using nn.Module(X).argmax(1) - get IndexError

[b0r1ngx]

Hello there, I'm student of NN course, I'm try to implement FFNN (or TDNN) to work on prediction of AR(2)-model, im using PyTorch example, and on my data and NN architecture i got pred.argmax(1) - error:

Traceback (most recent call last):
  File "/home/b0r1ngx/PycharmProjects/ArtificialNeuroNets/group_00201/lab01/lab01_pytorch.py", line 116, in <module>
    first_method()
  File "/home/b0r1ngx/PycharmProjects/ArtificialNeuroNets/group_00201/lab01/lab01_pytorch.py", line 87, in first_method
    test(test_data, time_delay_nn, loss_function)
  File "/home/b0r1ngx/PycharmProjects/ArtificialNeuroNets/group_00201/lab01/lab01_pytorch.py", line 70, in test
    c1 = pred.argmax(1) == y_pred
IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)

where it's used in you're examples:
here in test_loop function - https://github.com/pytorch/tutorials/blob/master/beginner_source/basics/optimization_tutorial.py

I'm also doesn't think that i get best Hyperparameters / loss_function / optimizer - cos i get bad Accuracy / Avg loss in my case,
please help me with that:
U can check my code here:
(now im using how its recommended -1 or 0, but there is always 0)
https://github.com/b0r1ngx/ArtificialNeuroNets/blob/master/group_00201/lab01/lab01_pytorch.py

Thanks!

cc @jerryzh168 @z-a-f @vkuzo

[svekars]

The link to your code returns 404. Can you please provide the correct link?

[z-a-f]

Assuming your training data is (y), and the prediction is (y_pred), maybe it's supposed to be c1 = y_pred.argmax(1) == y? Check your variable names.

[b0r1ngx]

The link to your code returns 404. Can you please provide the correct link?

https://github.com/b0r1ngx/ArtificialNeuroNets/blame/291a9eba04ccecdbda73d8b8c336339a12cdbc03/group_00201/lab01/lab01_pytorch.py

[b0r1ngx]

Sorry that I am lost here, I just delete part of code that I got as a problem:|

[b0r1ngx]

Assuming your training data is (y), and the prediction is (y_pred), maybe it's supposed to be c1 = y_pred.argmax(1) == y? Check your variable names.

Thanks. Ok I check it out and back with answer

[svekars]

@b0r1ngx were you able to test this? If no more questions/suggestions, can we mark this issue as resolved?