I’m trying to set up part 2 on Colab after doing part 1. Since there are dependencies between files for this notebook, how do I setup in a way that will allow me to use colab like I did for Part 1. Should I clone the repository onto my colab instance, or is there a way to use the files directly from Github?
What I’ve been doing is building the library in my google drive and calling from it. You can run the exportation and everything. Just change the working directory to somewhere in the google drive. I can show an example here in a moment when I have access to my computer
Thanks a lot. Used that and works well.
For the last part (Lesson 13-14). Swift on Colab setup is a bit different. You need to…
- Copy from blank_swift.ipynb or blank_swift_gpu.ipynb
- Update Swift to the latest version
- Copy all Swift materials to the current directory
!git clone https://github.com/fastai/course-v3.git !mv course-v3/nbs/swift/* /content
I built this Jupyter notebook that you can use to set up and run the “Deep Learning From the Foundations” notebooks on Google Colab. @okhursheed has verified that it works. Try it and let me know if you run into any problems!
Thanks a ton!
@okursheed Did the notebook work for you, i.e. did it allow you to successfully run the course notebooks in Colab?
Hi @jcatanza. The notebook works well for the first 00_export.ipynb, but it doesn’t when I move on to the next one. I’m unable to import variables from the first converted python script.
If you have time, could you please show the error message that occurred in the next notebook?
Sure. Here’s a link to the notebook.
I’ve been using Colab+Drive to go through most of part 2, it works well. I just released a blog post describing all the steps necessary to make this work:
If you have any trouble following the steps please let me know!
@jcatanza, I redid the entire process step by step, and it seems to work great now! Thank you!
You can use this Jupyter notebook to set up to run the course notebooks on Google Colab.
I did a follow up on that tutorial. When you’re working with (larger) image datasets some extra steps are necessary:
“Speed up your image training on Google Colab” von Oliver Müller https://link.medium.com/mz2qwYkaz0