Confusion regarding data conversion in experiments.py

[arslansadiq]

Hi, Excuse my ignorance, I want to ask that in models_code/experiments.py, function test_eval() has a line y_.cpu().data.numpy()[:, :num_classes].argmax(axis=1).

Now my question is that, why are you using y_.cpu().data.numpy() why not y_.cpu().numpy(), I tested both returns the same output? First of all what does it mean, secondly, both produces same results, why are you using it in this way then? One more thing I want to ask is that, for queries related to your paper, should I write issues here or email you guys?

Thank you Regards

[MSusik]

Now my question is that, why are you using y_.cpu().data.numpy() why not y_.cpu().numpy(), I tested both returns the same output?

I don't have any reason to use one over another. Just like you mentioned, both usually do the same thing.

One more thing I want to ask is that, for queries related to your paper, should I write issues here or email you guys?

If it concerns the implementation, probably Github is better, email otherwise.

[arslansadiq]

Alright Thank for your reply.

It's regarding the math of the paper, I can't really understand the mathematical Intuition behind Inhibited softmax i.e. the introduction of c parameter and the certainty factor. I share more details in the email.