Dependency Management

I am wondering if someone else as well tried to run all the labs on the local machine and has figured out the set of dependencies that work well with all the labs across the courses or even a single course. All the tensor flow components have too strict dependencies which managing through PIP seems to be difficult. At times I went through changing even the Jupyter versions back and forth in the same course. I believe it is a crucial issue while discussing productionizing any project as we can’t go back and forth on library versions for the new features. Wondering if there is something equivalent to maven BOM here for various TensorFlow components.

Hi Sagar! Thank you for bringing this up. I’m assuming you’re referring to the ungraded labs. These were made to run in Coursera Labs so learners will be able to readily apply the concepts taught in class. It is indeed a good point to at least know the package dependencies. I’ll bring this up with the team and we’ll update the content as needed. Thank you!