Loss while training ANN is becoming huge negative?

Question

The loss given by function is coming out to be a huge negative while training. The model is given below:

model = tf.keras.models.Sequential()
model.add(tf.ker