RuntimeError: view size is not compatible with input tensor‘s size and stride解决记录

1、报错

Traceback (most recent call last):
  File "main.py", line 419, in <module>
    main()
  File "main.py", line 209, in main
    loss_temp, train_prec1_temp, train_prec5_temp = train(train_loader, model, criterion, optimizer, epoch)
  File "main.py", line 275, in train
    prec1, prec5 = accuracy(output, target, topk=(1, 5))
  File "main.py", line 395, 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

2、分析原因

这是因为view()需要Tensor中的元素地址是连续的，但可能出现Tensor不连续的情况，所以先用 .contiguous()。将其在内存中变成连续分布即可。

3、解决方案

correct_k = correct[:k].contiguous().view(-1).float().sum(0, keepdim=True)

运行，成功解决问题