Why my validation lost is “val_loss: 0.0000e+00” and val_accuracy is “val_accuracy: 0.4826” always

Question

I was making a face classification program. When I was using just 2 faces I was using "binary_crossentropy". I want to add new faces and prepared data and I change