Evaluation error on binary classification

[jerryzhang-ss]

Thanks for your great work, it is really inspiring.

I tried applying CoTNet onto a binary classification problem, but I got this error:

Traceback (most recent call last):
  File "train.py", line 378, in <module>
    main()
  File "train.py", line 350, in main
    top1_acc, top5_acc = evaler(epoch, model, amp_autocast=amp_autocast)
  File "/home/jerry/git/CoTNet/evaler/evaler.py", line 50, in __call__
    top1_num, top5_num = accuracy(output, target, topk=(1, 5))
  File "/home/jerry/git/CoTNet/utils/meters.py", line 16, in accuracy
    _, pred = output.topk(maxk, 1, True, True)
RuntimeError: selected index k out of range

If I understand correctly, the code assumes there're at least 5 classes in the schema? Is there any way I can get around with it?

[YehLi]

You can modify the code in evaler/evaler.py: accuracy(output, target, topk=(1, 5)) -> accuracy(output, target, topk=(1,))

[jerryzhang-ss]

Thanks, it works!