I need help to jimplement it

J updated my code to

total_loss=tf.reduce_sum(tf.keras.losses.categorical_crossentropy(labels,logits,from_logits=True))

Not runing

Hi there, can you try categorical cross entropy instead of binary?

1 Like

I updated th code to

total_loss=tf.reduce_sum(tf.keras.losses.categorical_crossentropy(labels,logits,from_logits=True))

The same wrong …

Did you divide the sum by total number of samples?

To avoid potential misleading, the implementation now is changed to `compute_total_loss`

so we don’t need to divide it by the total number of samples.

Actually i already used it …