# Wrong output results

why is the output gives me wrong results?

The result is 99.99% “5” which I think is the correct answer?
I’m not familiar with the model so maybe I’m missing something.

yes the element 5 should be about 6 - 10, but he give under than 1 in output

I think its probability. 1 is 100%.
“5” got 99.99% and other values are all below 1e-4 (<0.0001).

How can this help me?

So Element 0 - 9 correspond to number 0 - 9.
Element 0 : 5.80e-09 = 0.000000006
Element 1 : 1.86e-13 = 0.0000000000002
Element 2 : 6.05e-11 = 0.00000000006
Element 3 : 4.10e-05 = 0.00004
Element 4 : 1.55e-12 = 0.000000000002
Element 5 : 0.99996

So AI is thinking that the given letter is most likely (99.99%) “5”.
You might be misunderstanding what “e-09” or “e-13” mean?

You didn’t understand the idea at all

hmm maybe.
Could you explain what AI model you are using and what result you are expecting?
From the image you posted, to me it looks like you are trying to classify hand written number.
The image in the middle is the input and result field in the inspector is the classification result?

1 Like

Hi,
Seems like you’re using the MNIST model. The output in the screenshot is correct.

Are you expecting the result before applying a softmax? Check the sample code that’s running on the script, we demonstrate adding a Softmax layer to give probabilities in the output window.

but i draw number 5, and the output should be in the element 5 > 6, and in the other less than 1

hell yeah, thats make it work well, thx bro

1 Like