Hello,
I have a batch of samples for which I have to compute the loss for each sample individually(using a for loop). Is there any workaround to compute the loss efficiently, or is it possible to compute it in parallel on a GPU?
Hello,
I have a batch of samples for which I have to compute the loss for each sample individually(using a for loop). Is there any workaround to compute the loss efficiently, or is it possible to compute it in parallel on a GPU?