Transfer_learning_with_MobileNet_v1 - tensorflow errors

I am running into different kinds of errors, mostly to do with newer tensorflow definitions, is this noticed by others as well? If so, how should we go about finishing this assignment? I had no such issues with other modules within the specialization.

Please don’t use tensorflow 2.18.
Use package versions from coursera jupyter environment using pip list as the starting point if you’d like to do the labs offline.

Even with tf=2.3.0, I am not able to get the tests passed for:
Implement a function for data augmentation. Use a Sequential keras model composed of 2 layers:

  • RandomFlip('horizontal')
  • RandomRotation(0.2)

Reinstalling tf=2.3.0 and using the functions directly worked well. Thanks!

It is always a mistake to manually install a different version of TF than you are given in any of the notebooks here.

If you are running the notebooks locally on your computer, instead of the course website, then you’ve got some problems to solve with the versions of everything. But on the course website, everything just works as it is given to you. If you think you need a different TF version, then it means you are making some kind of mistake. E.g. in this particular case, please examine the “import” block early in the notebook to notice that they gave you versions of RandomFlip and RandomRotation that work with their chosen version of TF.

2 Likes

Thanks @paulinpaloalto . I should not have mentioned, reinstalling in my earlier message. You were right, the code worked fine with the original version. However, I updated tf to a newer version, that’s why I mentioned reinstalling back to the original version, worked well. Everything is fine now

1 Like