I was reading a simple CNN model to output [1,10]. Training has crossEntroyLoss with Adam optimizer and labels is of shape (210, ). But before passing or computing loss it has basic operations.
for i in range(len(images)):
target=torch.tensor(labels[i])
target = target.long()
target = torch.argmax(target)
target = target.unsqueeze(0)
In this i am not able to get why they used argmax (why their is need of index or the max value index) as label[i] will only contains 1 value.
Labels are as follows-
[0 0 2 0 0 1 6 0 0 0 0 0 0 7 7 1 0 0 6 0 2 4 7 4 5 6 2 5 6 6 3 6 5 0 3 8 5
9 2 8 9 1 7 3 1 4 7 3 8 1 3 4 7 9 3 6 5 8 6 8 2 1 7 8 0 5 6 3 6 4 9 7 9 1
5 3 6 6 8 3 1 4 3 9 8 5 2 4 6 4 7 1 5 2 1 5 8 5 8 3 1 2 4 5 1 2 8 3 8 3 5
4 2 9 5 0 8 6 0 8 5 2 4 5 8 3 2 0 8 6 9 2 8 4 5 8 0 6 2 4 9 4 5 5 2 7 8 4
9 3 2 4 7 5 9 3 1 8 1 3 6 9 1 2 8 2 7 9 9 5 9 8 3 9 8 5 1 4 2 7 0 5 8 6 3
9 6 1 3 7 4 7 1 9 8 3 6 5 6 4 1 3 8 5 4 6 0 4 6 1]