I’m trying to classify two very similar fonts using a binary classification network. I used BCEwithlogitLoss, but the training loss is either very large (e.g., 0.69) or very small (e.g., 0.002). Can you tell me what could be the reason for this?
I’m trying to classify two very similar fonts using a binary classification network. I used BCEwithlogitLoss, but the training loss is either very large (e.g., 0.69) or very small (e.g., 0.002). Can you tell me what could be the reason for this?