No need for this line "test_pred = torch.softmax(test_logits, dim=1).argmax(dim=1)" in traning loop for the model

[aryamahadik]

As the work of softmax function to classify the output so there is no need to add "test_pred = torch.softmax(test_logits, dim=1).argmax(dim=1)" this line in traning loop, nothing to do with this line in the loop! if iam wrong pls corret me.

[mrdbourke]

Hey @aryamahadik ,

You are correct, there is no need to add that line:

test_pred = torch.softmax(test_logits, dim=1).argmax(dim=1)

In the testing loop.

You could do the same with:

test_pred = test_logits.argmax(dim=1)

This will give you the same result.

However, if you want prediction probabilities, you can use:

pred_probs = torch.softmax(test_logits, dim=1)

[aryamahadik]

thank you bro to letting me know this i got confused in that line, bro your pytorch course is really helping me thanku for this quality content ! lots of love !