SparseCategoricalCrossentropy not giving exact answer

true_labels = [0,1,2,0]
predicted_logits = [[-2.3,-0.51,-1.20] , [-1.61,-0.36,-2.30], [-2.30, -0.69,-0.92], [-0.92,-0.92,-1.61]]
It gives answer tf.Tensor(1.1242604, shape=(), dtype=float32)

Correct answer for this loss should be 4.5/4 = 1.125

Choose the right category for your query.

Also give brief description about assignment name, exercise you are working and which unittest you are failing.

You can share a screenshot of your output if it is different from expected output but do not post any grade cell codes.