I don’t know why, I already used a couple model and found the most promising model but the grader still gave me an output “Almost. Your model has NOT passed the required level of accuracy on the test set. The minimum required accuracy is 0.9500. Please try again.” even though I already got the minimum requirement.
keep your model architecture simple, refer to videos week 1 and notice instructors model, you will know how you need to have your model architecture. Do not add too many layers. if there ungraded labs then that gives you also hint.
Thank you!
i have this problem too. Is there any way to run tests WITHOUT having to upload a submission everytime and wait for the fail? It is very timeconsuming. My model is reporting:
Epoch: 98 Loss: 0.00010003003990277648 Accuracy: 0.9999999403953552
…however all 4 submissions have failed with the identical error that original poster had, here. Almost. Your model has NOT passed the required level of accuracy on the test set. The minimum required accuracy is 0.9500. Please try again.
The thing is , if at least told us what accuracy was being achieved, then you would know if you were making the model better or worse. So how would i emulate this on the browser before submission? Is there any other validation files available?
The idea behind this is probably understanding overfitting and underfitting. Although your model training accuracy is 99%, it is probably not passing on test set. So go back check your units, parameters, compare with ungraded labs and videos of week1
you will find the solution.
1 Like
i am going to proceed to week 2 and if i cant pass that assignment, then i will just abandon this course and find a different one as i just cant spend this time on it.
1 Like
Hi Deepti, can you give me more hints please for this first weeks exercise? Its the only one that i could not pass (8 attempts failed with: - Almost. Your model has NOT passed the required level of accuracy on the test set. The minimum required accuracy is 0.9500. Please try again…) The rest of this course is now complete. I attach all my code.
Many thanks! 
Cormac
TensorflowJSCourse_C1_W1_assignmentCode.pdf (45.6 KB)
ok kindly do these changes
Use unit of 10 for the first layer with relu activation
and no input shape need to be mention in the next layer with unit 1 and activation sigmoid.
then use adam optimizer with learning rate around 0.001 or 0.01 depending on desired output required.
you should be able to get passed on this now. please make sure to use only 2 sequentials layers.
Thanks Deepti, i still cant get this to pass - and i tried both learning rates. Is there something wrong with this?:
Many thanks for your support on this
Cormac
Cormac please don’t post codes here.you can dm the code discussion.
Check your dm