Facebook/Udacity PyTorch Challenge 2018-2019 Complete!

PyTorch Challenge 2018-2019 Completed

PyTorch Challenge 2018-2019 Completed

Congratulations to all the students that finished the challenge on Jan. 9th 2019. It was an exciting 2 months culminating into a final project performing image classification on specific types of flowers. I was able to get a 95% accuracy overall on my test set. Now that the program is over, I wanted to go over a few key things I learned from the course.

It’s not a sprint, it’s a journey.

As so many people that have participated in scholarships like this have shared, it’s not a sprint, it’s a journey. I took my time with the course material, always trying to finish the exercises without looking at the solutions first. It’s very easy to reverse engineer code after you’ve seen it, but understanding why is much more important. On the other hand, I took it to the other extreme at times getting deep into the weeds on subjects. This prevented me from moving on to the next lesson so the best pace is somewhere in the middle. Give yourself a time constraint so you continue the lessons at a pace where you can finish within the allotted time. I pushed to finish 1 week before the due date to give me time to fine tune any code if needed.

There is no “I” in team.

Cliché, I know, but what makes this course different from other courses I’ve taken is that there is a huge focus on community participation. Being a two month long course with a due date also gives it a sense of urgency and motivation versus self study. Putting my Product Management hat on, this is definitely something other online course providers should consider doing. With the invisible carrot (of an additional scholarship to Phase 2) being dangled in front of all the students, that motivation creates a huge online community in which you are able crowdsource any question you may have. Slack channels and Facebook groups created by regions also give a great mix of online and IRL meetups to help you along the way. I got to meet people from all different walks of life including men, women, students, parents and working professionals. My personal favorite was meeting someone that worked for a company that analyzes your fecal matter, what!?

Don’t sweat the details.

A lot of students were overly concerned about how much online participation is needed. With 10K students worldwide and only 300 moving on to the next phase, my advice would be focus on learning the course material first and foremost. Then participate online when and where you can. It’s like the aircraft safety videos, you have to put on your own  mask before you can help others. I posted often to our facebook group and slack channel whenever I had something to share, but I always made the course material the first priority. Finishing the final lab challenge is the first requirement for moving on so if that’s not finished, any additional participation is moot.

Create a network.

Also don’t forget to network! Make a LinkedIn account and add all your fellow students so you can keep in touch. You never know when you’ll need that connection to a fecal analysis expert. I joked in our meetups that you could use AI to make a recommender system for food recipes based on the data from the analysis. If I ever decide to go down that rabbit hole, I now have a connection to discuss its’ viability. It is always good to have people you can bounce your ideas off of. It might just lead to the next unicorn startup.

Thanks for reading! In my next article, I’ll talk more about the key AI concepts I learned from the course, stay tuned. If you want to learn more about the scholarship program you can read more on Udacity’s blog page.

https://blog.udacity.com/2018/10/introducing-the-pytorch-scholarship-challenge-from-facebook.html

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.