Don't average the loss

Models
1

Hi,

I was wondering if there was a way to not average the loss over all examples of a batch- I want to get the losses as a vector of (batchsize , 1) and not (1, 1).

Is that possible?

1 Like
2

I would like to do the same thing while the direct way I’ve found this moment is to forward batchsize times (stupid enough for me :dizzy_face:). Is there a more appropriate way to achieve this result? Please share with me, thanks!

1 Like