Week 2. Why we use K.mean here?


I don’t understand why we use K.mean in Contrastive Loss in class and don’t use in other places?
This is screen from video “Coding Contrastive Loss”

It’s an interesting question. Notice that averaging is not being done in this fragment from the TensorFlow implementation…

return y_true * tf.math.square(y_pred) + (1.0 - y_true) * tf.math.square(
tf.math.maximum(margin - y_pred, 0.0)

It isn’t mentioned in the video?

The teacher skipped this line.

I did some research on this and it seems the instructor has mistaken that calculation with the RMSE calculation, seems to me. I will raise an issue in github so they can have a look at it as well.