import tensorflow as tf

out = tf.random.uniform([4, 10])
print(out)

y = tf.range(4)
print(y)
y = tf.one_hot(y, depth=10)
print(y)

loss = tf.keras.losses.mse(y, out)
print(loss)

loss = tf.reduce_mean(loss)
print(loss)