0
$\begingroup$

While going over the pseudocode of the CURL paper, the method to identify labels from the logits wasn't clear to me. I believe this technique might be common in other PyTorch/Deep Learning tasks. I have attached the pseudocode below -

enter image description here

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

Given arange returns a 1-D tensor with values from 0 to logits.shape[0], then the labels is a vector of $0$ to $N$ where $N$ is the number of classes predicted by the output layer of f_q.

The CrossEntropyLoss then finds the difference between the predictions and the target labels, which the encoder weights f_q.params is updated according to. I haven't read the paper, but this particular part is a standard multi-class classification approach.

Improve this answer
$\endgroup$

Not the answer you're looking for? Browse other questions tagged .