **kwargs)
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\parallel\data_parallel.py", line 168, in forward
    outputs = self.parallel_apply(replicas, inputs, kwargs)
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\parallel\data_parallel.py", line 178, in parallel_apply
    return parallel_apply(replicas, inputs, kwargs, self.device_ids[:len(replicas)])
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\parallel\parallel_apply.py", line 86, in parallel_apply
    output.reraise()
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\_utils.py", line 425, in reraise
    raise self.exc_type(msg)
RuntimeError: Caught RuntimeError in replica 0 on device 0.
Original Traceback (most recent call last):
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\parallel\parallel_apply.py", line 61, in _worker
    output = module(*input, **kwargs)
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\modules\module.py", line 1051, in _call_impl
    return forward_call(*input, **kwargs)
  File "D:\Projects\Crack\UCTransNet\models\UCTransNet.py", line 123, in forward
    x1,x2,x3,x4,att_weights = self.mtc(x1,x2,x3,x4)
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\modules\module.py", line 1051, in _call_impl
    return forward_call(*input, **kwargs)
  File "D:\Projects\Crack\UCTransNet\models\CTrans.py", line 348, in forward
    emb1 = self.embeddings_1(en1)
  File "C:\Users\techno v\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\modules\module.py", line 1051, in _call_impl
    return forward_call(*input, **kwargs)
  File "D:\Projects\Crack\UCTransNet\models\CTrans.py", line 43, in forward
    embeddings = x + self.position_embeddings
RuntimeError: The size of tensor a (1600) must match the size of tensor b (196) at non-singleton dimension 1

[McGregorWwww]

Hi, please check your image size, the size we used in the code is 224*224, so the number of patches is (224/16)^2=196. If you do not want to resize your images, you can change the "img_size" parameter in "Config.py". Hope this can help!

[SeyedAliRezMousavi]

thanks.
https://github.com/McGregorWwww/UCTransNet/commit/c136ee5530e46fe40c250a029783387a39082546