This is part of fastai's Practical Deep Learning for Coders's Lesson 2 on building a Binder app from a Bear Detection Model.
I mainly use Google Colab to study and use their notebooks but Colab doesn't support Voila for this section. With Paperspace Gradient, I used a Jupyter Notebook instance backed by a free GPU.
Initially, I used Binder but I encountered several failed launches and opted to try deploying it to Heroku. However, I think there were compatibility issues with some of the requirements (I didn't get any errors in the log of my last attempt but the last launch still failed) so I switched back to Binder with the help of this forum. If you also encounter issues, you might want to read through the thread as well.