Hi!
That might be cause by a slight variance in trained weights of the model and is acceptable. To be more precise, you can recheck you previous exercises and make sure that you are using tf functions everywhere.
This variance should not impact your Ex5 grades. Let us know if it does.
hi, may i ask how did you had the accuracy and confusion matrix by vectorize? I used for loop and get accuracy of 0.748 and stuck in the confusion matrix part.
I’m getting the same accuracy and cm as above.
How was this solved for Jatin?
I suspect that the pre-trained model and the unittest might be out of sync.
There is a difference of 13 fewer correct predictions than expected, our of 10240 total predictions. That is odd, and doesn’t seem to point to an error in my function.
Note that the input to the function is a pre-trained model.
How do I give you access to my code in a secure and ethical manner?
If you have updated your lab code with the latest notebooks and you are still not able to get the correct confusion matrix, please send me a copy of your notebook to me privately. You can go to my profile and send me a private message.
The latest notebook has sufficient comments for a learner to pass the test cases. However, there is always room for improvement. Please elaborate on where you are stuck and which part is confusing.
To those coming here and still find the section confusing, please update your assignment as it was recently updated with better instructions. Go to “help → get latest version” to do that.
Note: To get a new copy of the notebook using “Help → Get latest version”, you first have to rename your existing notebook (using the File menu).
Please remove your solution code since it’s against the rules.
Also, very important, that you do not remove code comments like ### START CODE HERE ### and ### END CODE HERE ###, because they are important for grading.
As for your question, you could refer how to get v1 and v2 please check this post or this post