Inconsistent results between PyTorch loss function for `reduction=mean`
In particular, the following code block compares using
Why I am getting IndexError
Why following code:
Implementing the Contrastive Loss’ Temperature with PyTorch
I am attempting to implement the temperature parameter within a contrastive loss, and I have no idea if I’m doing well. However, it seems to be working because the learning seems to be harder than without the temperature parameter.