I am trying to create a custom loss function using the tensors y_true and y_pred (shape= (batch_size, 1)) and it looks something like this:
minima = tf.reduce_mi