Completely stumped by Tensorflow introduction assignment exercise 6

Exercise 6 instructions where not very clear, I think I am following them and I have tried to understand what needs to be done. But it does not seem to work. I am using the two functions tf.reduce_sum and tf.keras.losses.categorical_crossentropy with from_logits set to True. I have read the tensor flow documentation to make sure I am using them right but that has not helped. I suspect I am applying one of them wrong. Can I get a mentor to look at my one line of code? Many thanks in advance

It sounds like you’ve got most of the issues correctly handled, but the one thing that is perhaps least well explained is that you need to transpose the labels and logits. Here’s a recent thread about that.

BTW no-one can simply examine your code directly. Only the course staff have that “superpower”. If the mentors can’t help without looking at your code, we can share code using private DM messages.

1 Like

if the instructions had mentioned that we would need tf.transpose I think I would have seen I needed to use it. The comment in the instructions should went right by me. Thanks for the help.