view size is not compatible with input tensor's size and stride

[bemoregt]

Hi, @authman @vfdev-5 @consilium538 @ojdo @cove9988

I met this error when run main.py

python3 main.py data -a 'efficientnet-b4' --pretrained --batch-size 8
main.py:99: UserWarning: You have chosen a specific GPU. This will completely disable data parallelism.
warnings.warn('You have chosen a specific GPU. This will completely '
Use GPU: 1 for training
Loaded pretrained weights for efficientnet-b4
=> using pre-trained model 'efficientnet-b4'
Using image size 380
Traceback (most recent call last):
File "main.py", line 443, in
main()
File "main.py", line 117, in main
main_worker(args.gpu, ngpus_per_node, args)
File "main.py", line 265, in main_worker
train(train_loader, model, criterion, optimizer, epoch, args)
File "main.py", line 311, in train
acc1, acc5 = accuracy(output, target, topk=(1, 5))
File "main.py", line 437, in accuracy
correct_k = correct[:k].view(-1).float().sum(0, keepdim=True)
RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

What's wrong to me?

Thanks , in advance.
Best,

@bemoregt.

[lukemelas]

Hi, this is easy to fix: replace the instance of .view in the definition of the accuracy function with .replace

Also, for future reference, this is a PyTorch ImageNet code thing, not an EfficientNet thing

Hope this helps!